IN
DEGREE PROJECT DESIGN AND PRODUCT REALISATION, SECOND CYCLE, 30 CREDITS
STOCKHOLM SWEDEN 2019,
Motion analysis as a service to
prevent musculoskeletal disorders in forestry
ANTHON BREMER
Motion analysis as a service to prevent musculoskeletal disorders in forestry
Anthon Bremer
Master of Science Thesis TRITA-ITM-EX 2019:479 KTH Industrial Engineering and Management
Machine Design
Master of Science Thesis TRITA-ITM-EX 2019:479
Motion analysis as a service to prevent musculoskeletal disorders in forestry
Anthon Bremer
Approved
2019-06-25 Examiner Sofia Ritzén Supervisor
Jenny Janhager Stier
Commissioner
Anonymous Contact person
Anonymous
Abstract
The purpose of this thesis was to investigate how motion analysis can be used as service to reduce the rate of upper-body musculoskeletal disorders (MSDs) in forestry, and propose a design based on the identified requirements. The project was carried out in collaboration with a Power Tools and Accessories Manufacturer (PTAM). Work-related upper-limb musculoskeletal disorders cost the European Union about 0.5-2% of its Gross Domestic Product yearly, cause immeasurable human suffering, and strain societal resources, not to mention the effect on the finances of the firm employing the afflicted individual. Forestry is the most prolific industry in terms musculoskeletal disorders, with a prevalence rate of circa 6 recorded MSDs per 100 workers and year. MSDs can be prevented, if individuals are aware of the risk. However, the traditional ways of creating awareness are clearly not working, considering the high rates of MSDs. Self-tracking technologies are therefore proposed as a new, more effective, way of increasing risk awareness among forestry workers. By increasing awareness, exposure and subsequent risk can be reduced.
This thesis was initiated by a literature study, user observations and an interview study at PTAM, followed by a technical evaluation and synthesis of the ergonomic parameters. Early concepts were developed, tested with users, and reworked according to their feedback. Some relevant ethical dimensions of this innovation were also considered as the technology has great potential for both positive and negative influence, which can have profound effects on the users and affect their willingness to engage in self-tracking.
The proposed solution is called the Ergonomic Risk Assessment service (ERA). The ERA is a conceptual service that uses motion analysis to assess the risk of developing musculoskeletal disorders caused by hazardous working postures. The ERA is composed of two parts: the tracking unit and the data analysis. The tracking unit is a sports shirt with integrated Bluetooth and IMU sensors which gathers motion data that is used in the automated analysis to assess exposure and risk. The assessments are communicated back to the user via real-time indications and long-term overviews. As a service, the ERA has potential to generate considerate value and have a significant impact on the social and economic sustainability for nations, firms, and individuals, if designed and implemented in an ethically responsible manner.
Keywords: Motion analysis, forestry, musculoskeletal disorders, ethics in technology
Examensarbete TRITA-ITM-EX 2019:479 Rörelseanalys som en tjänst för att förhindra muskuloskeletala besvär inom skogsindustrin
Anthon Bremer
Godkänt
2019-06-25 Examinator
Sofia Ritzén Handledare
Jenny Janhager Stier
Uppdragsgivare
Anonym
Kontaktperson
Anonym
Sammanfattning
Syftet med den här rapporten var att undersöka hur rörelseanalys kunde användas som en tjänst för att minska förekomsten av muskuloskeletala besvär (MSDs) i överkroppen bland skogsarbetare, samt att föreslå en design av ett sådant system baserat på de identifierade behoven och kraven. Projektet utfördes i samarbete med PTAM, ett företag som tillverkar motordrivna verktyg och tillbehör. Arbetsrelaterade muskuloskeletala besvär i överkroppen kostar årligen EU ca 0.5–2% av dess BNP, orsakar ett omätbart lidande och belastar samhällets resurser, för att inte nämna dess inverkan på ekonomin bland de drabbade individernas arbetsgivare. Bland de olika industrierna är skogsbruk värst drabbat av MSDs, med en prevalens på ca 6 anmälda MSDs per 100 anställda och år. MSDs kan förebyggas, om individerna är medvetna om risken. De traditionella sätten att skapa riskmedvetenhet fungerar uppenbarligen inte, med tanke på den höga förekomsten av besvär. Därför föreslås teknik för självövervakning som ett nytt och mer effektivt sätta att öka riskmedvetandet bland skogsarbetare. Genom att öka medvetenheten kan exponeringen och risken att utveckla muskuloskeletala besvär minskas
Den här masteruppsatsen inleddes av en litteraturstudie, användarobservationer, och en intervjustudie på PTAM, som följdes av en teknisk utvärdering och en syntes av de ergonomiska parametrarna. Preliminära koncept utvecklades, testades med användare, och anpassades efter deras återkoppling. Några relevanta etiska dimensioner av den här innovationen beaktades, då tekniken har potential till positiv så väl som negativ inverkan, vilket kan ha djupgående effekter på användarna och påverka deras villighet att använda sig av självövervakningssystemet.
Den föreslagna lösningen kallas för ERA (Ergonomic Risk Assessment service). ERA är en konceptuell tjänst som använder sig av rörelseanalys för att uppskatta risken att utveckla muskuloskeletala besvär orsakade av skadliga arbetsställningar. Lösningen består av två delar:
spårningsanordningen och dataanalysen. Spårningsanordningen består av Bluetooth- och tröghetssensorer vilka samlar rörelsedata, som i sin tur används i analysen för att utvärdera exponering och risk. Utvärderingen kommuniceras sedan till användaren via realtidsindikationer samt långtidsöversikter. Som en tjänst har ERA potential att generera avsevärt värde och ha en betydande påverkan på den sociala och ekonomiska hållbarheten för nationer, företag, och individer, och det designas och implementeras på ett etiskt, ansvarsfullt vis.
Nyckelord: Rörelseanalys, skogsbruk, muskuloskeletala besvär, etik inom teknik
FOREWORD
In this chapter, those who have assisted, aided, contributed to or otherwise helped this thesis are acknowledged.
This thesis was written in the spring of 2019 at KTH Royal Institute of Technology within the area of Innovation Management and Product Development in collaboration with a power tool and accessories manufacturer, PTAM, who will remain anonymous for the purposes of this report.
I have several people who made this thesis possible to give my thanks to. To begin with, I would like to thank my academic supervisor Jenny Janhager Stier and my industrial supervisor PTAM Janhager provided me with guidance on the academic aspects and overall content of the thesis, while the industrial supervisor provided industry insights throughout the project and was of great assistance in finding interview subjects.
I would also like to take the opportunity to thank all the respondents at PTAM for giving me their time and allowing me to interview them. I am very grateful for their responses. Their names have been pseudonymised, but they know who they are, and they can be assured that their answers helped move the project forward. I am also very thankful to everyone at PTAM who recommended or helped me find relevant individuals to interview.
I am also obliged to the arborists who let me test some ideas and discuss concepts with them. Their comments and insights helped improve the value of the concept.
Anthon Bremer Stockholm, June, 2019
TABLE OF CONTENTS
1 INTRODUCTION 1
1.1 Background ... 1 1.2 Purpose... 2 1.3 Delimitations ... 3
2 FRAME OF REFERENCE 6
2.1 The cost of musculoskeletal disorders ... 6 2.2 User perspectives & ethics in self-tracking ... 7
3 RESEARCH QUESTIONS 12
4 METHOD 13
5 TECHNICAL EVALUATION 17
6 ERGONOMIC PARAMETERS 24
7 RESULTS 28
7.1 Results from the PTAM interviews ... 28 7.2 Results from the user observations ... 30 7.3 Results from the user interviews ... 31
8 ANALYSIS 35
8.1 Design considerations ... 35 8.2 Incentives ... 37 8.3 User perspectives & ethical considerations ... 40
9 PROPOSED SOLUTION 44
9.1 Tracker configuration ... 44
9.2 Analysis of motion data ... 47
9.3 Other functionalities to include ... 53 9.4 Motion analysis as a service ... 54
10 DISCUSSION AND CONCLUSIONS 56
10.1 Discussion ... 56 10.2 Recommendations for future research ... 59 10.3 Conclusions ... 61
11 REFERENCES 63
1 INTRODUCTION
In this chapter, the background for the issue at hand is presented, along with the purpose of the thesis. Following the background and purpose, the relevant delimitations are described and argued for.
1.1 Background
Work-related disorders are incredibly costly, cause great suffering, and strain societal resources, but they are preventable. Only in the EU, 476 billion Euros, about 3.3% of its GDP, or just barely the entire GDP of Sweden in 2018 (EU-OSHA, 2017; SCB, 2018), and 7.1 million disability- adjusted life years are lost due to various work-related disorders (EU-OSHA, 2017). Among the work-related disorders, a group of occupational diseases called musculoskeletal disorders (MSDs) is the most frequently reported health complaint in the EU (Hauke et al., 2011). Musculoskeletal disorders represent more than half of all occupational diseases in the EU and are the leading cause of absence from work due to sickness, work disability, ‘presenteeism’ (working with symptoms of disability) and loss of productivity. Half of all absences from work longer than three days are due to MSDs, as well as 49% of sick leaves lasting longer than two weeks. About 60% of all reported cases of permanent incapacity caused by work-related conditions are due to musculoskeletal disorders. Two thirds of those with painful MSDs report a significant reduction in their quality of life (Bevan, 2015). Musculoskeletal also have significant impacts on the economy. Estimates of all the costs associated with work-related upper-limb musculoskeletal disorders place the total costs at between 0.5-2% of Gross Domestic Product (GDP) (Bevan, 2015; Schneider et al., 2010).
Considering that the EU spends 7-8% of its GDP on health care in total (Eurostat, 2018), the cost for work-related upper-limb musculoskeletal disorders is substantial.
Conditions, body regions & risk factors
Musculoskeletal disorder is a broad term, compromising more than 150 diagnoses that affect various parts of the human musculoskeletal system, which consists of muscles, bones, joints, and associated tissues such as ligaments and tendons (Bohgard et al., 2015; WHO, 2018). While the entire locomotive system can be affected, the most frequently reported regions afflicted by musculoskeletal disorders are the back, neck, shoulders, and upper limbs (Bevan, 2015; Calvo, 2009; Dunning et al., 2010; Parent-Thirion et al., 2007; Schneider et al., 2010). The conditions caused by an MSD be anything from sudden and short-lived issues to more permanent or even lifelong disorders (Bohgard et al., 2015; WHO, 2018).
Because it is such a wide group of disorders, the causes for work-related MSDs are usually multifactorial and cover physical as well as ergonomic and psychosocial factors. The main risk factors for MSDs are: repetitive and monotonous motions, awkward or asymmetric postures, carrying or moving heavy loads, lifting, prolonged walking or standing, and vibrations, as well as strict time limits, long working hours, and high working speeds (Schneider et al., 2010). Exposure to these physical hazards increases the risk of developing an MSD. It is worth noting that these diseases develop over time, and symptoms of more severe disorders are sometimes not felt until after the damage has been done (Bevan, 2015; Bohgard et al., 2015; Schneider et al., 2010; WHO, 2018).
2 Risk awareness
Active programs to prevent the development of musculoskeletal disorders have existed for years now on a local, national, and European (see e.g. Arbetsmiljöverket, 2019; Lima et al., 2011;
OSHA, 2019; Nunes & Bush, 2012). Despite these initiatives and the frequency of the disease, few seem to be aware of the risk factors of musculoskeletal disorders. A general understanding of the danger exists but a comprehensive awareness of the risk musculoskeletal disorders appears to be lacking (Al-Qahtani et al., 2016; Deros et al., 2015; Feng et al., 2016; Garbin et al., 2017; Hatch et al., 2017; Ng et al., 2016; Unver-Okan et al., 2017). A lack of awareness may be one of the key reasons why MSDs exist and occur at such high rates, and why they will likely persist unless different actions are attempted. Increasing awareness via the ways of old, using pamphlets, educative material, instructive movies, coaching etc., appear to be inadequate. To extend the capacities of the individual to better detect exposure to physical hazards and become more aware of the risk, self-tracking technologies could be used.
Self-tracking to increase awareness
Motion analysis and other self-tracking systems that monitor hazardous physical exposure, the posture, repetitive motions, or the strain on the worker, to mention a few, have been proposed as means to increase risk awareness among workers (Cao et al., 2019; Lee et al., 2017; Lind et al., 2018; Rawashdeh et al., 2016; Valero et al., 2016; Yan et al., 2017). By becoming more aware of their physical exposure with feedback from an automated analysis of their motions, a self-tracking system could enable workers to reduce their risk of developing musculoskeletal disorders by identifying risk factors for them to minimalize. It is a developing field of potential products but, no advanced motion analysis system for self-tracking is commercially available at the time of writing. What does exist, however, is a significant interest, and tests in practical applications within construction and manufacturing have shown great potential in reducing the users exposure and subsequent risk (Lind et al., 2018; Yan et al., 2017).
1.2 Purpose
The purpose of this thesis is to investigate how motion analysis can be used as a service to prevent the development of MSDs in forestry. This thesis will focus on preventing upper-body musculoskeletal disorders caused by ergonomically incorrect static postures in the trunk, shoulders, and arms, within the forestry industry, for reasons that are presented in chapter 1.3, Delimitations. The project was carried out in collaboration with a Power Tools and Accessories Manufacturer (PTAM), which will remain anonymous in this thesis.
While specific designs and configurations have been tested, no commercially available systems with the purpose of preventing MSDs exists. The lack of existing systems also means that no standards exists. Therefore, the technology that enables motion tracking, and the ergonomic parameters with which motion data is analysed, must be evaluated and synthesised. Sensor configuration, design requirements, and necessary functionalities of the tracking unit based on the users’ needs must be investigated and developed, along with how exposure and risk ought to be communicated.
Because it will be used as a service, it is also relevant to investigate the value of the users and their incentives for use. Some potential ethical issues will also be analysed and discussed. Most, if not all, of the papers on the ethics on self-tracking at the workplace approach it from a sociological perspective without considering a specific system. Papers that combine both the technical and ethical aspects appear to be absent at the moment, and it is the intention of this thesis to combine both to create a functional and ethically sound solution.
1.3 Delimitations
Over the course of writing this thesis, several limitations had to be made. In order to fairly assess what to focus on, a lot of information had to be gathered that may not be pertinent to the rest of the report. The arguments for the selected focus and the limitations are therefore presented in this chapter.
Focus on posture & upper body
This thesis focuses on musculoskeletal disorders in the upper body caused by inadequate or poor working postures. Musculoskeletal disorders can affect the entire body and the causes are usually multifactorial, covering physical as well as ergonomic and psychosocial factors. To focus on combatting all factors in the entire body would be too broad for this thesis. Therefore, the scope is adjusted according to which regions or parts of the body are the most commonly afflicted by musculoskeletal disorders, and what is possible to track in a convenient and unobtrusive manner.
Neck excluded for breaking self-containment
With regards to location of disorders, the back, neck, shoulders, and upper limbs are by far the most frequently reported regions affected by musculoskeletal disorders (Calvo, 2009; Dunning et al., 2010; Parent-Thirion et al., 2007; Schneider et al., 2010). Focus is therefore on upper-body musculoskeletal disorders, since they are the most prevalent. More specifically, due to the limitations posed by the self-containment criteria (that tracking should be carried out without the need for multiple tracking units placed on the user or in the surrounding), only the back, shoulders and upper limbs are monitored. While neck and wrist issues are common, attempting to track the neck or wrists using the selected technologies would require sensors outside the tracker shirt, which would break self-containment and therefore disregarded.
Repetitive work and the issues in assessing it
With regards to risk factors, the only one that can be easily assessed using a wearable motion- tracking system is posture. Posture is a key factor in the development of MSDs, but it is not the only one present in forestry. Aspects such as stress, repetitive work and heavy lifting are also risk factors for MSDs (Schneider et al., 2010), especially within forestry. To only focus on posture could thereby neglect a large sum of the total exposure. One could therefore suggest that it would be reasonable to attempt to measure these as well. While it is a valid suggestion, it would not be reasonable to attempt monitoring of these at such an early iteration, for two reasons.
The standards and ergonomic guidelines that concern repetitive work, such as the ISO 11228, details recommendations and threshold values for repetitive work, such as cycle times, number of repetitions per minute, maximum cycle time one should stay in certain postures, distance moved etc. (ISO, 2014). While it is possible to measure aspects such as cycle time, repetition and so forth, the issue is that the recommendations are weight dependent. For example, the recommended number of repetitions per minute varies with the weight that’s being handled, since the strain differs greatly with the external force.
Posture can easily be assessed using only motion data, but accounting for the exact weight that is being handled during each task is, at this stage, impossible. The technology for it simply does not exist. Attempting to design a functionality that could assess the weight of all items in every task via e.g. computer vision is a master’s thesis in itself. Beyond weight, the exertion is also considered. Exertion is measured on the Borg CR-10 scale, which is a subjective scale. (Bohgard et al., 2015). For an accurate and reliable assessment of physical exposure in repetitive motions, the system must not only be capable of automatically assessing the weight, somehow, the user would also have to enter their perceived exertion after a completed set of working cycles, which
4
issue within forestry and tree care to focus on. Stress, repetitive work and heavy lifting are also valid factors to consider, but they are, at the present, difficult to assess automatically. Focusing on static postures also means that standards that focus on other risk factors, such as repetitive work, will be excluded from the ergonomic parameters unless justified.
Selection of industry
Forestry was selected as the industry of focus for this thesis, but after much research. In order to fairly assess and be able to motivate which industry to direct efforts to, the injury rates and conditions in each industry had to be understood. There are several industries which have high rates of musculoskeletal disorders, such as construction but also manufacturing and certain service professions (Bevan, 2015; Schneider et al., 2010). The industries that report the greatest number of musculoskeletal disorders are presented in table 1, which show that industries such as construction and forestry, but also industries like health and service work, report the highest rates of MSDs in the EU. It needs to be noted that the numbers in the table represent prevalence rates, which are not the same as the total amount. The prevalence rates indicate how many MSDs are reported per 100 000 workers every year.
Table 1. Industries in the EU with the highest prevalence of MSDs
Prevalence rate (per 100 000 workers) of
MSDs in the EU (Schneider et al., 2010) Top five industries in the EU with the highest prevalence of upper-limb MSDs (Bevan, 2015)
• Health and social work (4283)
• Transport and communication (3160)
• Construction (3158)
• Agriculture, hunting and forestry (2895)
• Other community, social and personal services (2666)
• Average rate in the EU (2645)
• Agriculture, forestry and fishing
• Manufacturing
• Construction
• Wholesale and retail
• Hotel and catering
All of these industries are relevant targets but, the company that this thesis is written in collaboration with, PTAM, only makes products for forestry and construction. On paper, construction looks to be more hazardous with a significantly higher rate of disorders; Construction has a prevalence rate of 3158 recorded MSDs per 100 000 workers, which is higher than the rate of 2895/100 000 in forestry (Schneider et al., 2010).However, in the study by Schneider et al.
(2010), forestry is aggregated with hunting and agriculture, which may affect the prevalence rates.
By observing the fatality rates in forestry on its own and compared these with the fatality rates in the aggregated category, a more accurate assessment can be made.
Proportionalities
Statistically, a certain number of near misses occur before a minor injury happens, and a serious or fatal injury occurs happens first after a certain amount of these minor accidents have occurred.
In forestry, this hierarchy is approximately 1000:100:1, meaning that higher rate of fatalities corresponds with a geometric increase in number of injuries (Garland, 2018; Jacobsson, 2018).
Injuries are of course not the same as occupational diseases, but a proportion between recorded injuries and diseases of 6:1 exists in forestry (Arbetsmiljöverket, 2018), meaning that there exists a proportion between the fatalities and the occupational diseases in forestry.
Double the rate of disorders
When observed as an individual category, the number of fatalities per 100 000 workers in forestry is twice as high than the fatality rate in the aggregated agriculture-hunting-forestry category (Garland, 2018). By this reasoning, considering the proportionalities between fatalities, injuries and occupational diseases, the actual rate of recorded MSDs in forestry is double the rate in the aggregated category, being closer to 6000/100 000 than 3000/ 100 000. Forestry is thereby considered to be most egregious industry with regards to MSDs, with a rate of about 6000 reported musculoskeletal disorders per 100 000 workers each year, or 6%. The efforts in investigating the prevention of work-related MSDs via motion analysis systems have neglected forestry workers in favour of industries such as construction or manufacturing. It is probable that focus is on these industries because they are much larger than forestry and, on paper, appear to have the highest rates of MSDs. However, forestry is clearly far more egregious and deserving of attention, considering the high rates of disorders.
Focus on sensor configuration, not the data analysis
Another limitation in this project is that while the service developed consists of two parts, the tracking technology and the data analysis, emphasis is on the tracking technology. The data analysis is, for the purposes of this thesis, treated as a black box. The reasoning for this was that developing a suitable tracker configuration was deemed more critical than creating algorithms for ergonomic assessments for two reasons. Firstly, users will likely not engage with the data analysis beyond the results they receive, which are the exposure and risks assessment. Interaction with the tracking unit would be much more intense and crucial to the user experience and was therefore prioritized. Secondly, specific algorithms were not developed for the same reasons specific components was not selected. The purpose of this thesis was to investigate the possibilities of motion analysis and propose a concept and not a detailed design.
6
2 FRAME OF REFERENCE
In this chapter, the frame of reference is presented. It summarizes existing knowledge and prior research in the relevant subjects. It is later used in the analysis to interpret and discuss the results and design considerations of the proposed solution.
2.1 The cost of musculoskeletal disorders
Conditions caused by MSDs are typically characterised by pain, which is often persistent, and limitations in dexterity, mobility and functional abilities. The individual’s capacity to work and participate in social roles is also reduced, which in turn impacts financial status and mental wellbeing. The firm and the prosperity of entire communities are also affected by the social, psychosocial, and economic impacts of musculoskeletal disorders (WHO, 2018).
Direct & indirect costs of MSDs
For the individual, a musculoskeletal disorder could reduce their productivity at work and at home, which can negatively affect their financial situation and cause psychosocial stress (Bhattacharya, 2014; Schneider et al., 2010). For the firm employing the individual, a distinction between direct and indirect costs must be made when examining costs related to upper-body MSDs. Costs directly related to the management of the identified disorder, such as insurance, compensation, medical and administrative costs, are direct costs. These are generally visible and easier to relate to the MSD than the indirect, hidden, costs, which can be attributed to costs relating to the sick leave, worker compensation, hiring and training new employees, and the potential losses in productivity and quality of work (Schneider et al., 2010). The hidden costs are greater than the direct costs, although the estimates are uncertain. Some estimate them to account for circa half the total cost of an MSD (Bhattacharya, 2014) or almost two thirds of the total cost (Ahlberg, 2014), while Schneider et al. (2010) estimate them to be 10 to 30 times higher than the direct costs.
Difficulties in assessing costs
This variation in indirect costs highlights the difficulties in assessing the costs of MSDs for a company. Different researchers and companies may have different methods for calculating the related costs. Even the definition of an MSD may differ between countries, which affects how the firm reports the disorder. There are also intangible costs involved for the company and the individual, although these will not be considered due to the difficulties associated with quantifying these (Bevan, 2015; Schneider et al., 2010). Furthermore, the condition, severity of symptoms and the duration of the absence or disability also affect the overall cost of the disorder, which emphasises the need to observe the cost in the industry of focus. Accurate or reliable MSD cost assessments regarding forestry and tree care are difficult to attain. Indications from other industries have to be used, in this case manufacturing and slaughterhouses are used as proxies, due to their similarities in MSD rates (Bevan, 2015; Fauconnier et al., 2005; Schneider et al., 2010) and type of labour performed (Fagan & Hodgson, 2017). The following cost estimates should therefore be handled with care but, due to the similarities in prevalence rates of upper-body MSDs, they can be considered accurate enough to serve as surrogates.
Estimates of the firm-related costs for MSDs
In a study investigating the real costs of MSDs in three manufacturing companies making engines and electronics in France, each with over 500 employees, the average total cost for upper-body MSDs for the firm was found to be between 6 800 to 11 200 Euro per afflicted person and year.
These costs are primarily due to high rates of absenteeism and a productivity loss of about 7%,
and to a lower degree social costs, medical management costs, and administrative costs (Fauconnier et al., 2005). However, that may be a cautious estimate. The cost of MSDs at a slaughterhouse are also investigated in the same article by Fauconnier et al. (2005). For slaughterhouse workers the average total cost for a recognised MSD is about 38 200 Euros (ibid).
The firm also suffers in the form of lost working hours. Around 38% of all absenteeism, and 24%
of working days lost, are due to MSDs in the EU (Blatter et al., 2005; Schneider et al., 2010). In manufacturing, a mean average of 215 working days per afflicted individual are lost due to musculoskeletal disorders (Dunning et al., 2010). An MSD may therefore incur great costs for the firm and incapacitate a worker for almost an entire working year.
High rates of unrecorded disorders
It is worth mentioning that these estimates are only concerned with the number of reported disorders. One should be aware that few occupational diseases are reported. For example, Swedish Arbetsmiljöverket (2018) assesses that barely 10% of all occupational diseases are reported.
Valero et al. (2016) state that there exists a “large number of unreported [work-related] injuries”
in the UK, and in the US, neglecting or under-reporting (down-playing the severity) of work- related disorders is estimated to occur in 20% to 70% of all cases, and about 50% of employers fail to fully report all disorders, according to Fagan & Hodgson (2017). This implies two things.
Firstly, it implies that the prevalence rates may contain a high amount of uncertainty and should be handled with care. Secondly, since far from all occupational diseases are reported, it also implies that the actual rate of disorders in forestry may be even higher than 6%.
2.2 User perspectives & ethics in self-tracking
Quantifying the self via self-tracking is not a new practice; humans have been self-tracking since at least the 16th century (Swan, 2013), nor is implementing self-tracking at work. Simple self- tacking devices have already been implemented in wellness programs of hundreds of thousands of firms in the US. These initiatives often utilise fitness devices like Fitbits to track and quantify their employees in terms of heartrate, calories used, time spent on a task, mood, and so forth (Karkar et al., 2015; Moore & Robinson, 2016; Rivera-Pelayo et al., 2017). Although these are simple modes of self-tracking and technological advances have allowed for a greater degree of more convenient and sophisticated self-tracking than ever before (Crawford et al., 2015; Day, 2016; Wolf, 2016).
But technological advances alone are insufficient reasons for why companies are so interested in self-tracking.
Wellbeing and data
Companies want healthy and productive workers (Ajana, 2017). However, employee wellbeing and productivity have been in decline over the past decade (Moore & Piwek, 2017). Self-tracking programs can be used to improve employee health and encourage greater reflection and learning, which in turn lead to higher productivity and quality of work (Karkar et al., 2015; Lind et al., 2018;
Mettler & Wulf, 2019; Moore & Robinson, 2016; Rivera-Pelayo et al., 2012; Rivera-Pelayo et al., 2017; Vignais et al., 2013 Zhao, 2016). Because of its perceived objectiveness, the tracking data is considered more valuable and useful in self-optimisation than “vague” instincts or emotional notions (Feiler, 2014). In fact, the motto of the Quantified Self movement, the largest community of self-tracking enthusiasts, is “Self-knowledge through numbers” (Ajana, 2017). Therefore, Moore & Piwek (2017) argue that firms have begun to turn to implementing employee self- tracking programs as a means to efficiently encourage self-optimisation, which in turn leads to improved wellbeing, health, and productivity.
8 User perspectives and ethics in tracking
But, while there exist tangible benefits to the employer, the perspective of the employees, those who will subject themselves to quantification, needs to be taken into consideration. The increased presence of self-tracking technologies also presents ethical issues (Lupton, 2014; Mettler & Wulf, 2019; Moore & Piwek, 2017; Moore & Robinson, 2016), which must be discussed. The self- tracking technologies have great potential to be of benefit, but they also have the potential to be of harm. Most research in self-tracking technologies focus on the technical possibilities, but it is also important to be mindful of the detriments brought by this kind of innovation. Even very simple self-tracking systems can be harmful, as evidenced by the case of Amazon warehouse pickers.
Example of abusing self-tracking technologies
Warehouse pickers at Amazon, the online retailing company, are instructed by a wrist-mounted computer where their next item is located and how soon they need to reach it. Failure to meet the set time result in a warning, and repeated failure leads to termination. Time limits are not inherently problematic in self-tracking, but the programming assumes that the picker can maintain a jogging pace throughout their entire shift, which usually lasts 10-12 hours, with little to no leeway for taking breaks or going to the bathroom. Pickers often report that they are being “treated like robots”
(BBC, 2018), and the high demands on productivity over such a long periods of time are so stressful it often harms their mental wellbeing (BBC, 2013; Ghosh, 2018; Gracely, 2014; Picchi, 2018; Shapiro, 2018). The self-tracking system has even become capable of automatically firing individuals who fail to meet the high expectations, which further increases the pressure on the employees (Bort, 2019).
Granted, it is not the technology in itself that is causing harm to the pickers, but rather the poor management practices. However, the technology is being used to enforce these practices and normalise harmful working behaviour. If a simple self-tracking system with only a timer and activity meter can have such an impact, one can only imagine what abuse of a more sophisticated monitoring system could cause. The quantified workplace, where self-tracking technologies are common, is at the verge of being implemented at a much larger scale, and the “future normal” will likely involve even more tracking devices (Moore & Piwek, 2017). For these reasons, and reasons which will be discussed, the user’s perspectives and the ethical dimensions of implementing self- tracking at work needs to be considered. While all dimensions of the ethics and the user perspectives cannot be presented in this frame of reference, the most pertinent aspects can and will be considered.
User perspectives
For self-tracking programs to even function, the employees need to engage in it. It is therefore important to understand their perspective. However, literature on the users’ perspective on self- tracking in the workplace is somewhat scarce. In order to address this gap, and to better understand the employee’s perspective and attitudes towards self-tracking, Mettler & Wulf (2019) designed a survey. The survey investigated the attitudes of individuals about to engage in self-tracking programs, gauging their interest in what the self-tracking system should do, or not do, and potential incentives or deterrents for using it. This survey was designed so that respondents answered how much they agreed or disagreed with a certain statement. The most agreed and disagreed statements from the survey by Mettler & Wulf (2019) are presented in table 2. Future users agreed the most with the statement was that use of self-tracking systems at work should be “completely voluntary”.
The most disagreed statement was that users would like to be “monitored during all [their] job activities”, even if it is carried out for the sake of their wellbeing.
Table 2. The 5 most agreed and disagreed statements from the survey by Mettler & Wulf (2019), in descending order from most agreed/ disagreed to fifth most agreed/ disagreed.
Most agreed statements Most disagreed statements
• The use of algorithmic decision- making tools within the scope of (health prevention) at work should be completely voluntary
• I do not need a tool to tell me when I feel sick or stressed
• There should be no peer pressure when it comes to using algorithmic decision- making tools for (health prevention) at work
• I would only use algorithmic decision- making tools if an independent
organisation watches over my personal information
• I would like to be monitored during all my job activities, knowing it is carried out for my personal health
• I would only use an algorithmic decision-making tool if it grants me more vacation days
• I would only use an algorithmic decision-making tool if I would be financially compensated by the firm
• It is hard for me to determine my health status accurately without the help of technology
Self-tracking cultures
Some of the most agreed statements in the survey by Mettler & Wulf (2019) tangent the social dimensions of self-tracking. Despite being focused on the self, self-tracking is a “profoundly social practice” (Lupton, 2014). This is especially important to recognise in the context of the workplace, which is an inherently social setting. Because it is a social practice, it creates different types cultures, or modes. Lupton (2014) categorises the self-tracking cultures into five different modes:
Private, communal, pushed, imposed, and exploited.
The first, the private mode, is the lone enthusiast who quantifies solely for personal use, while the communal mode is when users share data, trade tips and discuss quantification with other self- trackers, as previously discussed. The pushed mode is similar to the communal mode, but here the user is encouraged by others to engage in self-tracking rather than by their own volition. In an imposed self-tracking culture, the user is compelled by others to engage in self-tracking, and the line between imposed and pushed is thin. Some elements of self-interest may still be in place, but individuals might not always have full choice over whether to self-track or not and may be coerced into participating. In the final mode, exploited, the user is not only coerced into participating, but their data is repurposed for use by third parties (Lupton, 2014).
Social effects
As it is a social practice, there may also be social repercussions for those who do not participate.
In a social context where self-management and life optimisation is encouraged, rewarded or even idealised, those who reject or do not engage in it could be disadvantaged. Objectors may be disadvantaged in terms of financial losses but may also attract moral judgement from their peers (Lupton, 2014). If, for example, an employee chooses to not engage in a wellness program that involves self-tracking, other employees might think that the person is trying to be different or
“does not want to be well“ (Moore & Robison, 2016). There are also other social effects of implementing self-tracking technologies, such as normalization. Establishing goals is an effective way to communicate the values and priorities of the firm but using self-tracking to enforce these is taking it one step further.
10
Self-tracking initiatives at work promotes normalization (Sanders, 2017). Thus, in a workplace setting, self-tracking is not a neutral technology. A self-tracking initiative acts as a means of normative power in the workplace; the technology becomes a way of governing individuals to act in certain ways via establishing norms (Foucault, 1991).
Awareness of trackers
Beyond functionalities, the awareness of the tracking technologies, and the fact that one is being tracked, may have an impact on the user. A tracking device that is too heavy, too cumbersome, uncomfortable to wear or too fragile may draw attention to it (Ruckstein, 2014), and the design of the device, and the consciousness or the lack of it, is integral to the user experience (Torgan, 2012).
In tests where users had to wear tracking devices for a longer period of time, becoming hyper- aware of the trackers on the body could make them become unpleasant, even disturbing, to wear.
Some users noted a desire to “tear off” the devices after becoming conscious of them. Neither do all who engage in self-tracking perceive the technologies as an extension of their capacities, and this can be discomforting (Buse, 2010; Freund, 2004; Ruckstein, 2014). As stated by Darmour (2014) “no one wants to be weighed down by a mountain of electronics strapped to their bodies”.
Visibility & integration
The design and visibility of the devices matter. Darmour (2014) argues that if these technologies are too obvious on a user, if they are “bolted” onto their bodies, then it’s visibility may deter usage.
Users with devices “bolted” onto their bodies may be perceived as “cyborgs” (Lamontagne, 2011), something less than human, and less obtrusive pieces may be preferable. Darmour (2014) goes on to argue that discrete devices should “recede into the background, so we can focus on our activities while still being connected”, by seamlessly integrating wearable devices into the “fabric of our lives”. Via e.g. integrated sensors in clothing and communication, and with the help of lights, colours, shapes, or vibrations rather than numbers and text (Lamontagne, 2011), the technologies would become not only wearable but also valuable and meaningful in their interaction with the user (Darmour, 2014).
Data doubles
Implementing self-tracking in the workplace creates a lot of data generated by the users’ data trail (Moore & Robinson, 2016). The information from the self-tracking devices creates a data double of the individual. Data doubles are digital duplicates of an individual in an information system (Haggerty & Ericson, 2000), and they are created by the conversion human bodies into data that can be “reassembled for the purposes of personal reflection and interaction” (Ruckenstein, 2014).
In a self-tracking context, data doubles are the digital incarnations of the self-tracker, which are constantly updated and re-configured using the tracking data (Haggerty & Ericson, 2000; Moore
& Robinson, 2016; Sjöklint et al., 2015). Interacting with the data double is a powerful way to visualise otherwise less tangible aspects of everyday life, in order to improve knowledge of self and well-being (Sjöklint et al., 2015). Yet, while it is designed for reflection, increasing the presence and importance of the data double may have profound effects.
A recursive & reflective relationship
The relationship between the individual and its data double is described as “recursive and reflective”; the self-tracking user creates a data double with data from whatever it is they measure.
By then reflecting on the data double, the user changes their behaviour accordingly. Since the data double is formed by the tracking data, changing behaviour reconfigures it. The user then again reflects and adapts their behaviour, which in turn reconfigures their data double, and so it goes.
The data doubles are therefore “both constituted by the body and self and in turn serve to reconstitute the body and self” (Lupton, 2014). Such a relationship may have far-reaching effects on the self-tracking individual.
Take the case of Kaiton Williams as an example of the strange effects self-tracking may have on the individual. In an attempt to lose weight, improve diet and exercise more, technologist and designer Kaiton Williams begun self-tracking multiple aspects of his life. After a while, Williams found that he begun trusting the derived data above his own intuition and noted that it affected him as well as his self-perception (Williams, 2013). As Williams (2013) state: “We (the Apps and I) had co-constructed a digital model of my self, and here I was, managing myself, it seems, by proxy.
When I didn’t eat ‘enough’ protein I felt weaker, and when I had too much sugar, I felt fatter.
These were delayed reactions; a re-reading of my body from the model. I’ve yet to decide: is this model pushing me closer in contact or further away from myself and my world”. Williams (2013) also notes that he changed his diet in accordance to the tracking algorithm, prioritising certain foods or recipes that best fit the capabilities of the food database.
Emotionally charged data
Self-tracking is of interest in self-optimisation because the data it generates is considered to be more objective, and therefore more reliable (Ajana, 2017; Feiler, 2014), but this does not make it emotionally neutral. The numbers presented are not just objective ‘facts’ but can be emotionally charged for many users. For e.g. patients with multiple chronic conditions, self-tracking may be a way of self-managing or sense-making, but there may also be negative emotional responses. It might be perceived as a constant reminder of sickness rather than self-improvement which can ultimately demotivate the user from engaging in self-tracking entirely (Ancker et al., 2015).
Entanglement
On a more abstract level, concerns have been raised that an increased presence of self-tracking have begun to blur the line between the online and offline self, as in the case of Kaiton Williams.
With the increased presence of these technologies, the real and digital world begin to intersect and reality itself becomes augmented (Jurgenson, 2012). As the “future normal” involves an increased amount of tracking technologies (Moore & Piwek, 2017), self-tracking devices will continue to transmute into smaller and more easily wearable configurations, which enables higher levels of quantification. As more devices are placed on, near or inside the body, and as the data doubles become more refined, it becomes less obvious where the body ends and where the technology begins. In this future, the body, the self, and the technology have become entangled (Bode &
Kristensen, 2016; Lupton, 2014; Moore & Robinson, 2016; Pink & Fors, 2017), leading to a situation where “flesh, identity, and technology are porous and intermeshed” (Lupton, 2014).
12
3 RESEARCH QUESTIONS
In this chapter, the research questions that guided this thesis are formulated and presented.
In order to achieve the purpose of investigating the use of and developing a conceptual motion analysis system to prevent the development of musculoskeletal disorders, several research questions were formulated. These were:
• What or which are the most suitable means of motion tracking within forestry?
• What ergonomic parameters should be used to analyse the postures of the tracking subjects?
• How should a self-tracking system to be used by forestry workers be designed?
o What functionalities should the system provide?
o How should the system interact and communicate with the user?
• What are the incentives for adopting a motion analysis-based MSD-prevention?
• What are some of the potential ethical issues of implementing self-tracking technologies, and how, if possible, can these be countered?
4 METHOD
In this chapter, the working process is presented. An overview of the method is described, which is followed by more detailed descriptions of each step in the process.
This thesis is based on a literature study, a qualitative interviews study, user observations and a technological evaluation. The project was initiated with a primary literature study, in which disorder rates in the relevant industries were investigated, along with the possibilities and limitations of motion tracking technologies and ergonomic standards. Interviews at PTAM were performed simultaneously. From the primary literature study and interviews at PTAM, forestry was selected as the most relevant industry, with MSDs caused by ergonomically incorrect working postures as the main issue.
Once the relevant industry had been selected, a secondary literature study was performed to attain more industry-specific information, along with literature on some ethical considerations and user perspectives on self-tracking at work, and created a frame of reference with which the results from the interviews and observations could be analysed. User observations and user interviews were carried out in parallel to the secondary literature study. This secondary literature study and results from the user observations and interviews, along with material from previous steps, created the analysis. The analysis ultimately resulted in the proposed design which was called the Ergonomic Risk Assessment service, or the ERA. The various steps of the method are described in greater detail below.
Literature study
Because motion analysis as a means to prevent musculoskeletal disorders is an interdisciplinary subject, literature in multiple fields, such as epidemiology, ergonomics, human motion analysis, and ethics, was required. More specifically, focus within the epidemiological area was on the risk factors, development, and frequency of MSDs in the relevant industries. The studies in ergonomics focused on acceptable and not recommended postures of specific body parts, as well as holding times. With regards human motion analysis, the main areas of human motion tracking were investigated, with emphasis on the possibilities and limitations of wearable sensor technologies.
In the area of ethics, some of the possible effects of introducing self-tracking at work were investigated, with special emphasis on how it could affect the individual and management.
Literature in all these areas was found using search engines such as KTH Primo and Google Scholar. Phrases that were used to find literature were: “Human motion analysis/ tracking”,
“human motion analysis review”, “injury rates construction/ forestry/ gardening”, “occupational disease/ illness construction/ forestry/ gardening”, “musculoskeletal disorders in construction/
forestry/ gardening in the EU”, “self-tracking”, “quantified self”, “self-tracking at work/
workplace”, “quantified self at work/ workplace”, “ethics of self-tracking”, “ethics of self-tracking at work”, “data doubles”, “data doubles self-tracking”, “data doubles ontology”, “self-tracking normative”. Some literature was also found using the references from certain articles, or papers referring to said article. Further literature was found by going through the publications of various government organisations (e.g. Arbetsmiljöverket and the EU-OSHA).
14 Interviews
As indicated, two kinds of interviews were carried out: PTAM interviews and user interviews.
Both types of interviews were semi-structured and qualitative (Maxwell, 2008), and an overview of all the interviews is presented in table 3. Interview guides were constructed prior the interviews to ensure that the vital questions were answered, see appendix A-F for all interview guides. Caution was taken to ensure that questions were open and not leading the interview subject. Respondents are anonymous and therefore have their names pseudonymized. A total 9 different individuals were interviewed in 7 interviews, of which 6 were at PTAM. The interviews at PTAM were carried out at an early stage in the thesis and served a few purposes. They were primarily intended to increase the understanding of the different markets by investigating who the users were in each market, what their needs, values, and attitudes were, and the possibility of applying a self-tracking system in each. This was important because PTAM makes products for construction, commercial lawn and garden, and private users, as well as forestry.
When the interviews at PTAM were carried out, focus had not yet been selected and so all markets had to be investigated in order to fairly assess the relevance of all industries. Because the focus of this thesis is forestry, only the results relevant to that industry will be presented. The results that do not pertain to forestry helped limit the scope of the thesis but were left out of the report as they were otherwise not relevant. It needs to be noted that the interviewees in Accessories and Pro- Grade Users had extensive experience with forestry. Therefore, even if their main area was not specifically forestry, they could still contribute with information and insights on the forestry industry.
A second round of interviews was also performed, but with users, once the industry had been selected. These user interviews were carried out in order to better understand the users and to test some conceptual prototypes to validate or falsify assumptions, appendix G-I for the conceptual designs used during these interviews. Insights from these interviews were used to improve the design of the system.
Table 3. Overview of the interviews, both PTAM interviews and user interviews
Primary role, affiliation, and
industry. Medium for
interview Time
[min] Deviations/ comments Project leader. PTAM.
Commercial lawn & garden. Telephone 25 -
Product specialist A. PTAM.
Pro-grade experts. Skype, mic
only 55 Double interview with product specialist B.
Product specialist B. PTAM.
Pro-grade experts. Skype, mic
only 55 Double interview with product specialist A.
Director. PTAM. Forestry. Telephone 55 -
Director. PTAM.
Construction. Skype 20+7 Interview interrupted ⅔ through Product manager. PTAM.
Accessories. Skype 55 Double interview with the developer.
Developer, R&D. PTAM.
Accessories. Skype 55 Double interview with the product manager.
Developer, R&D. PTAM.
Accessories. In person 90 Second interview with the developer.
Informal interview discussing the concept and the technology.
Arborist A. User in forestry. In person 100 Double interview with Arborist B.
Arborist B. User in forestry. In person 100 Double interview with Arborist A.
16 User observations
User observations were performed once the relevant industry had been selected, as to better understand how the users worked and what tasks were carried out during a workday. Special attention was paid to their posture during various tasks and how long it was held, as well as what kind of gear and protective equipment was used. Observing users could also help validate or falsify any statements made by the interviewees. The observations were carried out by watching footage of individuals in tree care and forestry. Some of it was educative material, on topics such as how one should take down trees, limb, and divide them, as well as how to safely climb and descend trees as an arborist. Other footage was shot by foresters or arborists displaying how they worked out in the field. Footage from competitions was also reviewed.
Several interviewees at PTAM recommended watching competitions such as the World Logging Championships or Tree Climbing Championship, as a way to observe the elite in action. While the tasks are carried out at a much greater intensity than they would in real life, they are all practical tasks ordinary workers would carry out in their everyday labour. Contestants in these competitions were judged on not only speed but also technique, safety and quality of work. All videos were found on YouTube on channels such as “Skogskunskap”, “World Logging Championship”, “Stihl”
and “Climbing Arborist”, and similar.
Technological evaluation & ergonomic parameters
A portion of the work also included an evaluation of the available methods and technologies for tracking motion. In order to assess the risk of developing MSDs due to ‘poor’ working postures, their movements of the workers must be tracked and analysed. Since no standards for motion tracking in forestry exists, the main methods and their primary means of tracking had to be evaluated, to attain the ‘best’ technology based on its suitability of application. Relevant ergonomic standards also had to be evaluated in order to synthesis fitting ergonomic parameters with which the system could interpret the postures of the user and assess their risk.
5 TECHNICAL EVALUATION
In this chapter, the technical evaluation is presented. The three main categories of motion tracking are discussed in order to fairly assess the most suitable method. A table summarising the pros and cons of each method and primary means of tracking are presented at the end.
In human motion analysis, there are three main categories of motion tracking: Visual methods, computer vision, and wearable sensors (Xu, 2018). As no standards for tracking technologies in forestry exists, the ones best suited to the context have to be selected. In order to make that decision, the main methods, along with their primary means of tracking, need to be evaluated.
Visual analysis
In visual analysis, the tracking and analysis is performed by a trained observer and is the most traditional method of motion analysis. An expert observes a subject in motion and, by using some tool or framework, assess their movement and postures, and communicate their insights to the subject. There exists a multitude of methods one can use in this assessment, see e.g. the framework by Snook & Ciriello (1991) to assess the risk of lower back disorders, the Assessment of Repetitive Tasks and Manual Handling Assessment, or The Manual Handling Method (Valero et al., 2016).
More objective visual methods & fundamental flaws
These observer-based methods have been criticized for being too subjective and other, more objective methods have been developed in response (ibid). The more objective methods commonly record a subject and then sort their posture into different, predefined poses to more objectively assess strain. Despite this, the core issue of their subjectivity persists. It is not certain that two different observers will make the same assessment, especially if they use different evaluation methods (Alwasel, 2017; Valero et al., 2016). No matter what method is used by an observer, the assessment will be fundamentally flawed because they must infer motion and posture rather than measuring them directly. This can be a significant source of error in the analysis. The nature of the data that is generated also presents issues, since it can be difficult to quantify, store and compare it objectively (Alberto, 2018; Valero et al., 2016).
Moreover, the methods are labour intensive. An expert must observe every subject to make an assessment. While the observer might be able to make a relatively accurate judgement somewhat rapidly, depending on method used, one can only observe and assess a limited number of subjects per day (Alwasel, 2017; Robert-Lachaine et al., 2019; Zhao & Obonyo, 2018). In other words, even with more objective methods, visual analysis is insufficient to accurately capture and analyse human motion on a larger scale over a longer period. One must therefore turn to the help of tracking technologies to achieve more accurate and quantifiable readings on a larger scale over a longer period of time. Out of the available technologies, computer-vision analysis is probably the most common way of gathering motion data (Aggarwal & Cai, 1999; Alberto et al., 2018; Poppe, 2007).
Computer vision
Computer vision-based movement analysis extracts data from sequential images to detect, analyse and interpret motion. The images are attained via optical sensors, often some form of camera, with various systems using anything from a cell phone to CCTV or specially designated devices (Colyer et al., 2018; Manomotion, 2019; Wang et al., 2003). These systems are usually extremely accurate, depending on placement and number of cameras, and are often used in motion capture for films or other media (Eichelberger et al., 2016).
18
Some computer vision systems require a marker to be placed on the subject for accurate results or recognise the subject. Most modern systems, however, do not need them and the current trend is moving towards markerless systems (Coyler et al., 2018). If markers are not required, then multiple subjects can be observed without any wearable hardware with a single unit, but the feasible applications of computer-vision tracking are limited, especially when it comes to forestry.
Limits of computer vision tracking
Despite computer vision-tracking being so common, there are certain limitations with it. Certain assumptions must be made with computer vision-based systems, and the most frequent assumption made is that the tracked subject remains within sight of the sensor (Moeslund & Granum, 2001), meaning that any form of activity performed out of view or in occlusion from the camera cannot be recorded and analysed. This is not a problem in a controlled environment, such as a sports arena or a recording studio. In an area with multiple obstructions, however, where the working area frequently changes and conditions such as visibility and lightning are difficult to control, this becomes an issue. These drawbacks make it unreasonable for a computer-vision based system to be used at forestry sites, because of the sheer number of cameras required. Even with only a few workers on site and a relatively static environment, multiple cameras would be required and probably have to be moved regularly to keep track of everyone. Computer-vision based solutions are therefore disregarded, leaving only the wearable sensors as viable means of tracking in forestry.
Wearable sensors & the tracker-on-a-chip
A plethora of different technologies exist within the wearable sensor category, with no single
‘perfect’ tracker. To be able to assess what the ‘best’ technology might be, it is necessary to establish the characteristics of an ideal motion tracking sensor. The ideal sensor will act as a benchmark with which all other technologies can be compared to. Welch & Foxlin (2002) denominate the ideal sensor as a “tracker-on-a-chip” (ToC), which is a theoretically ‘perfect’
tracking device. Since a solution that embodies all properties is unattainable with current technology, and because not all properties are as relevant for the intended purposes, the intention is to develop a system with as many of the pertinent properties as possible. Therefore, the properties are prioritised. The properties of the ToC, along with its priorities with regards to intended purposes on a scale from low to high priority, are as follows:
• Tiny (Medium-high) - The ToC should not be larger than a transistor or small chip.
• Self-contained (High) - No other parts need to be mounted in the environment or on the user.
• Complete (Medium-high) - The ToC is able to track all six degrees of freedom (position and orientation).
• Accurate (Medium) - A resolution better than 1mm in position and 0.1 degrees in orientation is required.
• Fast (Low) - Being able to run at 1000 Hz with a latency less than 1ms, no matter how many ToCs are in use.
• Immune to occlusion (High) - Direct line of sight to anything should not be required for the ToC.
• Robust (High) - The ToC resists performance degradation from light, sound, heat, magnetic fields, radio waves, and other ToCs in the environment.
• Tenacious (High) - The ToC tracks its target no matter how fast or far it goes.
• Wireless (Low-medium) - Being able to run for several years on a coin-sized battery is a must for the ToC.
• Cheap (Low-medium) - A ToC should cost no more than one euro per unit.
It is imperative that the system is robust, immune to occlusion, tenacious, and self-contained since it will be used in demanding environments and should not require multiple, separate devices to function. It is also important that trackers are tiny, complete, and accurate to be easy to wear and give reliable results. Fast and wireless are of low priority since latency is not crucial and it does not need to run for several years on a single battery. Cost is not prioritised as highly as other properties since it will be a product by a quality brand. Using slightly more expensive trackers is therefore acceptable so long as standards are kept high.
With the optimal tracker and the operating conditions in mind, the various technologies can be evaluated. To go through all available technologies in detail would be too extensive, so only the most common means of motion tracking will be evaluated. The most common wearable sensor technologies for motion tracking are mechanical, magnetic, acoustic, radio & microwave, and inertial sensing (Vlasic et al., 2007; Welch & Foxlin, 2002), and are described and discussed below with regards to previously mentioned properties and priorities. An overview of the technologies, their advantages and drawbacks and suitability for application in forestry is presented in table 4 at the end of the evaluation.
Mechanical tracking
Mechanical systems were the first wearable motion tracking devices to be developed, emerging in the 1940s (Valero et al., 2016). They usually measure movement with a series of two or more rigid parts moving in relation to each other. The parts are interconnected with electromechanical transducers, commonly potentiometers or shaft encoders, or even simpler devices such as goniometers. (Welch & Foxlin, 2002). These systems are highly accurate because they measure joint angles directly rather than estimating them, which most other solutions do. However, mechanical systems come with severe drawbacks. Some form of bracket or exoskeleton is required to mount the tracking devices on the user. These are generally uncomfortable to wear for longer periods of time, impede motion, and can only cover a relatively small range of motion, making them unsuitable for any form of application outside of a clinical environment (Vlasic et al., 2007;
Welch & Foxlin, 2002).
Magnetic tracking
Magnetic systems use magnetometers or current induced electromagnetic coils passing through a magnetic field, which can be a generated field or the Earth’s natural magnetic field. The resulting magnetic flux is measured and used to calculate position and orientation. It is highly accurate and insensitive to occlusion, but it is also expensive, power consuming and sensitive to disruptions.
Ferromagnetic or conductive materials, or other magnetic sensing units in the environment can interfere with the magnetic field, thereby disrupting the measures. If one uses a generated field, then the range is limited since the field strength decreases with the distance from the generator. To utilize Earth’s magnetic field, the system must be calibrated to account for local variations since the field is heterogenous (Vlasic et al., 2007; Welch & Foxlin, 2002). While magnetic trackers have some significant strengths, they are unreasonable to apply in forestry due to the sensitivity, power requirements, and potentially low tenacity.
Acoustic tracking
Acoustic motion tracking relies on the transmission of acoustic signals to assess movement by measuring time of flight. Some, usually older methods, function by mounting emitters on the users, which regularly send signals to a receiver. This makes it relatively low power but limits the tenacity of the system. Modern systems often function by bouncing soundwaves off the various body regions of the user, using the resulting doppler shift to estimate distance, which reduces the need for multiple components (Mao et al., 2016). Acoustic sensing is, however, sensitive to occlusion, although not as sensitive as computer-vision systems. Noisy environments can also disrupt the
