the Driver’s Line of Sight

(1)

EXAMENSARBETE

LISA SANDBERG KARIN SENNVALL

Visual Information in

the Driver’s Line of Sight

An Evaluation of Advanced Driver Display and Head-Up Display

MASTER OF SCIENCE PROGRAMME

Luleå University of Technology Department of Human Work Sciences

Division of Industrial Design

(2)

Abstract

Driving a car is visually highly demanding, since it includes searching for, detecting and reading information. The information shall then be interpreted and valued, as a part of the decision making process – often within a short period of time. From time to time the amount of information exceed the capacity of the driver, and this increases the risk of an accident.

Volvo Car Corporation wanted to find of if the driver’s situation, and safety, could be helped by presenting some information in a higher position than today, above the steering wheel. It was of interest to see what kind of visual information would benefit the most from this position and how the information should be designed. Two technical solutions where evaluated:

one display with perceived depth, ‘advanced driver display’ (ADD),

developed by Jaguar, and one ‘head-up display’ (HUD), designed by BMW.

Literature studies and interviews with people experienced in developing the driver area, have formed a base on which the evaluations are built.

A HUD solution is recommended for presenting information which seeks the driver’s attention and is related to the driving task, example of this is warnings and navigation help. This kind of information should be salient and placed in a high location, between 4^◦ and 8^◦ down from a horizontal line of sight in the driver’s field of view (vertically), to be quick to read and to enable the driver to use the peripheral vision to detect danger and traffic.

It is important to keep the amount of information as low as possible. A focal distance greater than 2.5 m makes it easy for the driver to refocus.

Green is recommended as the main colour and red is recommended for acute warnings. The height of alphanumerics is recommended to cover the driver’s field of view at least 0.40^◦ vertically. Since symbols often hold more information per area it is recommended that they are larger than the alphanumerics.

(3)

(4)

Preface

This master’s thesis was written at the Department of Human Work Science at Lule˚a University of Technology, during 20 weeks in the autumn 2004 and the spring 2005. The work has been carried out at Driver Information &

Interaction Design, a part of the department Research and Development at Volvo Car Corporation. The project lies within the domain of Human Machine Interaction.

We would like to thank Johannes Agardh and Robert Brostr¨om, our supervisors at Volvo Car Corporation, and Peter Bengtsson, our supervisor at Lule˚a University of Technology, for their great support and commitment.

We would also like to thank every one who helped us with our interviews at Volvo for invaluable information and many interesting viewpoints.

Last we would like to thank Jennie Nilsson and Per Blommeg˚ard. Jennie has written her master’s thesis at Volvo during the same period of time as we.

She has always been there for us in discussions and when developing ideas.

Per has been very supportive with technical matters in the process of writing the report.

Lule˚a, March 2005

Lisa Sandberg Karin Sennvall

(5)

(6)

1 Introduction

1.1 Volvo Car Corporation

Volvo Car Corporation (VCC) is owned by Ford Motor Company since 1999.

It is a part of Ford’s Premier Automotive Group (PAG), where also Jaguar, Landrover and Aston Martin are included. VCC was founded in 1927 in Gothenburg by Gustaf Larson and Assar Gabrielsson. Their fundamental idea, that all enterprises in the company are based on human beings, is still valid today. One criterion for a Volvo car is that the driver always has an unobstructed view of what is happening around him or her and that the car’s different parts work together to avoid an accident.

1.2 Background

Driving a car is visually highly demanding, since it includes searching for, detecting and reading information. The information shall then be interpreted and valued, as a part of the decision making process – often within a short period of time. From time to time the amount of information exceed the capacity of the driver, and this increases the risk of an accident.

This can mean several different things: too much information is presented at the same time and in combination with the traffic situation, the information requires too much processing, irrelevant information is more salient than relevant information and takes it’s place, information is presented in a way that requires the driver to look away from the traffic for too often or too long.

Volvo Car Corporation want to find out if the driver’s situation, and safety, can be helped by presenting some information in a higher position than today, above the steering wheel. It is of interest to see what kind of visual information would benefit most from this position and how the information should be designed.

Different display solutions, which may be a useful help to this problem, is currently coming on to the market. VCC are looking into two different solutions to see if they are capable of making the driver environment better and the driving safer in future cars, therefore this project was initiated.

One solution is a dot matrix screen, with high spatial and colour resolution, and a perceived depth. The image appears to be at a different distance than the actual (physical) image. It is developed by Jaguar and is called ‘advanced driver display’ (ADD). The other solution is an image projected onto a transparent screen. The image has high spatial and colour resolution and a

(12)

perceived depth other than the distance to the screen. This solution is a

‘head-up display’ (HUD) and is designed by BMW.

1.3 Purpose

The purpose of our assignment is to look at what information should be presented above the steering wheel, how it should be presented and to evaluate Jaguar’s ADD and BMS’s HUD to see in what ways they may be a solution in a future Volvo.

1.4 Goal

Our goal is to answer the following questions:

1. What information should be presented?

2. How should this information be placed and designed?

3. How well and in what way does Jaguar’s and BMW’s solutions meet these requirements?

1.5 Delimitations

We will only look at the visual information presented to the driver within the car today. We do not look into the technical aspects of the solutions.

Jaguar’s and BMW’s displays together with view angles in XC90 sets the physical limits for design and placement of the image/display. We will not do any work on our own to find out how much and how often different pieces of information is used. Instead we will get help from experienced people at Volvo and use our own reasoning to sort out how different kinds of information should be positioned and their importance.

(13)

2 Method

We have used the method ‘Systematic problem solving’ to guide our work. It is built up by the following steps:

• Gathering information (Necessary facts are gathered.)

• Determine the problem, investigate the problem (Facts and problems are made clear and analysed.)

• Generating ideas (A range of different ideas to solve the problem are produced.)

• Evaluate ideas (These ideas are evaluated objectively.)

• Completion (The best solution is chosen and carried out.)

To adjust this procedure, and use the parts relevant, our planning of the work follows.

2.1 Gathering information

Our most important sources of information is literature and interviews with people experienced in developming the driver area. Visits to a BMW retailer and a test drive with a HUD give us valuable experience of BMW’s solution.

2.2 Determine the problem

We use ‘the question method’, by answering the following questions the problem will take form seen from several important points of view.

1. WHAT is the problem? WHY does the problem exist?

2. WHERE is the problem? WHY is it there?

3. WHEN does the problem appear? WHY at that time?

4. WHO is the problem concerning? WHY are those concerned?

5. HOW common is the problem? WHY is it of this magnitude?

6. WHICH parts does the problem consist of? WHY these parts?

For the answers see appendix A. From this we formulate a question which will be answered in this work.

(14)

2.3 Research of the problem

The first questions were: ‘What information should be presented?’ and ‘How should this information be placed and designed?’, because the information and functions, and their position in the car, vary form model to model we focused more on different types of information than on actual pieces of information. We used literature and we interviewed people from the following departments at Volvo: Driver Information and Interaction Design, Testing Complete Vehicle, Ergonomics, Active Safety, Infotainment, Product Planning and Design.

We used the visual information presented to the driver in a V70 today as a base to form groups of information and order them after priority. To do this we looked at what kind of tasks the information was related to and how it was used.

Which information is the most important is of course dependent on the traffic situation, among other things, and since our main goal is not to classify different kinds of information we have decided to look at one situation/scenario: driving in daylight in moderate traffic, not on the highway. Every piece of information was weighted against every other piece of information in a matrix, see appendix B.

The next step was to look into what factors is most important for the ‘outer design’, such as display layout etc. To generate a list with these factors we used literature and the answers from the interviews. To find out if any of the factors might have higher priority than others we weighted each factors against all others in a matrix, in the same way as with information, see appendix B.

From the results of both of these matrices we formed a list of the most important factors to look at. The last step was to determine recommended values, were this was possible, for all of the factors so as to be able to evaluate the different solutions.

2.4 Evaluation of solutions

To evaluate each of the solutions The list of factors and their recommended values were used. When values can not be measured the judgement is of course subjective.

Some factors depend on the environment in which the solution is mounted,

(15)

2.5 Completion

From conclusions drawn both from literature and the evaluations recommendations was formed.

(16)

3 Theory

This section consists of four parts: psychological, physical and layout aspects and a description of the main principles for how the two categories of displays work. The first two parts describe the psychological and physical functions and limits of the human being, and how they are affecting the driver’s performance. The layout section is about how to design the driver environment to make the human machine interaction as good as possible and the last part is meant to go give an idea of how the two different categories of displays the solutions represents works.

3.1 Psycological aspects

This part describes the human information processing, how we perceive and attend to information from the outer world. It also describes how factors such as stress and workload affect our performance.

3.1.1 Stress

The Yerkes Dodson Law describes how performance depends on the level of arousal.

Level of arousal

Low High

Poor Good

Performance

Figure 1: The Yerkes Dodson Law. After (Wickens and Hollands, 2000).

(17)

also more and more relevant information is lost, at this point performance begin to decrease. This phenomenon is called stress-induced tunnelling. The optimal performance, the highest point of the curve in figure 1, varies with task complexity and person.For complex tasks the optimal performance is reached at a lower level of arousal than for simple tasks (Wickens and Hollands, 2000).

3.1.2 Workload

To reduce the workload the density of the information displayed should be minimized and the user should have control over the rate of information displayed, especially if it is auditory information, since it cannot be scanned or ignored (Alves-Foss, Dingus, Hulse, Jahns, Confer, Rice, Roberts,

Hanowski and Sorensen, 1996).

3.1.3 Decision making

When people are making decisions they cannot handle more than about two or three different cues. With more cues the quality of the decision will be poorer. This is not something people realize and therefore they will often want as much information as possible. When there is plenty of information at hand this can result in ‘information overload’ and poorer performance as a result.

Different pieces of information usually are of different value, but this is seldom taken into account when people weigh cues against each other. Other features, such as how salient a cue is (the more salient the more weight it is given) or how easy it is to interpret (complicated language or the need of calculations seems to give the cue less weight), is more important (Wickens and Hollands, 2000).

3.1.4 Cognitive capture

The human brain can not take in all information around us in parallel, therefore we have the ability to focus our attention on one, or a few, things at a time. The way we do this can be described as involving the use of:

selectiv attention When the observer selects appropriate aspects of the environment to process.

divided attention When the observer gives attention to two or more sources of information at the same time.

attention switching means that the observer switches attention between different sources of information.

(18)

When driving the driver has to switch and divide attention between the world outside and the instruments inside the car. This seems to be a problem when the workload is high and when objects in the outer world is less salient than the HUD (Wickens and Hollands, 2000).

3.1.5 Attention and detection

The number of objects a driver must see and keep his attention on is typically high, and the distances of the various objects from the driver vary. To present all information at one distance (as a HUD typically does) may both clutter the display and distort the driver’s perception of object distances in the scene (Wickens and Hollands, 2000).

Most drivers do not seem to recognize almost any information that lies more than 8^◦ from the direction of travel of the vehicle, and no significant difference could be noticed between people who were searching for

information and people driving without searching for anything (Cole and Hughes, 1990).

Things which are large, bright, colourful or changing (e.g. blinking) draw attention to themselves. This should be used in visual warning signals and taken into account for information which are not meant to draw attention to itself. Irrelevant signals in the field of view, or a cluttered display, is a hindrance when a critical target is to be located or responded to. Since it is mostly warning signals which shall draw attention to themselves they should preferably be colour coded and other information should be switched off during the time the warnings are shown (Wickens and Hollands, 2000).

For a visual warning to be salient the following should be true:

• At least 2 times brighter than surrounding signals.

• Less than 15^◦ eccentricity for high-priority signals.

• Must subtend at least 1^◦ of visual angle.

• Flashing against a steady background.

• High-priority signals should be coloured red (Gish and Staplin, 1995).

We tend to focus on items which are placed at the top of the display, this may have to do with the way we read, from top to bottom, left to right (Danielsson, 2001).

(19)

nonconformal. Conformal imagery work together with it’s background, or related information, and form a unit. This helps the pilot if the different pieces of information are connected and are supposed to be processed

together. This is shown as ‘object benefit’ in figure 2. Nonconformal imagery is not adjusted to it’s background or related information. This is a drawback for pieces of information which are supposed to be processed together, but if there is no connection between the different information this is of course positive since it helps prevent mix up of information. This is shown as

‘clutter cost’ in figure 2.

Figure 2: Conformal versus nonconformal imagery. After (Wickens, 1997).

For information which benefit from conformal imagery a HUD is generally a better solution than a head-down display (HDD). The opposite is true for information which benefit from nonconformal imagery. See figure 2.

In some situations a HUD solution has drawbacks even if it is best from the information point of view, see figure 3. In a situation where the pilot are to detect an unexpected event, for example another plane on a runway which are supposed to be empty, the existence of a HUD will lower the odds that the pilot will notice the other plane rather a lot compared to a situation without a HUD. For most other situations a solutions with a HUD is more

(20)

beneficial than one without it, see figure 3, (Wickens, 1997).

CONFORMAL NON- CONFORMAL

PROCESSING DETECTION EXPECTED

DETECTION SURPRICE 0

+

- HUD BENEFIT

Figure 3: Above 0, on the vertical axis, the HUD is more beneficial than the HDD, below 0 the HDD is more beneficial. After (Wickens, 1997).

3.1.7 Gestalt

Gestalt is ‘an arrangement of parts which appears and functions as a whole that is more than the sum of its parts’. To describe this phenomena one possibility is to use different factors. The factors relevant in this case is: the similarity factor, the proximity factor, the common movement, the inclusion factor and the experience factor.

The similarity factor (things which share the same properties form a gestalt without the need of being close to each other spatially), see figure 4.

(21)

The proximity factor (things which are placed close together are seen as parts of a whole and form a gestalt together), see figure 5.

Figure 5: The proximity factor. The difference in space between the squares make them into tow groups, the same goes for the buttons on the panel.

After (Mon¨o, 1997).

The common movement (if a group if things are moving together they are seen as gestalt), see figure 6.

Figure 6: The common movement. The birds common movement make them into one gestalt. After (Mon¨o, 1997).

(22)

The inclusion factor (an area enclosed by a unbroken line is seen as a whole more easily than a broken line), see figure 7.

Figure 7: The inclusion factor. The left figure is more of whole than the right figure. After (Mon¨o, 1997).

The experience factor (some gestalts, for example letters, are learned and are only recognized in a specific form seen from a specific point of view), see figure 8, (Mon¨o, 1997).

Figure 8: The experience factor. ’Y’ upside down does not mean anything, but the right way the meaning is clear. After (Mon¨o, 1997).

(23)

3.2 Physical aspects

This part describes the human eye and vision, how it works and it limitations and how these changes with age.

3.2.1 Field of vision

Figure 9: Our field of vision is built up by several zones. After (Andersson and Ekfjorden, 2000).

The field of vision can be described as having several zones. In the middle is the fixation point. Around it is the central field, where details are perceived.

We can only see sharply in the small central field of view, which covers about 2^◦of visual angle from the centre, see figure 9. Outside the central field is the functional field where we can identify information with good accuracy.

The angle of the functional field is approximately 5^◦. Outside of the

functional field is the peripheral field, it goes all the way out to 90^◦ from the centre. Here we do not see clearly or any details, but we can see movements (Andersson and Ekfjorden, 2000).

(24)

LINE OF SIGTH

RIGHT MONOCULAR FIELD 150°

BINOCULAR FIELD

120° ROTATION 30°

ACCEPTABLE EYE

EASY HEAD TURN 45° MAX. HEAD TURN 60°

LINE OF SIGHT

MAX. UPWARD ROTATION 45°

COMFORTABLE EYE ROTATION 15°

15° LINE OF SIGHT

MAX . DO

WNW ARD ROTA

TION 65°

30°

EASY HEA D TILT

Figure 10: Field of view depending on head rotation and tilt. After (Peacock and Karwowski, 1993).

0º horizon

15º line of sight at rest recommended

angle

30º acceptable angle

(25)

Our maximum field of view covers 180^◦ horizontally and 110^◦ vertically. To be able to see things with our central field of vision we rotate the eyes for small movements and both the eyes and the head for larger movements.

To rotate the eye upward or downward up to 30^◦ counts as comfortable, maximum rotation is 45^◦ upwards and 65^◦ downwards. A head tilt up to 30^◦ up or down is considered easy. Horizontally 30^◦ outward rotation is acceptable, an 45^◦ head rotation outward is easy and a 60^◦ outward rotation is a maximum head turn. See figures 10 and 11 (Peacock and Karwowski, 1993).

3.2.2 Focal distance

When the driver changes his/her focus from a far away object to a near object (or the reverse) the eyes has to accommodate. This takes some time (usually less than a second), how long depends mainly on three things: the distance between the focal points, the age of the driver and in how much detail the object has to be seen (Peacock and Karwowski, 1993).

3.2.3 Where drivers look

For a person driving a car different areas of the road scene is of more or less interest. Figure 12 shows the percentage of eye fixation in different regions of the road scene.

(26)

Figure 12: Where the driver look in the road scene. (The study was carried out in left hand traffic.) After (Cole and Hughes, 1990).

There seems to be little difference between drivers which search for information and drivers which do not, when it comes to where they direct their eyes. One thing that can be said about detection is that the most important factor is how close to the driver’s line of sight the event is. As much as 95% of detection is within 8^◦ from the centre point of their field of view. The further out the lower the probability of detection, see figure 13, (Cole and Hughes, 1990).

(27)

0 5 10 15 20 25

0,5 1,5 2,5 3,5 4,5 5,5 6,5 7,5 8,5 9,5 10,511,512,513,514,5 >15 ANGULAR SEPARATION deg

FREQENCY%

Figure 13: ‘Frequency distribution of the angular separation between fixation location and the focus of expansion (the aiming point of the direction of travel). Ninety-five percent of fixations are within 8 deg of the direction of travel.’ After (Cole and Hughes, 1990).

3.2.4 Saccadic eye movement compared to accommodation The longer distance the eyes have to move the longer time the movement takes, e.g. to move the gaze 10^◦ takes about half a second.

The time needed to accommodate depends both on how large the

accommodation is and of in how much detail the object has to bee seen. For pilots, who change focus from almost infinity to a reading distance of about 70 cm it takes almost 4 seconds to get a sharp view.

In a Japanese study the affect of the location of information (in the cars windscreen) had on sleepy drivers were measured. Information presented directly in front of the driver was compared to information presented to one side. It was found that ‘. . . when the driver reported feeling sleepy, the display 2 degrees down and directly forward kept capturing the driver’s gaze as measured by eye location.’ and ‘. . . the time at which the eyes are focused on the display is elongated in comparison with normal conditions.

This indicates that the driver’s viewpoint is unconsciously attracted to the information displayed at the front position.’ This phenomena did not occur when the display was placed aside (Weintraub and Ensing, 1992).

(28)

3.2.5 The ageing eye

Vision changes with age. The lens grows less agile and cannot focus on things as close as before. Older drivers often cannot focus on distances closer than 2 m without difficulty. The lens also tends to become more yellow which makes it harder to distinguish hues containing blue. The ability to see well in darkness also changes for the worse with age, and the time it takes to adapt to darkness increases

( http://www.hf.faa.gov/webtraining/VisualDisplays/HumanVisSys2a.htm).

3.2.6 Colour blindness

About 8% of the male population and 0.5% of the female population has a, in some way, defect colour vision. Most of the people with defective colour vision do see colours, but their response to a stimulus is not always the same as for the majority (i.e. they experience another colour). To make this as small a problem as possible some the following should be taken into account:

• Intense colours are often easier to distinguish.

• Colours put together in patterns are harder to make out.

• Do not put something red on a green background. Red and green can often be used by themselves, but not together.

• To get attentions use yellow instead of red.

• Avoid many different colours in a small area.

• Colours which are easy to read for many are black, white, yellow and blue (T¨ornqvist, 1997).

3.2.7 Target search

Target search is driven in part by cognitive factors related to the expectancy of where the target is likely to be found. Frequently watched instruments should therefore be placed in prominent locations or close to each other (Wickens and Hollands, 2000).

(29)

3.3 Layout aspects

This part describes the layout and design of the display, based on the physical and psychological aspects of the human being.

3.3.1 Placement of the display

‘If the display is placed far from the normal driving forward field of view, none of the driver’s peripheral vision can be utilized to detect unexpected movement forward of the vehicle. Another disadvantage of placing a display far away from the forward field of view is increased switching time’ (Dingus and Hulse, 1993).

The display is best placed in a high location on the instrument panel, in the area directly in front of the driver. A HUD could be a good solution since it is in the driver’s forward field of view (Alves-Foss, Dingus, Hulse, Jahns, Confer, Rice, Roberts, Hanowski and Sorensen, 1996).

3.3.2 4^◦-rule

The driver’s 180^◦ forward direct field of vision must not be obstructed down to a plane passing through the point V2, see figure ??, and declining forward 4^◦below the horizontal. The accepted exceptions are:

• ‘A’ pillars

• fixed or movable vent or side window division bars

• outside radio aerials

• rearview mirrors

• winscreen wipers

• embedded or printed ‘radio aerial’ conductors

• ‘defrosting/demisting’ conductors (VVFS 2003:22, 2003) and (77/649 EEG, 1977).

(30)

V2

Figure 14: The plane passing through the point V2 and declining forward 4^◦ below the horizontal, in XC90.

3.3.3 Digital versus analogue

If it is important that a value is read with high precision, or if it is an absolute value, a digital display is usually preferable. If it is a magnitude or the rate of change of a value which is of interest it is better to use an analogue display. This means that for example speed indicators is best presented by an analogue display (Wickens and Hollands, 2000).

3.3.4 Status report or action command

If a decision is to be made under stress or time pressure information in the form of a command is better than in the form of a status report, since the mental work of converting the status report into a decision of what action to take, as a response, need not be done. However, this is true only as long as the command given by the system is always completely reliable. When the given command is not completely reliable or if the time pressure is lesser it is in many cases better to present a status report instead.

Redundancy can be a way of getting the best from both by presenting both a status report and a command of action. It is important that they are clearly distinguished from each other, for example by being presented in different modes, otherwise it is better to choose only one of them (Wickens and Hollands, 2000).

(31)

3.3.5 Depth cues

We live in a three dimensional world and to be able to perceive depth and distances we use perceptual cues. These are divided into, observer-centered cues and object-centered cues, where observer-centered cues consist of signals from the visual system and object-centered cues are linked to the object and our knowledge of the object. Different cues has different strengths, the strength of a cue may also depend on the distance from the observer to the object, see figure 15.

Vista space

High

Low

Cue effectiveness

Depth (meters)

1 10 100 1000 10000

Action space Personal

space

Occlusion

Convergence and accommodation

Binocular Disparity

Motion parallax Height in visual field

Aerial perspective

Relative size

Texture density

Figure 15: Different depth cues and their relative importance depending on the distance to the object. After (Wickens and Hollands, 2000).

Observer-centred cues, for examples see figure 16:

Binocular disparity Each eye gets a slightly different image, the

judgement of distance is based on how large the difference between the images is. This principle is used in three-dimensional displays.

Convergence To focus on an object close by, the eyes turns slightly inwards. This is sensed by muscles and used to estimate the distance.

Accommodation To bring the image of the object into focus on the retina, muscles shape the lens and this provide information to estimate distance.

(32)

Object-centred cues,for examples see figure 16:

Linear perspective Converging lines in a picture or landscape are assumed to be parallel. (See the road lines.)

Occlusion (interposition) Our knowledge of the world tells us that when one object obscure another the obscured one is further away. (See the buildings to the right.)

Height in the plane Objects higher in the visual field are assumed to be further away.

Light and shadow Show the objects orientation and three dimensional shape.

Relative (familiar) size If we know that two objects are the same size but one look smaller we assume that object is further away. (Compare the two trucks.)

Textural gradients The texture goes from coarse to fine with growing distance. (See the field.)

Proximity-luminance covariance Objects close by are generally brighter than objects far away.

Aerial perspective Objects close by seems to be clearer than objects at a distance which seems more hazy. (See the mountains.)

Motion parallax The relative motion is greater for objects near us compared to objects at a distance (Wickens and Hollands, 2000).

(33)

Figure 16: Examples of different depth cues (Wickens and Hollands, 2000).

For a HUD in a car with a focal distance of slightly more than 2 m most important cues are those in the personal space in figure 15, namely:

Observer-centred cues:

• Binocular disparity

• Convergence

• Accommodation Object-centred cues:

• Occlusion/Interposition (See the buildings to the right in figure 16.)

• Motion parallax

• Relative (familiar) size (Compare the two trucks in figure 16.)

• Texture density (The field to the right in figure 16.)

(34)

3.3.6 Display background

ISO-standards recommend colour combinations between symbol and background in accordance to table 1 below.

Background colour Symbol colour

White Yellow Orange Red^∗, Green, Blue^∗, Black purple cyan violet

White - 0 + + ++ ++

Yellow - - 0 0 + ++

Orange 0 - - - 0 +

Red^∗, purple + 0 - - - +

Green, cyan + 0 - - - +

Blue^∗, violet ++ + 0 - - -

Black ++ ++ + + + -

++ Preferred + Recommended

0 Acceptable with high differences - Not recommended

∗Pure red and blue should be avoided because the eyes may have trouble focusing on these colours because of eye chromatic aberration.

Table 1: Recommended combinations of colours for symbols and back- grounds. After (ISO/FDIS, 2002).

If the HUD image has no background this will affect the way the image is perceived in several ways. It is impossible to choose a colour that will work in all conditions. There is an increasing risk that the image will ‘disappear’

against a heterogeneous background with an image made up of several colours. The image that meets the eye is made up out of the light from the background and the light from the projected image additively, therefore the colours of the image will change slightly with the changing background. How much depends on the strengths of the light from the image compared to the strength of the light from the background. It is probably easier to distinguish the HUD image from the background if it is monochrome.

Given that HUD symbology is represented at optical infinity, it can perceptually merge with the outside environment (Weintraub and Ensing, 1992).

3.3.7 Colour setting

(35)

3.3.8 Colour code

Colours can be processed parallel to other information. Because of this colour coding does not increase the workload almost at all but it still conveys a lot of information, more than for example shape, size or brightness does.

Because of this colour coding is a powerful form of redundant coding. Colour code also works well for tying together spatially separate areas or elements, and can help the user build a better mental model of the system this way.

Many colours has a symbolic meaning in populations, so called population stereotypes, these should always be taken into account and be used as much as possible (Wickens and Hollands, 2000).

3.3.9 Symbols

Results from empirical studies show that well designed symbols are recognized quicker and more accurately than text conveying the same message.

Symbols should be used which the user already associates with the object or idea. Many times there are icons that already exist and that have become acceptable through use over time. Also, take advantage of population stereotypes whenever possible.

Symbols should identify an object and show the ‘effect’ or action that will occur when the control is actuated. When a symbol conveys action, it is important that the resultant action of the mechanism is what is displayed.

Arrows and speed lines are examples of how we can best show action or motion.

Things which makes a symbol easier to read:

• Smooth and continuous outline.

• Closed figures.

• Symmetry

• Simplicity (Details tend to make the symbol more ambiguous and harder to interpret.)

• Do not let the symbol cover more than half the area.

• No patterns in the background.

• Use saturated, warm colours to the foreground and unsaturated, cool colours to the background (Carney, Campbell and Mitchell, 1998).

(36)

3.3.10 Alphanumerics

ISO-standards recommend heights of alphanumeric characters, measured as the subtended angel from the farthest design view point, in accordance to table 2, (ISO/FDIS, 2002).

Arcminutes Radians Suitability level

(multiplied by viewing distance)

24 arcmin. (≈ 0.40^◦) 6.98 × 10⁻³ Recommended

20 arcmin. (≈ 0.33^◦) 5.82 × 10⁻³ Acceptable if colour is a coding dimension

18 arcmin. (≈ 0.30^◦) 5.25 × 10⁻³ Acceptable if colour coding is not a coding dimension 15 arcmin. (≈ 0.25^◦) 4.36 × 10⁻³ Conditional^∗

∗When requirements for accuracy and speed of reading are modest, or when readability is incidental to the task (e.g. subscripts).

Table 2: Recommended height for alphanumeric characters (1 arcmin. = 1/60^◦). After (ISO/FDIS, 2002).

It is harder to recognize letters and digits made out of horizontal and vertical lines or dots than if they follow their conventional forms.

For short texts, one or two words long, capital letters are recommended. For longer sentences or words, lower case should be used to make the message easy to read. To make numbers easy to read they should contain no more than three or four digits (Wickens and Hollands, 2000).

3.4 The technical solutions evaluated

A description of HDD with appeared depth and HUD, their main features and examples of what they can look like.

3.4.1 Head-up display

The HUD is a system which can project information, both text and graphics, on a glass shield directly in front of the driver/pilot, see figure 17. This can be used to show instrument readings and ‘augmented reality’ superimposed on the forward view, so that the driver can concurrently monitor the display and maintain a view of the environment outside.

(37)

Figure 17: An example of a the image presented to the pilot on a HUD ( http://airlinesgate.free.fr/cdg/links.htm).

The image the driver/pilot experience does not have to be at the same distance as the surface where it is projected. In aircrafts the focal distance is set to infinity and in cars it is often at a distance greater than 2 m but less than infinity, (Gish and Staplin, 1995). In figure 18 the yellow part of the windscreen represents the ‘projection surface’ and the yellow arrow represents the lights way from the projector to the drivers eye (that is the actual image). The vertical red line represent the image the driver experience.

(38)

Figure 18: The difference between the appeared image and the actual image in a HUD.

The HUD has been in use in military air forces for some 50 years now. It was first designed to ensure that information inside and outside an aircraft could be processed simultaneously without visual scanning. Recently it has become a technical reality even for cars.

3.4.2 Three-dimensional displays

A ‘head-down display’ (HDD) which appears to have a depth is often called a three dimensional display. This means that the whole image or parts of it appears to be at a distance other than the physical display. This is done by making the display show two, slightly different, images at one time - one image fore each eye.

Normally the brain gets one image from each eye and then puts them together to make us see depth in our three dimensional world. The same occurs when the brain gets these two images from the display and the image appears to have a depth.

There are several different technical solutions for these displays, all

following the this main principle, one solutions will be described here as an example. It uses a ‘parallax barrier’ to make each eye see it’s own image.

(39)

( http://www.dgp.toronto.edu/gf/Research/Volumetric UI/3-D Displays A review of current technologies.htm).

Observer

Display

Parallax barrier

Figure 19: The eyes of the viewer, the parallax barrier and the display shown from above. With the parallax barrier at the right distance the image on the display will be divided into two image. After ( http://www.dgp.toronto.edu/gf/Research/Volumetric UI/3-D Displays A review of current technologies.htm).

Placed in a car the solution may look like figure 20 with the appeared image, the one image the driver sees at a much further distance than the physical image which is the display in the dashboard.

(40)

display appeared image

the actual image's way to the eye

Figure 20: A three dimensional display placed in a car with the physical image near the driver but the appeared image much further away.

(41)

4 Interviews

We have interviewed people from seven different departments at VCC: Driver Information and Interaction Design, Testing Complete Vehicle, Ergonomics, Active Safety, Infotainment, Product Planning and Design, to get their views and ideas of information presented above the steering wheel: Would the driver benefit from information in this position, from a safety aspect?

How may focal distance effect the driver? How much and what kind of information would the driver benefit the most from in this position? There is a, more or less, common trail of thought among the answers from all of the interviewees, which this section consists of.

All in all we interviewed eleven people, between one and three from each department. About seven of the interviewees had some kind of personal experience of different HUD solutions, mainly BMW’s. This naturally results in a way of thinking where the personal experience becomes a reference, and this must be taken into account. For all the answers see appendix C.

4.1 Information placed above the steering wheel

It can be a good solution to present information above the steering wheel.

With the right information presented in the right way at the right time, the time the driver can keep his eyes on the road can be increased.

With the information closer to where the driver looks most of the time, he/she does not have to look away far and therefore not for a long time and while he is looking at the display he can see the traffic peripherally at the same time. If the focal distance of the image is at a distance closer to the traffic and at least 2 m in front of the car it will be easier and quicker for the driver to adjust focus between the traffic and the display.

4.2 Placement

From the line of sight and 4^◦ down the driver’s view must not be obscured.

The driver should not need to look down, or to any side, at an uncomfortable angle. The driver environment is complex enough as it is today and must no be made worse. It is important that the driver feels he is in control, therefore he must be able to turn on and off the display easily and, to some extent, be able to decide what and how much information should be presented.

This will possibly decrease the driver’s workload and increase safety. The information in the display must be redundant.

High priority information has to do with traffic and driving (e.g. critical warnings, navigation and possibly speed), this kind of information should be

(42)

placed in a high position. Information with lower priority (e.g. less critical warnings, the status of various systems in the car and information connected to controls placed on the steering wheel) can be placed in a lower position.

There is also a category of information which should be put in an even lower position: settings for climate and sound, and longer text messages.

Most people thought it natural to put the information straight in front of the driver, but it can be of interest to look at the possibility of information placed more to the right of the driver (closer to the middle of the car). This position could decrease the risk of the image drawing the driver’s gaze to it, when it is unwanted. If the information was positioned closer to the middle of the car a larger area could be covered since the steering wheel would not

‘steal’ any space.

4.3 The amount of information

The mental workload must not increase. This means that the number of pieces of information shown, totally and at the same time, should be kept to a minimum. A majority thought that Jaguar shows too much information, both parallel and in total, and that maybe even BMW had to much information.

To make the information easy to read it is a good solution to use symbols as much as possible. Of course one or two words will sometimes make the meaning easier to understand and should therefore be used, but on the whole there should be no text, and absolutely no longer paragraphs or texts. It is also important that the display is not cluttered or hard to read.

To make sure that warnings do not disappear in the rest of the information they can take over part of or the whole display, for a short period of time, when they appear.

One way to see to it that the driver does not become overloaded is to show only the most important information in every situation. This could be done by splitting the driving task into three different modes: driving on a highway, driving in town and driving at night. The workload will probably be higher when driving in town than when driving on the highway and therefore the number of information items should be lower in town than on the highway.

There will possibly be differences when it comes to what kind of information is of interest in the different modes and also what kind of light settings and colour settings that will be preferred.

(43)

situation and workload dynamically and that way help the driver keep his eyes on the road.

It must be easy and quick to learn how to use the system, at least all the most common features.

(44)

5 Conclusions from theory and interviews

To be able to evaluate the two solutions we compare them to literature. To do this in a way which at the same time would allow the two solutions to, to some extent, be compared to each other a list of the most important factors were formed. This is done by reasoning based on literature, the solutions contents and our knowledge of factors important to the driver’s performance.

From different sources in literature, our own experiences of driving and in the field of HMI and in discussions with our supervisors and with people at Volvo working with HMI in cars we have drawn conclusions of our own about different aspects of the information presented to the driver, namely: ‘Amount of information’, ‘Different types of information’ and ‘Display background’.

5.1 Workload

It is very important, for safety reasons, that the workload does not exceed what the driver can handle at any time. Together the sections: ‘Amount of information’, ‘Different types of information’ and ‘Decision making’ covers important aspects of workload for a driver.

5.1.1 Amount of information

Based on theories concerning the definition of information we reason, by ourselves, as follows to adjust it to the driver’s situation in this case.

One definition of information is ‘reduction of uncertainty’ (Shannon &

Weaver, 1949). If an event, e.g. an answer to a question, reduces the original uncertainty to a great extent then it conveys a lot of information. One way of comparing the amount of information conveyed by different events is to translate the event into true-false questions. The more true-false questions needed the more information is conveyed. (Wickens and Hollands, 2000) When looking at the driver’s situation it is not enough to see to the number of fields of information, such as the current speed, navigation information etc., an important factor here is to look at how much information each field conveys and how much information is conveyed in total.

However it is not practical to convert the information in each field into true-false questions. Instead we reason that it would be possible to use the theory above as a starting point and divide the information into larger

(45)

Figure 21: The instruments showing speed, revolutions, fuel gauge, engine temperature, time etc. in a Volvo V70.

By our way of reasoning the instrument showing engine temperature holds two pieces of information, in that it answer the two questions:

1. What is the current temperature?

2. How is the temperature changing?

While the line showing the current time only holds one piece of information, answering the one question:

1. What is the current time?

This can be one way to see the maximum amount of information a display can present to the driver at one time. For example, if the driver can choose from a list of different kinds of information some combinations will hold more information than others.

5.1.2 Different types of information

We find that if the great amount of information presented to the driver in the car could be categorized into larger but fewer groups the information would be a lot easier to handle and it would be less work to add new information into the groups rather than to redo the evaluation altogether.

Following this line of thought we reason as follows.

(46)

It can be useful to group information after how it is used and it’s characteristics. Our suggestion is the following types:

• Information which seeks the driver.

• Information which the driver seeks.

• Information related to the driving task.

• Information which is used often.

Information has usually more than one of these characteristics, and we think that these combinations are the most important ones, listed from the most important to the least:

1. Information which seeks the driver. + Information related to the driving task.

2. Information which the driver seeks. + Information related to the driving task.

3. Information which the driver seeks. + Information which is used often.

4. Other information.

Our conclusion is that the information in group 1 needs to be salient, which non of the other information should be, and draw attention to itself. The information in group 2 should be placed rather high, it is important that the driver does not have to look away further than necessary, and it should be easy to read and interpret. It is good if the information in group 3 is also placed as high as possible, but it still has a lower priority than group 2, most of the time. Information in group 4 can be placed in the lowest positions. As an example some of the most common information in XC90 today would be sorted like this:

1. warnings (immediate action is needed), information (less acute warnings, immediate action is not needed), navigation instructions 2. speed, cc, navigation settings, rpm, traffic information, headlight 3. time, phone, entertainment

4. fuel gauge, trip meter

Since the outside scene is of higher priority than the HUD image (most of the time) it is important to keep unwanted information out of this view.

This means that all information except warnings and navigation should be kept well apart from the forward view. The information presented in the

(47)

5.1.3 Decision making

The driver does constantly make decisions based on information derived from the traffic and the car. As literature shows, the amount of information a person base their decision on quickly grows too large (more than three different cues) and instead of being a help to a better decision the opposite occurs. This is something we do not notice ourselves either.

Information, or cues, used for decision making are also easily biased. The less salient or harder to interpret a cue is the less important it seems. This makes it important to make the right information salient at the right moment. In a car there are a lot of information to handle and it is therefore important that the driver gets the right information for each decision and does not get more information than can be handled. Therefore, the amount of information presented to the driver above the steering wheel should be kept to a minimum and only contain high priority information, see section 3.1.3.

5.2 Placement

The further the driver has to move his/her eye the longer time it takes and the more information is missed on the way, therefore is it important not to spread out the information too much. It takes longer time to move the eyes vertical than horizontal, therefore the vertical placement is more important than the horizontal, see section 3.2.4.

It is of interest to look at the type of information. As we see it there is information that seeks the driver’s attention (for example acute warnings) and there is information which the driver seeks (for example current speed).

Information which seeks the driver’s attention should be placed closer to the driver’s line of sight than information which the driver seeks. See section 5.1.1.

Drivers do not seem to detect much information more than 8^◦ out from the line of sight, see section 3.2.3. Since 9^◦ is the lowest angel of free sight forward in XC90, a HUD could not be placed lower than this without splitting the image by having part of the image background static while the rest of the background is changing. A display with a background would not suffer from this of course. It is important that the view is not covered down to 4^◦ below the line of sight, see section 3.3.2. Therefore information which seeks the driver’s attentions should be placed between 4^◦ and 8^◦ down, see figure 22.

An angle down to 15^◦ is considered to be comfortable and down to 30^◦ is acceptable but information lower than that should be avoided, see figure 11, see section 3.2.1.

(48)

Figure 22: Examples of view angles in XC90. Nothing should obscure the driver’s view down to the 4^◦-line. Very important information which seeks the driver and might be connected to the traffic in front of the car are best placed between 4^◦ and 8^◦ down. Information with less connection to the traffic in front of the car or which does not seek the driver can be placed further down. As little information as possible should be placed lower than 30^◦ down since that is considered uncomfortable.

Information linked to the traffic should be placed closer to the outside, which lies higher in the view than information inside, and therefore higher on the display. In the same way should information linked to the car be placed lower, see section 3.3.5. Information read frequently should be placed high so that the driver quicker can move the gaze back to the road and use his/her peripheral vision while reading the information.

5.3 Focal distance

Focal distance is connected to the type and form of the information

(49)

5.3.1 Accommodation

To have a focal distance of at least 2 m is comfortable for an ageing eye since it is considerably harder and takes longer time to focus closer than 2 m at an age over 50 years, if it is possible at all, see section 3.2.5.

5.3.2 The proximity factor

The proximity factor, one of the gestalt principles, works not only in two dimensions, but in three. Therefore it should be better to place information which is connected to things in front of the car further ahead of the driver than things connected to this inside the car, see section 3.1.7.

5.3.3 Conformal and non-conformal symbology

Augmented reality benefit from a HUD since it is a question of information conformal to the background. For information which is not conformal with the background, such as odometer readings, a HUD would make it harder for the driver to read and interpret the information and it’s meaning. In this case a display with a background is a better solution.

It also seems like the HUD image is a help when it comes to detecting expected events, but makes it less probable that the driver will detect an unexpected event than if it was not there, see section 3.1.6.

5.4 Attention

The basic principle in all of our work is that, under normal driving, the most important information is what is happening in front of the car. This means that information shown above the steering wheel must not draw attentions away from what is happening in the traffic. An exception is critical warnings which might have higher priority under a short period of time since it is likely that their cause will affect the safety of the people in the car or traffic outside.

Shall the information draw attention to itself or should the information be as non salient as possible to not distract the viewer? Information which is important for the driver to notice should be made salient. Information which the driver seeks does not have be salient to be found, on the contrary it may be important that this information does not draw attention to itself to decrease the risk for distraction.

(50)

5.5 Main colour setting

Colours should not be used without cause and the total number of colours should not exceed 5 ± 2, see section 3.3.7.

To make the display easy to read it is important to make sure that the contrast between the symbols and the background is good. A table of recommended combinations of colour can be seen in table 1, section 3.3.6.

5.6 Colour code

Colours should not be used without cause and the total number of colours should not exceed 5 ± 2. Cultural stereotypes should be taken into account and used where appropriate, e.g. warning signals etc, see section 3.3.8.

5.7 Colour blindness

It seems like colour blindness is not a big problem for displays with background, but could be larger for displays without background since patterns and many different colours in a small area often is a problem, see section 3.2.6.

5.8 Size

On account of the gestalt principles the size of the display sets a limit to the amount of information which could be shown at the same time. Different pieces of information presented too close to each other will be read and interpreted as if they are linked even if this is not the case, see section 3.1.7.

To be easy to read alphanumerics should at least cover the driver’s field of view 24 arcmin. (≈ 0.40^◦) vertically, at the greatest distance they are intended to be read, see table 2 in section 3.3.10.

Since symbols often holds more information per area unit it is reasonable to make them higher than the text.

5.9 Display background

Our conclusion is that the benefit of a background, in contrast to having no

(51)

front of the car benefit from a solution without a background but information linked to the inside of the car benefits from a solution with background.

At the same time it is easier to read a display with at static background without patterns, see section 3.2.6.

5.10 Symbols and alphanumerics

Well designed symbols are recognized quicker and more accurately than text conveying the same message. Sometimes a few words are necessary to make the symbol unambiguous or add a more specific meaning to it. The text should be short, one or two words, see section 3.3.9.

Digital displays are best if it is important to read the value precisely or if it is an absolute value. An analogue display is best if a magnitude or the change of a value that is important. This means that for example speed is best presented on a analogue display, see section 3.3.3.

The height of alphanumerics is recommended to be at least 24 arcmin.

(≈ 0.40^◦), see table 2 in section 3.3.10. Since symbols often holds more information per area it is logical that they should be larger than the alphanumerics.

(52)

6 The two solutions

In this section’s first part Jaguar’s solution; the advanced driver display (ADD), is described in it’s environment, mounted in a Jaguar. It is also evaluated and compared to literature. In the second part of this section BMW’s solution in the form of a HUD is described in it’s environment and evaluated and compared to literature.

6.1 Description of Jaguar’s ADD

Jaguar’s ‘advanced driver display’ (ADD) is a head-down display placed at the top of the dashboard, behind the steering wheel, from the driver’s point of view, see figure 23. The display has a focal distance of 2 m and magnifies the image 1.78 times by using a prism. The tilt adjustment of the display is

±2.5^◦.

display appeared image

the actual image's way to the eye

Figure 23: A sketch of the principle for jaguar’s glance-Down display.

The width of the display is 184 mm and the height is 45 mm. It is divided into four different areas, see figure 24.

(53)

Figure 24: Schematic picture over the display layout.

All information shown in the ADD is redundant, it is presented in other displays at lower positions in the car. Information shown in the different areas of the display, see also figure 24:

A (the driver can chose one or two pieces of information from the list, depending on the size of the item/items):

• Tachometer (default setting for manual mode)

• Changes to ICE settings (in car entertainment)

• Clock

• Compass

• Navigation turn-by-turn

• Nothing (makes it possible to leave this space blank if desired)

• Trip information

• Voice control (on/off)

• Red warnings, which are acute, and the driver needs to act immediately too.(will overrule other information in the scratchpad)

• Amber warnings, which are less acute. (will overrule other information in the scratchpad)

• Blue warnings, which are less acute than the amber warnings. (will overrule other information in the scratchpad)

B:

• Gear (if the car is in manual mode)

• Warning reminders

• Headway adjust reminder

(54)

C1:

• Speedometer C2:

• Active cruise control settings (including headway adjust)

• Active speed limiter The whole display:

• Rear view camera (the display changes automatically to camera view when the car is in reverse)

Classified

(55)

Classified

(56)

Classified

the Driver’s Line of Sight

EXAMENSARBETE

LISA SANDBERG KARIN SENNVALL

Visual Information in

the Driver’s Line of Sight

An Evaluation of Advanced Driver Display and Head-Up Display

MASTER OF SCIENCE PROGRAMME

Abstract

Preface

Contents

1 Introduction

1.1 Volvo Car Corporation

1.2 Background

1.3 Purpose

1.4 Goal

1.5 Delimitations

2 Method

2.1 Gathering information

2.2 Determine the problem

2.3 Research of the problem

2.4 Evaluation of solutions

2.5 Completion

3 Theory

3.1 Psycological aspects

3.2 Physical aspects

3.3 Layout aspects

3.4 The technical solutions evaluated

4 Interviews

4.1 Information placed above the steering wheel

4.2 Placement

4.3 The amount of information

5 Conclusions from theory and interviews

5.1 Workload

5.2 Placement

5.3 Focal distance

5.4 Attention

5.5 Main colour setting

5.6 Colour code

5.7 Colour blindness

5.8 Size

5.9 Display background

5.10 Symbols and alphanumerics

6 The two solutions

6.1 Description of Jaguar’s ADD