OpenStreetMap: crowd sourcing client

(1)

2009:180 CIV

E X A M E N S A R B E T E

OpenStreetMap

- Crowd sourcing client

Mattias Grape

Luleå tekniska universitet Civilingenjörsprogrammet

Medieteknik

(2)

Abstract

Location-based mobile services/applications [1] are becoming more and more popular both on the web and in the mobile phones, according to research firm Gartner. The demand for accurate maps is rapidly growing and the possibility for the user to contribute to add details to the maps. In many parts of the world maps cost a lot of money to get and each time you want to update the maps there is a fee to update them. OpenStreetMap (OSM) is an open source project that provides free wiki-based maps.

The OpenStreetMap has an engaged and active community of mappers that constantly improve the map data. One way is to record GPS tracks and upload them to the OSM web site. On the site there is a drawing tool that can be used to draw and tag objects like road, fields, houses, rivers etc. Also Points-Of-Interests (POI) can be added to the map like bus stops, train stations, stores, restaurants, tourist objects etc.

This thesis work shows the process on how to contribute to OSM and presents different

techniques used for collecting useful data that is used to build the maps. Also the process on how the Java ME client application was made for crowd sourcing contribution to the OSM. The client

“Mobile OSM Creator” let’s you easily record GPS track points, take waypoints and record audio clips. Each of these features is saved to a file that’s in GPS eXchange Format (GPX) format, which can then later be uploaded to your OSM-account or imported to JOSM or any other map editor. Points-Of-Interests can be uploaded directly to OSM from the mobile phone by logging in with your OSM-account. The client can be customized by the user in a lot of different ways. All in all the “Mobile OSM Creator” is a feature rich application where you among many things can set GPS tracking frequency, offline/online maps and render styles for the map. This client can be downloaded from Ericsson Labs (https://labs.ericsson.com/applications).

(3)

Preface

This thesis is a part of the Master of Science in media technology education at Luleå University of Technology (LTU). The purpose of the project was to make an application to the Ericsson Labs (https://labs.ericsson.com/) collections of applications and make it open source to let the community carry on the project after my thesis work is done. The project was developed at Ericsson Research in Luleå. I would like to thank my supervisor at LTU Kåre Synnes. I would also like to thank the employees at Ericsson Research, especially my supervisor Staffan Larsson E, Sebastien Pierrel for valuable feedback from an experienced OSM contributor, and Tor Björn Minde for giving me this opportunity to do this work. Finally I would like to thank everyone that have tested the program, especially Rikard Nilsson or read parts of the thesis report and provided comments, making the work better then it would have otherwise been.

(4)

Chapter 1 Introduction

Location-based mobile services/applications [1] are becoming more and more popular both on the web and in the mobile phone, according to research firm Gartner. These services needs a map that’s up-to-date and contain a lot of details e.g. Points-Of-Interests (POI) such as bus stops, train stations, stores, restaurants, and tourist objects etc. The quality of the maps is a deal breaker whether these types of services will be a commercial success or not. The OpenStreetMap (OSM) is an open source project that was started in August 9, 2004 that tries to build maps that have good coverage and are up-to-date. OSM is not built from a company like most traditional maps, instead OSM is like Wikipedia for maps, people that want to contribute or use the map is free to do it.

In this thesis it will be presented how the process of building up the OSM map looks like.

Identifying what features are used, what features are missing and creating a crowd sourcing client in Java Micro Edition (Java ME) which includes the features that was found to be useful.

1.1 Objectives

The goal of this master thesis is to make a tool to the persons collecting data and editing the map (OpenStreetMap). The application should be easy to use, and in the same time give more

advanced options for the experienced user. A clear intuitive GUI is also desired. The thesis’ main objectives are:

• Study OpenStreetMap crowd sourcing functionalities and map drawing tools.

• Identify features and possibilities with a Java ME client tool.

• Identify and describe a possible technical solution.

• Implement a Java ME client solution.

• Focus on making the application easy to extend with new features.

1.2 Limitations

Because a thesis work has limited time frame some limitations has to be made.

You can take Global Positioning System (GPS) track with software that is on several types of devices, like dedicated GPS, laptops, Personal Digital Assistant (PDA), Personal Navigation Assistant (PNA), Mobile phones and homebrew like Sony PSP. The focus in this thesis will be on mobile phones.

Mobile phones can have an internal GPS built inside the phone, use an external GPS that you connect to the phone with a cable or use a Bluetooth GPS. GPS sends out information using either National Marine Electronics Association (NMEA) or SiRF (company name) protocol. I choose to only implement support for Bluetooth GPS, external cable GPS and NMEA protocol, because not every phone has an inbuilt GPS and it’s cheaper to buy and add GPS support using another small device than to buy a new phone with inbuilt GPS. Sony Ericsson provides 26 mobile phone models with inbuilt GPS. From the feasibility study it was found that NMEA

(9)

protocol was used by most similar software’s. Because of this NMEA protocol was selected as the communicating protocol between the GPS and the application.

Every phone model has different support of libraries with their own structure. Building an

application that is working well on all mobile phones is a big project. The application is tested on a few Sony Ericsson phones that have SE Java Platform 7 (JP-7) support or better, which is a set of profiles describing particular Sony Ericsson Java implementation.

Some features are harder to realize than others, which mean that some features had to be balanced with respect to the other desired features. Thus work had to be limited by restrict functionality of features or excluding others. For instance the feature to upload your work directly to your

account proved to be hard to implement. This feature then got limited functionality, the feature works on JP-8.5 or later Sony Ericsson phones, and to give time to add other desired features.

The application should currently be considered as a prototype which will be further refined, since the envisioned final version lies well beyond the scope of this thesis, hence the focus on making the application easy to extend.

One of the focuses in the project was to implement as many features as possible to give a better hint of what the application is capable of in its final revision, instead of making an application work on all mobile phones with few features included.

1.3 Thesis structure

The structure of the report is:

• Chapter 1 introduces the thesis, its objectives and which limitations were made.

• Chapter 2 describes the background of the OpenStreetMap project.

• Chapter 3 discussion about open source.

• Chapter 4 provide the methodology used.

• Chapter 5 goes through the process of map making.

• Chapter 6 list which features that were discussed at the start of the project.

• Chapter 7 describes which features that are included in the final version.

• Chapter 8 is about how the implementation process was done.

• Chapter 9 possible future work.

• Chapter 10 discussion and conclusions.

(10)

Chapter 2 Background

Information about OpenStreetMap, ideas of how to collect data and edit the map presented in this report is a compilation of information that can be found in OSM wiki page

http://wiki.openstreetmap.org/wiki/Main_Page [2].

2.1 What is OpenStreetMap?

OpenStreetMap (OSM) is like Wikipedia for maps. OSM creates and provides free geographic data such as street maps to anyone who wants them. The project was started because most maps you think of as free actually have legal or technical restrictions on their use, holding back people from using them in creative, productive, or unexpected ways.

Figure 1: A map over Luleå using OSM data.

Contributors to OSM take handheld Global Positioning System (GPS) devices with them on journeys, or go out specifically to record GPS tracks. They record street names, village names and other features using notebooks, digital cameras, and voice-recorders. Back at the computer, contributors upload those GPS logs showing where they traveled, and trace-out the roads on the OpenStreetMap's collaborative database. Using their notes, contributors add the street names, information such as the type of road or path, and the connections between roads. That data is then

(11)

processed to produce detailed street-level maps, which can be published freely on sites such as Wikipedia, used to create handheld or in-car navigation devices, or printed and copied without restriction. The resulting free map of the world can be viewed at http://www.openstreetmap.org, a picture of how the maps can look is shown in Figure 1.

2.1.1 How can a project like this create accurate maps?

The wiki-style process gives no guarantee that the map will be accurate. But do proprietary maps carry no errors? In fact commercial map providers have for years used “Copyright Easter Eggs [3]” that is intentional errors in the map e.g. imaginary names and wrong streets. If the owner of the proprietary map finds these errors on another map they can be used as proof that they have violated the copyright. Maps from different providers are very similar to each other and a way to protect your maps is to use Easter Eggs. Proving that a plagiarized piece is either a copy or unlawful derivative of another's effort is often very hard. It’s always a possibility the map could just be another unique work based on the same facts, making it something that is completely protected.

Easter Eggs that is used in digital maps can be used in a way that is not obvious to a user of the map. The lower bits of the geographic coordinates can be mangled in some systematic way e.g.

all the coordinates have a remainder of 3 when divided by 7, or all coordinates are displaced from their real position by some small but real offset.

All users gain in having accurate data. If few persons put in inaccurate data, maliciously or accidentally, the other 99.9% of the people can correct the errors made by these few persons that puts in inaccurate data. This is one of the corner stones in the wiki-style process. Since April 21, 2009, a full editing history is stored for each user. When errors is found you can check the editing history from that person and if you find that user deliberately puts in inaccurate data, it’s easy to find all the changes that user made and fix them.

The Wikipedia project has shown that a large amount of good quality data can be collected but it can be difficult to weed out all the inevitable errors.

2.1.2 Why is the data sometimes inconsistent?

OpenStreetMap is a free editable map that is made by its users. The map is as consistent as the people that use the map contribute to it. The database will always be subject to the whims, experimentation, and mistakes of the community. But it’s also one of the strength of OSM because it allows the data to quickly accommodate changes in the physical world.

2.1.3 Why use OSM?

Dirk Snauwaert [4] says in the article that about 15% of the roads changes every year like, new roads, one-way restrictions and new speed limits. It’s also mentioned that it’s not economically profitable to update everything that often or even track small roads. This leads to outdated maps and missing coverage for the user. For each time you want to update your maps you need to pay for it. Even if you buy maps from companies that provide maps you are not free to do whatever you want with it. You can't correct a street name, or add a pub/bar into the maps. What about sending it to a friend, enclosing it in an invitation or posting it on a notice board? A lot of these are less legal than you might think. With OSM you just correct errors and add missing features you find and everyone is the owner of the map and can use it freely.

(12)

2.2 How to contribute to OSM community

There are different ways to contribute to the OpenStreetMap.

2.2.1 Public domain map

People that have access to public domain data can add it to OSM’s database. It’s wise to contact the community before adding any public domain data to make sure there is no licensing issues or if it’s a large amount of data.

2.2.2 OpenStreetBugs

People that feel like it’s too difficult to edit the map themselves can report errors in the maps with the easy to use http://openstreetbugs.schokokeks.org/ [5] website to mark errors they see on OpenStreetMap data.

2.2.3 Simple edits

It’s not that hard to learn how to do simple edits. All that needs to be done is to create an account on the OSM website and check the Map features [6] page to get an idea of how to tag nodes.

Choose an editor among the several choices that exist and start adding things like pubs, fast food restaurants and shops, that haven’t been added to the map in your neighborhood. Correct any mistakes you see, e.g. misspelled road names.

2.2.4 Creating maps from scratch

In areas that have no coverage in the map it’s a good start to learn different mapping techniques such as recording GPS tracks and audio mapping, which is discussed in Chapter 5.1 and the process of map making that’s whole chapter 5. People can edit from tracks others have uploaded or use Yahoo! Aerial Imagery [7], if they have no access to a GPS device.

2.2.5 Collaborating with others

Mapping together with others is more fun and by working together effort can be focused and formidable milestones become achievable. 3 common ways to do mapping together are.

• Mapping Projects: People with common goals and interests collaborating together in goal oriented activities, such as cycle networks or places. These projects have a long time span.

• Mapping party: Meetings where experienced mappers help novices on how to do the mapping. The people that attend receive instructions in how to use the GPS receiver, take waypoints with it and write down notes of the waypoints to a paper. The place that’s going to be mapped is divided up among the participants. When the newcomers return back to the venue, the experienced mappers help them out to transfer the waypoints and GPS track to an OSM editor. The waypoints gets converted to POIs and gets appropriate tags from the hand written information the participants have collected and roads are built from the GPS track. The data then gets uploaded to the map. It's a very social event where people can meet up and talk (usually at a pub) in between mapping sessions. Mapping parties often last for a weekend.

(13)

• Micro mapping party: Are well defined mapping objectives where a small group of mappers meet on a casual basis. London has used this method successfully where several mappers live close to each other. These parties last from a few hours to one day.

If you want to organize your own mapping party for the first time this guide [8] can be helpful.

2.3 How is OSM build up?

Figure 2 [9] is a technical overview of logical system/components of OpenStreetMap.

Figure 2: High level overview of OSM.

Figure 2 gives a hint of how OSM components connect to each other.

The database holds a table of the current representation of the map and another table with all revisions that the map has had. The database is using the PostgreSQL software.

The Application Programming Interface (API) 0.6 is a protocol that communicates between clients like map editing software and the database server. The API is fetching and saving raw geodata from/to the OpenStreetMap database and stores user account information. The API is based on the ideas of the Representational state transfer (RESTful) API. For instance Potlatch is sending request in the form of HTTP GET, PUT, POST, and DELETE messages to the OSM database server. Requests to modify the database are authorized using HTTP Basic

Authorization.

Planet dumps are made each week of the current representation of the map on Wednesday and released on Thursday morning.

(14)

Map tiles that the main website uses to display the map are rendered from 2 different render processes called Mapnik and Osmarender.

The application that is made in this thesis work uses a few of these components. API 0.6 is used to communicate with the OSM database. Map tiles are downloaded with the API. Map tiles are OSM data that are rendered, which are used to build the map. Features in the application are made in way that can easily be used by Potlatch and Java OpenStreetMap Editor (JOSM).

If you want to know more details about what the components do visit the Component overview [10] page.

2.4 OSM community

The OSM project was started in 2004. The last 18 months, the user base has grown a lot and the map gets better and bigger for each new contributor. In the figure 3 you can see this growth in registered users and track points over time from OSM database statistics [11].

Figure 3: Users and User GPX uploads diagram.

The OpenStreetMap community is large, and spread across many locations, speaking different languages, and focusing on different interest areas. There are several different channels for the community to communicate with each other. A forum, live chat at IRC channel, blog for

announcements and interesting developments, wiki page used for reference and over 20 mailing lists. It’s in the mailing lists most of the discussions take place and decisions are made.

GPS is not an accurate tool. Each time it’s used the results it gives can be several meters from each other. When you start from scratch and building a map and use GPS as a tool to build it up,

(15)

it’s good to have a big community that all can edit the map and change it when they find errors in the map.

In the Figure 4 you can see which program that is used to create changesets that is uploaded to the OSM database, the data is from April 21, 2009 to November 25, 2009. The data comes from the changeset dump that is taken every week on OSM homepage. Every upload you do to the OSM database needs to be wrapped inside a changesets, you can read more about changesets in section 5.5.3. In the table you see the number of changesets uploaded by each program,

“with_bboxes” is the number of changesets that have a bounding box and users are the number of distinct (public) user ids.

creator | Changesets | with_bboxes | users ---+---+---+--- Potlatch | 964387 | 718768 | 54823 JOSM | 1064183 | 1015659 | 12741 Merkaartor | 121672 | 114819 | 2057 BigTinCan Upload Script | 245 | 201 | 109 osm2go | 898 | 865 | 98 iLOE | 940 | 850 | 84 Osmose Raw Editor | 714 | 330 | 78 bulk_upload | 59740 | 55051 | 58 osmtools | 7914 | 7622 | 56 Vespucci | 542 | 359 | 48 Mapzen Alpha | 310 | 164 | 37 andnav | 207 | 191 | 31 Mapzen POI Collector | 738 | 709 | 27 QGIS OSM v | 114 | 98 | 23 PythonOsmApi | 935 | 753 | 18 upload | 35026 | 33431 | 14 KMLManager | 17443 | 17313 | 9

Figure 4: Table of which program that is used to create changesets.

The online editor Potlatch that you can find on the main web page is used by most users and JOSM have most created changesets. These 2 editors are used the most by a large margin and therefore they will be evaluated in this thesis work and the crowd sourcing client application should integrate well with them.

2.5 Why is Sony Ericsson interested in OSM?

To reach a larger user base by providing several maps for example Tele Atlas and OSM. Many of the users of OSM are people that Sony Ericsson has identified as potential users of “Ericsson Labs” and by providing OSM they might be more interested in “Ericsson Labs”. Since Ericsson is doing research on crowd sourcing and using OSM in their services, they want to give

something back to the community, hence the OSM crowd sourcing client.

2.6 Ericsson Labs API

At the start of the project Ericsson had an idea that this work should use one of the APIs that is provided at “Ericsson labs” - the Mobile Maps API. Mobile Maps API provides the functionality

(16)

access to the map data itself. The maps data are streamed from a server and rendered in the mobile phone, which enables the possibility to use effective compression and transfer algorithms.

Most maps today are rendered in the server and the rendered map tiles are sent to the mobile phone application, a side effect of this is that the map has fixed scale levels. Mobile Maps does not have fixed scale levels.

(17)

Chapter 3 Why to use open source as development method

Open source involves a broad body of collaborators, typically volunteers, whose every

contribution builds on those before. Just as important, the product of this collaboration is freely available to all comers and the source code is freely shared.

Open source harnesses the distributive powers of the Internet, parcels the work out to thousands, and uses their piecework to build a better whole putting informal networks of volunteer coders in direct competition with big corporations. It works like an ant colony, where the collective

intelligence of the network supersedes any single contributor.

3.1 License

Open source software doesn't just mean access to the source code. The distribution terms of open- source software must follow some definitions [12]. A program is open source only if it uses one of the approved licenses [13], these licenses follow the definitions set by Open Source Initiative (OSI). Some examples of approved licenses are GNU General Public License, GNU Lesser General Public License and BSD license.

OpenStreetMap data is published under an open content license, with the intention of promoting free use and re-distribution of the data both commercial and non-commercial. The license used is currently the Creative Commons [14] Attribution-Share Alike 2.0 license (CC-BY-SA), however legal investigation work and community consultation is under-way to re-license the project under the Open Database License (ODbL) from Open Data Commons [15] (ODC) which is more suitable for a map dataset. OpenStreetMap software is licensed under the GNU GPL [16] if not stated with another license. What this means to you (and this isn't legal advice) is basically you can do what you like with the data, so long as you mention the original creator and the license and anyone else can do the same with anything you produce based on it. A good example on how to credit your work that is based on OSM can look like “Data/Maps Copyright 2009

OpenStreetMap Contributors | License: Creative Commons BY-SA 2.0”.

3.2 Input data

All data added to the project needs to have a license compatible with the Creative Commons Attribution-Share Alike license. This can include out of copyright information, public domain or other licenses. All contributors must register an account to OSM site to be able to upload data and by creating an account; you agree that all data you submit to the OpenStreetMap project is to be (non-exclusively) licensed under Creative Commons Attribution-Share Alike license. Use of copyright data is an especially severe problem for maps, as the only way to prove that no data came from copyright sources would be to rollback the whole area in question to before the addition of the questionable data.

(18)

Chapter 4 Method

This is the method that is used to identify which features that should be included in the Java Me crowd sourcing client.

4.1 Literature Studies

The first step is to get to know what the OSM community is about. OSM has a wiki page with a lot of information about what OSM project is, how to start mapping and which software that exist etc.

4.2 Studies of Existing Applications

After establishing background from literature studies, the next step was to test out how mapping is done today with available software for mobile phones and how to edit the collected data. The purpose of this was to identify what can be done and find drawbacks and missing features that can be implemented.

4.3 Analysis of mapping techniques

In OSM wiki page they list several mapping techniques of how to collect data. An analysis of these techniques provides valuable knowledge of what features that will be liked by the OSM community.

4.4 Identify and priority features for a Java ME client tool

After testing available applications which features they provide and read the mapping techniques of how to use these features. Its time to prioritize the features and make a list for in which order these features should be implemented in the Java ME client and identify new features that could be added.

4.5 Discussion

The priority list of features is discussed in a Brainstorm meeting with 3 employees from Ericsson research in Luleå and 1 from Kista. Sebastien Pierrel from Kista has been out mapping and has some knowledge of what can be done to make his work easier.

4.6 Iterative implementation according to these priorities

Iterative and incremental development [17] is a cyclic software development process. Key steps in the process are to start with a simple implementation of a subset of the software requirements and iteratively enhance the evolving sequence of versions until the full system is implemented.

For every iteration of the application development time, design modifications are made and new functional capabilities are added.

(19)

The project was divided into 3 iterative implementation phases:

• First phase to make Core functions to collect data and a map you can zoom and pan around in.

• Second phase to add editing tools. POI, waypoints and upload data to OSM.

• Third phase to add more features and robustness.

4.7 Evaluate each implementation phase

After each implementation phase there is an evaluation of what’s been done and problems that have come up. The feature that was in the original plan is evaluated and a new priority list is made of features to implement in next implementation phase.

4.8 Beta testing

When the last implementation phase was done a few selected persons, some with experience with other similar software, was used to beta test the application to find weakness and strengths of the program. The comments made by the beta testers gave invaluable feedback on how they use the application to make final adjustments to the program before it’s released to “Ericsson Labs”

webpage.

(20)

Chapter 5 Map Making

The process of adding data to OpenStreetMap is done in 4 steps

• Collect data. Chapter 5.1-5.3

• Upload data. Chapter 5.4

• Edit data and add tags. Chapter 5.5

• Render maps. Chapter 5.6

5.1 Mapping techniques

When mapping you need to know where you are and what are at that position. There are different types of techniques to collect data that gives this information.

5.1.1 GPS track logs

The basic technique for mapping roads, footpaths, ferry crossings, ski routes, etc. is to use a GPS and let it record a track log. If the GPS takes the timestamps in World Geodetic System 1984 (WGS-84) datum you can upload the GPS eXchange Format (GPX) track log directly to your OSM account. Some applications let you record GPX tracks with your mobile phone, laptop or PDA, if a GPS receiver is connected. Make sure that your GPS unit has the 'Lock on road' or 'Snap to road' option switched off or else you will be recording a direct copy of the map loaded into the GPS unit. This is a copyright infringement. Turn off wrapping function if your GPS has one, to avoid that the first points will be removed when the memory is full. Set the GPS to record points as densely as possible and necessary.

Two different recording rule sets are used either time or distance. With the distance rule set every point is recorded after a threshold distance “x” is reached. The good thing is that no points will get recorded while standing still. The bad things is that, unnecessary points will be recorded when traveling on straight roads and it can miss the subtleties of curves and direction changes if “x” is too large. With time rule set a points is recorded every “x” seconds. The bad thing is that when standing still, points will be saved that is not needed and when moving slowly more points will be saved then necessary. The good thing are that relative speed can be judged by the spacing, There are more points when you slow down for curves or turns and less on straight ways where you need fewer points.

5.1.2 Satellite/Arial Imagery

Many cities have high resolution aerial photos available which can be used to aid mapping. Due to copyright concerns of the derived data, only public domain or CC imagery should be used.

Yahoo’s aerial imagery the editing applet can also be used, and in fact this aids building the map later if you're using the online Potlatch editor.

Select an area you want to map from Yahoo! Maps. Use the satellite tab to get the map which you have legal rights to use. Do not print out the street map. Print out the map and go out mapping.

Make notes of street names, street types and POIs etc. A good use of Yahoo Aerial maps is to

(21)

find areas of interest to be mapped. When you get home, open up the editing applet and draw everything as you've noted down.

5.1.3 GPS Waypoints

Waypoints are single points you can log separately with your GPS receiver that mark

geographical positions of relevant objects, such as pubs, ATMs, bus stops, restaurants and many more. To take waypoints with a GPS is usually just a button push. Many GPS receivers number each waypoint. Use a notepad or sound recorder to record what that waypoint corresponds to.

While moving and saving waypoint be aware of that, a GPS receiver usually sends position information every second. Saving a waypoint can happen anywhere from exactly after you have received new position information up to just before the next information, hence almost 1 second.

This is a big factor of inaccuracy when taking waypoints while moving. In one second, you move 10 m at 36 km/h, 20 m at 72 km/h, 30 m at 108 km/h etc.

5.1.4 Photo mapping

Digital cameras are useful for recording street names, village names, and other features. Most digital cameras have a timestamp of when a photo was taken. Provided the camera clock was reasonably in sync with the GPS logger, you can get an accurate position of where the photo was shot. This information can be put to the Exchangeable image file format (Exif) header inside the picture file and thus geotagging it. JOSM includes support for reading geocoded photographs A geocoded photograph is a photo which is associated with a geographical location. The image can be associated with latitude, longitude, or to a street address. Some methods to create geocoded photographs are

• Automatic using a connected GPS: With a digital camera or camera phones that support GPS. The GPS can be built into the camera or be connected to an external GPS receiver.

With full supported geocoding when a photo is taken relevant GPS data will automatically be put into the photo’s Exif header tag.

• Synchronizing with a separate GPS: Most cameras sold today do not contain a built-in GPS receiver; however, an external location-aware device, such as a hand-held GPS logger, can still be used with a non-GPS digital camera for geocoding. The photo is taken without geographical information and is processed later using software in conjunction with the GPS data. Timestamps made by the camera can be compared to timestamps in the recorded GPS information, provided that the clocks in the separate devices can be synchronized. The resulting coordinates can then be added to the Exif header information of the photo.

• Manually geocoding: can also be used even if no GPS device was present when the photo was shot. Latitude and longitude can be entered directly to the Exif header tag or by selecting a location from a map using software tool. A less precise method is to add descriptive data of the location to an image.

A tip is to take a photo of your GPS screen/display showing the time (on the GPS) when you start mapping. You can then later on sync the time by checking what time the photo was taken and the time shown in the picture.

(22)

5.1.5 Hand drawn maps of city streets

One big problem when making a map is to get the names of every street. Draw how streets intersect and write down names for them. The sketches are then compared with your GPX track.

The quality of the sketches has to be good enough for you to spot where on the GPX track you are, to be of any use. This can be done by using the Yahoo Aerial Imagery that you print out (this map has to have no tags or names to not brake the copyright) or OSM map of streets that is already mapped but not named. Then the sketches are already done and you only have to name the streets.

5.1.6 Audio mapping

Some of the benefits of using audio mapping technique are that it’s faster than pen & paper or typing in meaningful names for waypoints. It’s the only meta data technique that is suitable while driving a car and practical to use with bike (because you might not need your hands for

recording). Give details that the GPS logs can’t give, street types and names, access restrictions, POIs, linear features parallel to your track, like rivers. But it can be confusing when the sound can have several different spellings and it’s easy to miss apostrophes. Remember to spell names that differ from the usual/common spelling when they have an equivalent pronunciation, e.g. "M- O-R-E" is better than “MORE” so you don’t accidentally add "MOOR" to the OSM database. If your area is already mapped, or if you have good satellite images, you might use audio mapping on its own, but in most cases, you will use a GPS at the same time. When you are on the road, just say what you see.

When you are audio mapping start you’re recording by saying “Mark” when you enter a new road, while using continuous recording method described later on. This is done to know exactly where a road starts. An example on how this can look like is “Mark residential road called midsommarvägen speed limit 30”.

Your audio data will be worthless if you cannot tell what recording belongs to which location.

Because the GPS track and the audio recording might not have been started simultaneously you might have to synchronize them. You may also have paused the recording part way through, so you would need to re-synchronize at that point too. There ways to do the synchronization are:

• Implicit synchronization: If the same hard- and software is used for audio recording and GPS, they will be synchronized automatically.

• Time synchronizing: Uses the timestamp of the audio recording and looks up the GPS data to know what location you where on that time. If the sync is as accurate as a delay of 1 second between audio recorder and GPS, you can correlate voice and position as

accurately as mapping can be. Record “Mark cinema at right side, called Filmstaden”.

The keyword “Mark” is used exactly when you pass the entrance of the cinema, to know where to add the POI. If the timestamps between the audio recorder and GPS is too large you have to use spatial synchronizing first to get an accurate synchronization.

• Spatial synchronizing: Gives you second accuracy even if the clock lags or the device does not take timestamp. Start the GPS and wait for a fix with full signal strength before doing the spatial synchronizing. Choose a place to do the sync that is clearly identifiable (for instance a line that is painted on the ground) and pass that point at the same velocity as when you are mapping. When passing that point you record “Mark” and pass the same

(23)

point from another direction and record “Mark” again at the exact moment that you pass it.

Along with the synchronizing problem goes the question, do you record one big audio file or make a new audio file for every clip you take. To not get a lot of noise when making one big file, pause the recording when there is nothing to say. There is a huge range of variety in audio mapping techniques 3 of them are.

• Continuous audio with GPS waypoints: Make waypoints with the GPS that saves its location, time and its number/name. At the same time you dictate onto a continuous audio recording what each waypoint represents e.g. POI or street name. Synchronization merely helps to select correct part in the audio track for each waypoint. It’s easy to locate the audio description for each POI. But you have to do some button pressing on the GPS, which can be cumbersome when moving.

• Continuous audio with contextual waypoints: Instead of entering waypoints into the GPS you make audio notes for each waypoint. If you speak out every turn you make, you might follow where you are by looking at your track log. Synchronization with this method is critical. No button pressing is needed while you are moving. But you have to be clever while you make the audio recording to not get lost in the track.

• Audio clips with waypoints: For every waypoint you make record a separate audio file.

The name of the audio clip is saved as a <link> element in the corresponding waypoint into the GPX file, to know which audio clip that belongs to each waypoint. No

synchronization is required. If your GPS can do the linking automatically and create the GPX file this method will be easy to use. Otherwise you have to use another software that does the linking.

5.1.7 Video mapping

With Video mapping you don’t have to do anything while you’re out and collecting data beside mount the video capturing device on your car and start the video and GPS recording. The disadvantages are that, processing of the collected data will take more time and other mapping techniques; you can only map what’s captured in the video and takes up a lot of storage space.

This method is rarely used by OSM mappers. Commercial companies use it and more advanced cameras to get 3D street maps.

5.1.8 Dictacam

Dictacam are a combination of audio mapping and photo mapping where you take a photo and record a voice tag to that photo. When you’re audio mapping it’s hard to correlate the audio clip with a point in the GPS track. With a picture the correct point is easier to find. Dictacam use the strengths of both photo and audio mapping which makes it suitable for POI and street names recording.

5.1.9 Focus on one type of street

To get less information to keep track on while mapping you can focus on one type of street for every recording you take. If you decide before you start tracking to follow for example a

cycleway you don’t have to take notes of when you change to another type of street. If you are on a trip and just collecting data of where roads are and not taking any notes, you can at least tag

(24)

every road you edit with the street type. Now you have less information to collect for the second trip in that area and you can easier choose a suitable transportation method for that trip because you know the street type.

5.1.10 Always turn right

When recording tracks in urban areas, a good way to cover convolved housing developmentsis to always turn right (or, always turn left in the countries that have left side traffic). It is easier to cover all side roads when using this technique. It also avoids making turns across oncoming traffic which can be both time consuming and dangerous. This means that you will catch the side roads of one side of the main street on your journey outward and on your journey back to the starting place the roads on the other side of the main street will be covered.

When you arrive at a junction you always choose to take the street that is the right side of where you stand and if you have already been to the street at the right side, go straight. If that is also covered, go left. If all turns are already covered you make a u-turn and go back the way you came.

5.1.11 Set borders

To map a city, or a big area, start by driving all the main streets with the car to divide the area into chunks. Now you can choose a chunk and use the main streets as your borders while mapping. Use the “always turn right” technique within the confines of the borders.

5.2 Mapping style

Mapping is done in two steps: First, you need to know “where” things are, mainly the streets and roads. Then you need to know “what” there is, street names, POIs or which type of road it is. You can do these one after another, or both at the same time, but you can hardly do” what” before the

“where”.

Between either collecting both where and what in one trip or split it up in two trips where you first focus on where things are and later on make a second trip to map what is at that place. This section will discuss which mapping techniques are good to use for the two mapping styles.

5.2.1 One-way mapping

Combinations of the mapping techniques GPS track logs, audio mapping, GPS waypoints and transportation method walking/bicycle are good to use when mapping in one trip. Because you will make many stops while making notes of waypoints and audio clip, walking is preferred.

Bicycle can also be used which are easy to start and stop if you are planning to make a longer trip.

5.2.2 Two-way mapping

The first trip you make the GPX track logs of where the roads are and take few GPS waypoints to note which type of road you are mapping. For this trip you want to use the fastest transportation method that is suitable for the type of road you are mapping. For the second trip you focus on taking notes of what detail that is in that area. It’s always recommended to take GPX track logs for every trip you take to get more tracks to use from when building roads.

(25)

5.3 Transportation method

Which transportation method you use make different mapping techniques appropriate. Depending on the type of tracks which you intend to create, some transportation means are better suited than others.

5.3.1 Walk

For short distance tracks that have a lot of features to map, hand drawn maps of city streets technique can be used, because you have both hands free and no vehicle that’s in your way when you stop frequently to take notes.

5.3.2 Bicycle/ Inline skates

A bicycle is good to use for medium distance tracks when you want to make several stops to take notes. Inline skates are more suited for tracking residential areas with dense features to map.

5.3.3 Car

A car is good to track all the length of a motorway, where little special features appear. If there are a lot of special features along the motorway that you want to map, 2 persons are needed, one that is the dedicated driver and the other one does the mapping work. If you are recording in time intervals, drive slower (if possible) while turning to get more densely points where the road is curving.

5.3.4 Rail

While taking a rail trip, get a good fix before boarding. Once you got the fix, the GPS receiver can get a new fix very fast within tens of milliseconds for a period of 2-4 hours. This is valuable in harsh receiver conditions such as rail cabs. It can also help (though to a lesser extent) when traveling by bus or car. If possible place the GPS receiver near a window in the same height as the window to easier get the signals from the satellites.

5.4 Upload data

Before you can upload data and edit the map you need to create an account on the OSM website and login. Your GPX file must consist of track points with valid timestamps. Tracks will fail to upload if data formats are not correct. For example GPS setup occasionally throws in a NaN as an elevation value. Deleting those lines from the file fixes the upload. GPX 1.0 and 1.1 schema is accepted, example of how they look can be seen in appendix B. OSM uses the WGS-84

latitude/longitude datum exclusively. All uploaded tracks and edits should always be in WGS-84, the default datum for most GPS receivers.

5.4.1 Save to GPX

If you have collected your track data with a GPS receiver you need to save the file in GPX format. Check the manual of your GPS receiver or software to learn how to save the track. If GPX is available, save it as GPX file. If GPX is unavailable, save the track in another format and use a converter program to change it to GPX format. One such converter program that can convert from most formats is GPSBabel.

(26)

5.4.2 Uploading to OSM

Once you have a GPX file, upload it to the web site by clicking the “GPS traces” tab. After uploading your GPX file you can download the data from the editor JOSM, along with all other data in that area. Alternatively, you can open it into the editor Potlatch, by clicking the “edit” link to the right of the trace (not the one at the top).

5.4.3 Always upload your tracks

The accuracy of the points in a single GPS track may be out by several meters. This depends on lots of factors such as the positions of the satellites when the track was taken, tree cover,

proximity to nearby buildings, position of the GPS unit relative to the centre of the road etc. If many traces are taken for the same road, then the effect of errors in any one trace will have a much smaller impact on the average position of these tracks. To help with this, it's useful if you upload all your GPS tracks to the server, even if they cover roads already in the database. This allows others to use your tracks to average out the errors and should eventually result in lower position errors.

5.5 Editing

After you have collected the required information it’s time to edit the map with your new data.

OpenStreetMap's data is made up from the elements, node, way and relations.

• )ode: A node is used to mark points of interests or as segments to build up a way. A node consists of latitude and longitude that tells where the node is located. Standalone nodes should have at least one tag that describes what point of interest it is referring to e.g.

(amenity=pub).

• Way: An ordered list of nodes of at least 2 nodes and up to 2000 nodes, displayed as connected by line segments. They are used to describe roads, paths etc. If a way is over the upper limit it will not get accepted. Just divide it up in smaller segments and group the ways in a relation of necessary. Nodes can be a part of several ways.

• Closed Way: Is a special case of the element way, which is a way that goes in a complete loop, which is used to describe areas like parks, lakes or islands.

• Relation: From the data model's perspective, relations are basically groups of objects in which each object may take on a specific role. Relations may also have tags. Typically you would expect the 'type' tag to be set, specifying what type of relation it is. Relations can be used for representing things like, traffic enforcement devices; speed cameras, red- light cameras, weight checks or destination signs at or before intersections.

Figure 5: The elements that make up the OSM data.

(27)

All these elements must have tags like describing name or type of road etc. You have a choice of several editors for adding elements and tags, 2 of them are Potlatch and Java OpenStreetMap Editor (JOSM).

As a very rough rule of thumb, Potlatch is easier for quick edits, and better for edits which range over a wide area (e.g. cycle routes). JOSM is better for heavy-duty editing in a concentrated area (e.g. housing estates) and offers facilities for audio and photo surveying. But everyone will have their own preferences.

5.5.1 General tips for adding roads or other ways

The length of a way-segment can vary greatly. If a road is extremely curvy then it needs a lot of segments/nodes. If it's dead straight, with no adjoining ways, then 1 segment is sufficient for the entire distance. Note the straight bit in Figure 6 at the center of the way requires only one segment.

Figure 6: Example on how to create a road.

If there are multiple tracks you should put the way down the center of them all, as in Figure 7, image 1. But you should consider that tracks that comes from routes that was done on foot or bicycle is more likely to be accurate than routes that are done in a car. Because the slower means of transportation has had more time for the GPS to get signals from the satellites. It's best to rely on this data to be more accurate.

GPS data is far from a perfect drawing of the path. If the dots are all over the place, draw a smooth way down the center (in Figure 7, image 2 is incorrect, image 3 is correct. The average route has been drawn cleanly down the centre). If you do the editing shortly after you have collected the GPX track it’s easier to remember how the road looks. If the dots are really scattered you should consider if you should use that track at all.

(28)

The more GPS tracks you have over the same road and average between those the more likely it’s that you have an accurate road. The points that you see from all the GPS tracks is used as a guideline of where you will put your nodes that builds up the road.

When creating and editing dual carriageways, try to keep the nodes and segments aligned and parallel, as in Figure 8, image 1. If one of the carriageways curves off then nodes don't need to be wasted on the straight carriageway, as in image 2.

Figure 8: Example on how to create dual carriageways.

When connecting two roads together they must both link to a shared node this is showed in Figure 9, image 1. A common beginner’s mistake is to place nodes very close to one another, but not form a connection Figure 9, image 2, or overlap 2 ways and presume a connection, Figure 9, image 3. Only image 1 will work. To check if both ways have a shared node, move the assumed shared node and then undo it, if both ways move you have done a successful connection.

Figure 9: How connecting 2 roads should be done.

To learn more on how to edit the maps visit Editing Standards and Conventions [18]. Many questions can be answered by looking at the work of others. Download pieces that are very well mapped and have a look. Note that some countries have different standards; preferably look at your own country.

5.5.2 Tagging

Once you have a way, you need to add tags to the way in order to see them rendered on the map.

Each element may have an arbitrary number of tags which are Key-Value pairs (e.g.

amenity=theatre). All tags need to be in lower-case, except for names. On the Map Features [6]

page you can learn more on how to tag correctly. Many of the examples on the map features page, particularly those regarding roads, are specific to the U.K. Users in other countries will want to look at Tagging [19]for guidance in applying the tagging scheme to their country.

Anyone can propose tags and it's not a requirement to have them "approved" before you can use them, but it’s unlikely the tag will show up with an icon in the maps before it is approved. The best advice is always this:

• Check the map features or any other tagging related pages on the OSM wiki webpage to see if there is a logical tag that fits your needs.

• Check the Proposed features [20] pages to pick up the tagging ideas of others.

(29)

• If you find tags from one of the above pages that match your criteria and you believe they are logical then use those.

• If what you want to tag doesn't fit any of the existing documented tags or proposed tags then make up your own using a logical process. Use your new tag(s) and add a page on the OSM wiki webpage and explain what the tag is.

No tag needs approval or to become "official" before use. Any tag key or value is valid in OSM and the approval process simply offers a mechanism to discuss tags as they become used.

5.5.3 Changeset

Every modification of one or more of the elements has to be wrapped in a changeset. Changesets can be up to 50 000 edits and alive for up to 24 hours. The changeset needs to be open before anything can be uploaded to the OSM database. Closing a changeset can be done manually or automatically if nothing is uploaded for a certain time (1 hour currently). A changeset can have tags just like the other elements. A recommended tag for changesets is to give a short description of what you have modified in the map made in that changeset.

Changesets are used for rollbacks of the OSM database. If one person is committing something that other persons of the community is questioning and thinks is wrong it’s easier to identify the changes made by that person if it’s wrapped inside a changeset instead of rolling back a whole region. It can be a very complex process to rollback a changeset therefore support for direct rollback is not available in the API. Instead rollbacks need to be made manually by reverse merging, examine the changes and undo them to obtain the desired results.

5.5.4 Potlatch online editor in OSM webpage

Potlatch is an online editor for OpenStreetMap intended as an easy-to-learn editing environment for adding new roads, railways, rivers or other features to the map and to change existing ones. A clear, uncluttered user interface is paramount for this audience.

Potlatch uses Adobe Flash, so you will need a Flash Player to use it. It is running in your standard web browser. When you go to the website OpenStreetMap [21] zoom into the streets you want to edit, and then start Potlatch with the edit tab of the main map, a screen will pop up as shown in Figure 10. You will get a choice to “Edit with save” where the update of your work happens after you save it, or “Edit live” where your changes will be uploaded directly after you make them.

Edit live is good to use at mapping parties. Edit with save is recommended for beginners.

(30)

An increasing number of experienced mappers use Potlatch as a supplement to JOSM for instant editing, and a few use it as their principal editor.

There is no need to wait for the track to be imported into the database before you start working with a track you have uploaded. In Sunday evenings uploading things to the database usually takes longer when everyone is uploading their weekend’s mapping.

One of the more powerful features is to directly convert a GPS track to a road. On the “GPS Traces” tab on the site, click the “edit” link to the right of your track as shown in Figure 11.

Figure 11: GPS Traces tab in OSM webpage, click edit convert GPS trace to road.

When you open Potlatch in this way, wait until the GPS track has loaded and a light blue line will appear when this is done. Mark “Convert GPS track to ways” box to do the automatic conversion as in Figure 12. After a few seconds, a new, bright red way will appear.

(31)

Figure 12: Login screen to Potlatch when opened with edit button.

Your waypoints will also be brought in as POI that is locked, so they do not get automatically uploaded to the database if you use “Edit live” Figure 13.

Figure 13: Example on how a waypoint from a GPX file can look when opened in Potlatch.

If you do want to upload the POI, tag it up properly and click the little padlock to the right of the POI number. When you deselect the POI, it will be uploaded.

Figure 14: How to unlock the POI.

Like the waypoints the way will also be locked. Do the normal edit work of splitting up the way, join it to existing ways and tag it. To unlock the way click the little padlock to the right of the way number. As usual, deselect to upload. If the way has more than 200 points unlocking the way will not work, split it up more before unlocking it.

(32)

Figure 15: Example on how an converted GPS trace can look like.

Potlatch is using Douglas-Peucker algorithm to simplify your GPS track to remove unnecessary points on a long straight road. If you find the automatic conversion of the GPS track to coarse, you can hold Shift when clicking “Edit with save” or “Edit live”. This will run a finer

simplification over the track.

Before you start mapping and want to use the automatic conversion feature you need to know the pitfalls you can fall in. You can’t cross the same road twice because then you will get double ways of one way. This feature is not suited for mapping a housing estate because it’s hard to avoid crossing the same road several times. But if you are mapping a normal trip from a point A to point B e.g. you record your journey from your home to the town. This feature can save you an awful lot of tracing.

Pros

It's right there on the 'Edit' tab. It allows quick, easy and immediate editing. Map changes can be sent back to the server as you make them. Yahoo! Aerial Imagery is supported without installing a plug-in. GPS traces are available as a separate layer. Automatic conversion of uploaded GPS tracks.

(33)

Cons

It's not intended for power users or those who want the speed of a desktop client. Zooming and panning prompts a map fetch.Photo and audio mapping and other file accessing features are not possible. Flash needs to be installed. It is not possible to work fully offline.

5.5.5 JOSM offline editor

JOSM is a feature-rich editor for the experienced OSM mappers. It requires a little installation and configuration efforts. You need to download the editor, java 1.5 and quite a few plugins [22]

exist that you can install. In the preferences enter your username and password to OSM account.

Start by importing the GPX file then JOSM will automatically fit the workspace to the bounding box of your GPX track. Click on the green down arrow to fetch the data from

openstreetmap.org. In the following dialog (Figure 16) ensure that the checkbox to get OSM data is enabled and press the OK button. There is also a checkbox to download every raw GPS data for that bounding box. If the track is too long, the server will refuse to send you the OSM data for that area. Zoom in a bit to restrict the request to a reasonable size.

(34)

An alternative way to start if you don’t have a GPX file is to click and zoom in to the area you want to work on. Ensure that the checkbox to get OSM data and raw GPS data is enabled and press OK button.

Figure 17: Example on how the OSM data can look like in JOSM.

Once you have drawn your line segments as in Figure 17, its time to upload your changes to OSM. Follow this procedure.

• Re-download the area you have edited. This checks that the area you have been working on has not been altered by somebody else in the time you have been editing. Click on the green down arrow .

• If there are conflicts, compare them and pick which one you think is correct. Conflicts can be viewed in the side bar section, opened with alt-c. You should be informed when

conflicts occur.

• Upload your changes. Click on the green up arrow .

• JOSM will give you an overview on the elements it's about to upload. Enter a comment that describes what you have edited.

(35)

The longer you spread uploads apart, the greater the chances are that somebody else will edit the same area. It is best to upload changes every few minutes.

After you have committed your changes to the openstreetmap.org database, it will take some time until the new ways will appear in the main map. If you want to shorten this process, you want to get familiar with the Tiles@home [23] project.

JOSM have several more features that can be used to make the editing work easier some of them are, photo- and audio mapping. Aerial satellite imagery from various sources can be traced in JOSM. This requires the installation of the WMSPlugin that allows you to fetch images from any WMS source, and display them as a background image within JOSM. High resolution imagery for major cities is available from Yahoo!

Pros

The panning and zooming of the map is fast and fluid. You can zoom in to near infinite level for super-precise mapping. JOSM can work offline using downloaded data files, and can work with local photo and GPX files. Advanced editing functionality e.g. makes nodes of a roundabout into a circle. JOSM is highly configurable and extendible via plugins [22].

Cons

Must be downloaded before it can be used and it requires Java 1.5 to work. The finer points of the interface take a while to learn.

5.6 Render maps

You have now collected data, created ways, tagged it and uploaded the changes to OSM and want to see the new map with your changes. But because map images take some time to render, the map tiles are cached, and only updated on a periodic basis, rather than immediately after you edit it. On the other hand you can download uploaded changes to OSM database immediately using for example the JOSM editor.

The default map Mapnik is updated with hourly diffs so that most data changes should get rendered within an hour. When you try to view the map for that area, it will flag the area to the rendering software. Please keep in mind that "tiles" at particular zoom levels are not updated all at once. Therefore during the update period you may see your changes at some zoom level, while at another they will be visible after a while.

Alternatively you can choose the Osmarender layer. Select this using the + icon at the top right of the map. The Osmarender tiles are updated by many OSMers using their own computers

Tiles@home [23], which has results in different response speeds - usually less than one hour. If this does not work or you like to see quicker update, you can go to informationfreeway.org [24]

web page, go to zoom level 12 and press ctrl+click on the tiles to be updated. It will then be added to the list for Tiles@home computers to render.

If your changes does not appear even after some time has gone by or you have manually requested a new render you might have a tagging problem. Check that.

• All your ways are tagged with something appropriate that will be rendered

• Tags are in lower case: HIGHWAY and Highway will not work.

(36)

Chapter 6 Features discussed at the start of the project

This chapter discusses which features the persons that are mapping could have use of in a crowd sourcing client. This was discussed in a brainstorm meeting in the beginning of the project.

Several features that were discussed at the start of the project will be presented and assigned a priority when that feature should be implemented.

6.1 Core Features

Some features have to be included to be able to have a track log of the path you traveled. Those features belong to the core features and will be implemented first.

6.1.1 Support different connections of GPS

There exist many types of GPS devices. Some are connected with Bluetooth; others are serial connected to an external GPS unit for example Sony Ericsson HGE-100. JSR 179 API produces information about the device's present physical location to Java applications. You can use the internal GPS in the mobile phones, which support JSR 179. Mobile phones which support JSR 179 are far fever than JSR 82 which is an API for Bluetooth communication. This means that the most important GPS connection to implement is support for JSR 82 and therefore be

implemented first.

6.1.2 Map

A map that you can zoom, pan and center the map on your position that is built from OSM data.

To get any context of your GPS track nodes it’s a nice feature to be able to see them plotted against a map over the area you are located in, then you get confidence that the GPS works. The map can also be used to make POI.

6.1.3 Create GPX files

Each point must have a timestamp or else the OSM database will not accept the point. All other information about the point is optional to include like altitude of the points or signal strength from the satellites. GPX, or GPS eXchange Format is an XML schema designed for describing GPS data between software applications. Waypoints can be added to the GPX file. When all points and waypoints are collected that should be included, a GPX file is created and stored in the file system.

6.2 Upload GPX track from phone to OSM account

Instead of the normal way of either storing your GPX file in the mobile phone internal- or card- memory, transfer the GPX file to the computer from the mobile phones memory and upload it to your OSM account through the OSM website. You could upload your GPX file directly to your OSM account from the phone. Then you get fewer steps to do when uploading your collected data. This is one of the main features Ericsson wanted to be implemented and have high priority when the core functions are implemented.