Lessons learned & Future of FeedMAP Final Workshop 6.10.2008 Trento, Italy Hans-Ulrich Otto Tele Atlas NV
Lessons learned - FeedMAP in-vehicle client Positional accuracy of GPS receivers differs up to 30 metres Bad GPS reception quality generates too many MDRs Difficult to detect too many deviation types at client side Current map matching has been optimised for robust locating at a give map vs. detect deviations in a map Data amount is significant -> compression gains in avg. 66%
Lessons learned - GPS receiver variation Running 5 GPS receivers in parallel in the same vehicle
Lessons learned - Location referencing There is no 100% location referencing solution Map IDs work fine for a closed system, i.e. map of one supplier is used in the whole chain Plain co-ordinates work fine between maps of the same supplier, but is not sufficient between different maps AGORA-C provides reliable MDR locations between maps from any supplier, especially useful for public authorities with own map
Lessons learned - Simulation Fully automated algorithm to simulate realistic vehicle route is challenging, e.g. vehicle decelerates on corners, stops at cross-roads, acceleration variation etc. Applied simulation considers only a subset of the real conditions -> simulated GPS errors are not always realistic Field trials provide realistic reports in statistically relevant amount -> follow up project, e.g. FOT?
Lessons learned - FeedMAP service centre Problems with geometry received MDRs Huge differences between clients caused by different map-matchers, GPS receivers and detection algorithms Deviation types WRONG_ROAD_GEOM & MISSING_ROAD are often overlapping Where to draw the border? Different clients, different thresholds? Inner-city situations -> often ambiguous Complex changes -> difficult to fully cover by an MDR
WRONG_ROAD_GEOM vs. MISSING_ROAD Depending on FM client parameters this is a wrong or a new road Situation not fully covered
The future of FeedMAP Example Tele Atlas Navigation maps are availability on a continental scale (EU, US and beyond) Tele Atlas maps (navigation quality) Asia Pacific: 2.1 Mio km America s: 12.3 Mio km Europe, Middle East, Africa: 8.6 Mio km Constant downward price pressure Economic availability of safety maps requires traditional map updating processes to be reconsidered Community feedback will play an important role in future mapping business!
Tele Atlas Map Updating Redefined Mobile Mapping Fast recording of a reality Fast extraction of map relevant information (Public) Authority Feedback Use Public Authorities and Road Operators sources Standardized interfaces required ROSATTE project (FP7) Community Feedback FeedMAP Project Feedback from TomTom users Feedback from Google users In future any partner offering such feedback
Mobile Mapping: Map Data Collection at normal car driving speed Automatic collection with sensors Stereo Imagery Positioning sensors Sensors: Slope, Banking, Laser No omissions: Everything is captured With high positional accuracy: 2m Semi-Automatic Map Data Extraction in optimally tuned office environment Automatic Object Recognition
(Public) Authority Feedback ROSATTE project 7th Framework Programme Started 01-2008 Build European infrastructure to access, exchange and maintain road safety spatial data from national/regional/local public sources Validate in cross-border Pilots
The FeedMAP concept deployed Tele Atlas will use community feedback from TomTom users as source for map updates as a result of the transaction TomTom has rolling out MapShare across all its systems TomTom users can» Correct encountered map errors in their map» Share corrected errors via TomTom home with other users» Report back errors to TomTom Map corrections and reports will be made available to Tele Atlas Thousands of MapShare reports every day Tele Atlas will also get access to map edits from Google s community of users Google Maps, Google Earth
Correcting & Reporting a map error: TomTom
Correcting & Reporting a map error: Google
Conclusions Community feedback mechanisms as developed in FeedMAP, and data sourcing with public authorities will play an important roll in the future, which will be implemented by Tele Atlas using feedback from TomTom and Google users, thus will result in a map meeting desired quality standards (for safety applications) and price requirements (for mass market), which will be available for ALL current and future Tele Atlas customers.
Questions to everybody How do you estimate the potential of map feedback aka the power of community? Have you concrete plans in this field? What do you like from the FeedMAP concept? Which areas would you like to be investigated more in detail?