An Ontology-based Model to Determine the Automation Level of an Automated Vehicle for Co-Driving

Transcription

1 An Ontology-based Model to Determine the Automation Level of an Automated Vehicle for Co-Driving Evangeline POLLARD, Philippe MORIGNOT, Fawzi NASHASHIBI IMARA Paris Rocquencourt Fusion 13, Istanbul, July 10, 2013

2 Perception sensors Intelligent vehicle + low cost production = processing unit GPS actuators HMI laser camera radar gyrometer accelerometer odometer Data fusion techniques exteroceptif proprioceptif

3 Situation assessment camera X com model scene obstacles ego-vehicle environment laser Y

4 Autonomous driving 90 s: Cybercar concept 97: automatic shuttle at Shilport airport 2010: Vislab challenge 2007: Junior 2013: Link&Go vehicle

5 Goal: Demonstratethe technicalfeasibilityof fully automated driving at speeds below 50km/h on controlled infrastructures Means: Lane detection Obstacle detection and tracking Lane keeping system ACC -5

6 The final demonstration of ABV Votre facture du mois de juillet est de euros. Veuillez contacter votre service client. Your internet bill is euros. Please contact your customer service

7 Motivation Sensors have predefined operating range and are not free of breakdowns full automated driving cannot be ensured yet at once in all situations self-assessment of the perception abilities

8 Outline Ontologydescription Automation spectrum Situation assessment Conclusion

9 Outline Ontologydescription Automation spectrum Situation assessment Conclusion

10 Ontology description Knowledge representation: «A specificationof conceptualizationof a domain knowledge» [Gruber 92] A complete semantic network. Classes, individuals, properties. Tools embed an inference mechanism.

11 Outline Ontologydescription Automation spectrum Situation assessment Conclusion

12 Automation spectrum 5 automation layers have been defined: Longitudinal Lateral Local planning Global planning Parking

13 Longitudinal control layer 0: fully driving Levels of automation in terms of decisions to make about P2 P1 Long1: Cruise control Long2: Dynamic Set Speed Type Longitudinal control P3 C1 Long3: Autonomous CC Long4: Stop&Go CLong: Cooperative cruise control communication Increasing needs in terms of perception and communication

14

15 Logical rules for automation spectrum Non-communicative vehicles Communicative vehicles

16 Outline Ontologydescription Automation spectrum Situation assessment Conclusion

17 Situation assessment

18 Logical rules depending on weather conditions Maximum automation level Front obstacle detection Ego-lane estimation Speed limit estimation

19 Combining with the current driver state Highest automation mode measured by the system (without driver consideration) Highest automation mode measured by the system (with driver consideration) In this case, an alert is given to the driver to give him more control back.

20 Conclusion An ontologyon automation layersand an ontologyon situation assessment, for co-driving(its). Inference rules to infer the former given the latter. Implementation: Classes /properties in OWL using PROTÉGÉ rules in SWRL using PELLET. Future work: betterrepresentationof rules(e.g., negation) and individuals(e.g., floatnumbers); porting on CyberCars(RTMaps and MySQL).

21 Thank you! Contacts : Evangeline POLLARD Evangeline.pollard@inria.fr imara.inria.fr