www.pegasusprojekt.de Testing and Safeguarding – Stand 19a System Architecture, Algorithms, Software and Hardware iMAR Navigation develops and provides in PEGASUS solutions for real-time moni- toring and validation of test runs via pose estimation and scene interpretation using INS/GNSS technology and binocular vision (with and without a-priori known maps). Challenge: • Monitoring (real life) test runs on proving grounds using a multi-sensor system implemen- ted on the roof of all vehicles associated with the test run. • Detection of ego pose deviations and scene interpretation. System Architecture: • Implementation of multi-tier software architecture (driver, data, algorithms, visualiza- tion, appplication). Multi-Sensor Head: • Incorporating of RTK INS/ GNSS, binocular vision and high performance mesh communication module. Algorithms: • Visual Odometry for pose esti-mation, based on stereo image sequence analysis. • Machine Learning (Convolu- tional Neural Networks) for semantic interpretation of vehicle surrounding. • Scene reconstruction com- bined with Machine Learning for estimation of infrastruc- ture on proving grounds in order to estimate ego pose based on a priori generated maps. iMAR Doc-No. DOC171018146, Rev. 1.01, 10/2017 TOOLS FOR PROVING GROUND AND FIELD TESTS LOCALIZATION AND COLLISION AVOIDANCE
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
www.pegasusprojekt.de
Testing and Safeguarding – Stand 19a
System Architecture, Algorithms, Software and Hardware
iMAR Navigation develops and provides in PEGASUS solutions for real-time moni-toring and validation of test runs via pose estimation and scene interpretation usingINS/GNSS technology and binocular vision (with and without a-priori known maps).
Challenge:• Monitoring (real life) test runs
on proving grounds using a multi-sensor system implemen-ted on the roof of all vehiclesassociated with the test run.
• Detection of ego posedeviations and sceneinterpretation.
System Architecture:• Implementation of multi-tier
software architecture (driver, data, algorithms, visualiza-tion, appplication).
Multi-Sensor Head:• Incorporating of RTK INS/
GNSS, binocular vision andhigh performance meshcommunication module.
Algorithms:• Visual Odometry for pose
esti-mation, based on stereo image sequence analysis.
• Machine Learning (Convolu-tional Neural Networks) forsemantic interpretation ofvehicle surrounding.
• Scene reconstruction com-bined with Machine Learning for estimation of infrastruc-ture on proving grounds in order to estimate ego posebased on a priori generatedmaps.
iMAR Doc-No. DOC171018146, Rev. 1.01, 10/2017
TOOLS FOR PROVING GROUND AND FIELD TESTS
LOCALIZATION AND COLLISION AVOIDANCE
Related Documents
