UTC: Safety Research Using Simulation (SAFER-SIM)
Speaker: Asher Elmquist from the University of Wisconsin-Madison

For the foreseeable future, autonomous vehicles will continue to require extensive development and evaluation demanded by difficult to replicate conditions and safety critical scenarios. Against this backdrop, this SAFER-SIM project has focused on modeling and simulation techniques for simulating sensors to test the capabilities of autonomous vehicle control strategies. Under this project, a sensor framework has been developed as a module in an open-source, multi-physics simulation platform called Chrono. By including sensing, Chrono is now capable of modeling and simulating autonomous vehicles equipped with camera, lidar, GPS, and IMU, each augmented with noise and distortion models, which are implemented for increased realism. Simulation of the camera and lidar leverages real-time ray-tracing techniques for efficient and high-fidelity sensor data generation. This simulation module is a component of a virtual environment that aims to facilitate testing that is (i) safe as it allows the virtualization of the traffic landscape, removing real people from harm’s way, (ii) versatile in that numerous simulations can be run rapidly, sweeping across edge-cases of environmental condition and traffic scenario, and (iii) low-cost in that no vehicle or sensor hardware need be deployed when testing novel algorithms or evaluating hazardous conditions.

Read the final report here: [ Ссылка ]

свернуть