The MOSAR platform (Methods and Tools for Operational Safety Assessment and Robustness Analysis of Autonomous Vehicles) offers a methodology and a suite of tools for designing and validating the safety of autonomous vehicles through the use of a database of scenarios.
Automobile manufacturers and equipment suppliers are faced with a major scientific challenge: to demonstrate the safety of the autonomous vehicle in a complex environment, generating a very large number of driving situations in which it must react safely.
The MOSAR platform is the result of the IRT SystemX collaboration with leading French automotive players such as Renault Group, Stellantis, Valeo, AVSimulation and Expleo. This platform provides unique expertise for evaluating the behavior of an autonomous decision-making system based on scenarios derived from driving, accident statistics or expert assessments. MOSAR is available in three modules:
1. The management of scenarios representative of the situations encountered by the autonomous vehicle: the platform provides its users with capabilities enabling the management of a library of reference scenarios and the statistical analysis of situations for the design, validation and approval phases.
2. The combinatorial generation of test cases, the objective of which is to assist in the creation of a validation plan and to generate simulable test cases by exploiting the variability of the scenario parameters to represent the diversity of situations encountered by the vehicles.
3. The execution of tests in simulation and the analysis of the results and the behaviour of the autonomous system under test.