Case Study: Aptiv harvests realistic driving scenarios from sensor data with MathWorks

Harvesting Driving Scenarios from Recorded Sensor Data at Aptiv

Aptiv needed a practical way to test ADAS and automated driving algorithms safely and repeatedly without relying on expensive real-world road testing. Working with MathWorks, the team used MATLAB and RoadRunner products to turn recorded sensor data into virtual driving scenarios for simulation and validation.

MathWorks helped Aptiv implement a three-step workflow that reconstructed the ego vehicle trajectory from GPS, IMU, camera, and map data, then rebuilt surrounding traffic using radar data and a noncausal JIPDA tracking approach, and finally generated scenarios in RoadRunner and exported them to ASAM OpenSCENARIO. The result was lane-level accurate, repeatable virtual scenarios that Aptiv could use to verify ADAS/AD closed-loop algorithms and automate scenario generation for additional sensor datasets in their simulation pipeline.

Aptiv

Krishna Koravadi

Aptiv