Perception Stack Design for the EcoCAR Electric Vehicle Challenge

Abstract

In this paper, a new sensor configuration for the 2023 Cadillac Lyriq is proposed. The SAE automation level, Operational Design Domains (ODDs), and Advanced Driver Assistance Systems (ADAS) features are considered for the proposed perception stack. Data clustering and object tracking functionality is developed for the proposed sensor fusion algorithm, testing various different combinations of sensor suites, clustering algorithms, and tracking algorithms. The operation of the proposed sensor layouts and algorithm combinations are investigated in the context of two urban scenarios and two highway scenarios. The sensor suites, driving scenarios, and algorithms were modeled and simulated using Simulink and the MATLAB Automated Driving Toolbox. The results indicate the Lyriq sensor suite with an Ouster OS1-64 LiDAR added on the front of the roof outperforms the other considered configurations.