Efficient and Effective Generation of Test Cases for Pedestrian Detection – Search-based Software Testing of Baidu Apollo in SVL - Hamid Ebadi

Authors: Hamid Ebadi, Mahshid Helali Moghadam, Markus Borg, Gregory Gay, Afonso Fontes, Kasper Socha

Affiliations: Infotiv AB · RISE Research Institutes of Sweden · Chalmers and University of Gothenburg · Lund University

Venue: IEEE Third International Conference on Artificial Intelligence Testing (AITest 2021), pp. 103–110 — 2021 IEEE Autonomous Driving AI Test Challenge

Abstract

With the growing capabilities of autonomous vehicles, there is a higher demand for sophisticated and pragmatic quality assurance approaches for machine learning-enabled systems in the automotive AI context. Simulation-based testing properly complements conventional on-road testing by enabling inexpensive testing and the ability to capture critical corner-case test scenarios.

This paper presents a study on testing the pedestrian detection and emergency braking system of the Baidu Apollo autonomous driving platform within the SVL simulator. We propose an evolutionary automated test generation technique that generates failure-revealing scenarios for Apollo in the SVL environment. Our approach models the input space using a generic and flexible data structure and benefits a multi-criteria safety-based heuristic for the objective function.

Our evaluation shows that the proposed evolutionary test case generator is more effective at generating failure-revealing test cases and provides higher diversity between the generated failures than the random baseline.

Index Terms: Search-Based Test Generation, Evolutionary Algorithm, Advanced Driver Assistance Systems, Pedestrian Detection, Automotive Simulators

Award

This paper was presented at the 2021 IEEE Autonomous Driving AI Test Challenge, where the Infotiv-led Swedish team ranked in the top 13 out of 119 teams worldwide and qualified for the final phase.

Citation

Ebadi, H., Moghadam, M. H., Borg, M., Gay, G., Fontes, A., and Socha, K. (2021). Efficient and effective generation of test cases for pedestrian detection — search-based software testing of Baidu Apollo in SVL. 2021 IEEE International Conference on Artificial Intelligence Testing (AITest), pp. 103–110.

Download PDF Download BibTeX GitHub Repository