GENESIS-RL: GEnerating Natural Edge-cases with Systematic Integration of Safety considerations and Reinforcement Learning

Yang, Hsin-Jung; Beck, Joe; Hasan, Md Zahid; Chakraborty, Subhadeep; Wongpiromsarn, Tichakorn; Sarkar, Soumik

GENESIS-RL: GEnerating Natural Edge-cases with Systematic Integration of Safety considerations and Reinforcement Learning

File

2024-Sarkar-GENESIS-RLGEnerating.pdf (2.27 MB)

Date

2024-09-19

Authors

Yang, Hsin-Jung

Beck, Joe

Hasan, Md Zahid

Chakraborty, Subhadeep

Wongpiromsarn, Tichakorn

Sarkar, Soumik

Abstract

In the rapidly evolving field of autonomous systems, the safety and reliability of the system components are fundamental requirements. These components are often vulnerable to complex and unforeseen environments, making natural edge-case generation essential for enhancing system resilience. This paper presents GENESIS-RL, a novel framework that leverages system-level safety considerations and reinforcement learning techniques to systematically generate naturalistic edge cases. By simulating challenging conditions that mimic the real-world situations, our framework aims to rigorously test entire system's safety and reliability. Although demonstrated within the autonomous driving application, our methodology is adaptable across diverse autonomous systems. Our experimental validation, conducted on high-fidelity simulator underscores the overall effectiveness of this framework.

Academic or Administrative Unit

Mechanical Engineering

Type

Preprint

Comments

This is a preprint from Yang, Hsin-Jung, Joe Beck, Md Zahid Hasan, Ekin Beyazit, Subhadeep Chakraborty, Tichakorn Wongpiromsarn, and Soumik Sarkar. "GENESIS-RL: GEnerating Natural Edge-cases with Systematic Integration of Safety considerations and Reinforcement Learning." arXiv e-prints (2024): arXiv-2403. doi: https://doi.org/10.48550/arXiv.2403.19062. Published as Yang, Hsin-Jung, Joe Beck, Md Zahid Hasan, Ekin Beyazit, Subhadeep Chakraborty, Tichakorn Wongpiromsarn, and Soumik Sarkar. "GENESIS-RL: GEnerating Natural Edge-cases with Systematic Integration of Safety considerations and Reinforcement Learning." In 2024 IEEE International Automated Vehicle Validation Conference (IAVVC), pp. 1-8. IEEE, 2024. doi: https://doi.org/10.1109/IAVVC63304.2024.10786471.