CommonRoad-Game enables real-time human-in-the-loop testing for autonomous driving

Researchers Yunfei Bi and Youran Wang have released CommonRoad-Game, a lightweight human-in-the-loop simulation framework designed to test autonomous driving motion planners with real-time human interaction. Unlike existing platforms that rely on recorded datasets or lack dedicated interfaces for live participation, CommonRoad-Game is tightly integrated with the CommonRoad ecosystem. It features a multi-threaded architecture with a robust synchronization mechanism that aligns simulation time with wall-clock time, ensuring deterministic and temporally consistent interactions between autonomous and human-driven vehicles.

The framework includes a scenario generation module that records driving logs from human-in-the-loop experiments, allowing researchers to construct diverse and reproducible test cases. This makes it particularly valuable for early-stage autonomous driving research, where rapid prototyping and flexible experimentation are critical. Experimental results show that CommonRoad-Game achieves stable temporal synchronization, supports scalable multi-agent simulations, and seamlessly integrates CommonRoad-compatible motion planners to generate interactive driving scenarios. The system is computationally efficient, avoiding the high overhead of many existing human-in-the-loop simulators.

CommonRoad-Game is open-source, with its code available on GitHub. The paper has been submitted to arXiv and includes 15 pages, 18 figures, and 2 tables. This tool aims to bridge a key gap in autonomous driving research: systematically evaluating how autonomous systems behave when interacting with real human drivers in a controlled, reproducible environment.