Arena-Bench: A Benchmarking Suite for Obstacle Avoidance Approaches in Highly Dynamic Environments

The ability to autonomously navigate safely, especially within dynamic environments, is paramount for mobile robotics. In recent years, DRL approaches have shown superior performance in obstacle avoidance. However, these learning-based are often developed specially designed simulation environments and hard test against conventional planning approaches. Furthermore, the integration deployment of into real robotic platforms not yet completely solved. this letter, we present Arena-bench, a benchmark suite train, test, evaluate navigation planners on different 3D environments. It provides tools design generate highly evaluation worlds, scenarios, tasks autonomous fully integrated robot operating system. To demonstrate functionalities our suite, trained agent platform compared it variety existing model-based relevant metrics. Finally, deployed towards robots demonstrated reproducibility results. code publicly available at github.com/ignc-research/arena-bench