Bench-MR: A Motion Planning Benchmark for Wheeled Mobile Robots

Selection of environment types supported in Bench-MR

Abstract

Planning smooth and energy-efficient motions for wheeled mobile robots is a central task in applications ranging from autonomous driving to service and intralogistic robotics. Over the past decades, a wide variety of sampling-based motion planners, extend functions, and path-improvement techniques have been proposed for such systems. Choosing the best combination of components that fit an application is a tedious task, even for expert users. With the goal of aiding researchers in designing novel planning algorithms and evaluating path finding solutions, we present Bench-MR: the first open-source sampling-based motion-planning benchmark designed for nonholonomic, wheeled mobile robots. Unlike related benchmarks, Bench-MR offers an extensive number and variety of algorithm families, post-smoothing techniques, steer functions, optimization criteria, complex environments resembling real-world applications (such as navigating warehouses, moving in cluttered cities and parking) and performance metrics that make it a comprehensive comparison and analysis framework. Bench-MR is easy to use and extend, and as shown in the detailed examples, it significantly helps practitioners and researchers to analyze and compare their work against the state of the art.

Publication
To be published at IEEE Robotics and Automation Letters (RA-L) 2021 and presented at ICRA 2021

Related