Autonomous Ground Navigation in Highly Constrained Spaces: Lessons Learned From the Benchmark Autonomous Robot Navigation Challenge at ICRA 2022 [Competitions]

Abstract

The 3rd BARN (Benchmark Autonomous Robot Navigation) Challenge took place at the 2024 IEEE International Conference on Robotics and Automation (ICRA 2024) in Yokohama, Japan and continued to evaluate the performance of state-of-the-art autonomous ground navigation systems in highly constrained environments. Similar to the trend in The 1st and 2nd BARN Challenge at ICRA 2022 and 2023 in Philadelphia (North America) and London (Europe), The 3rd BARN Challenge in Yokohama (Asia) became more regional, i.e., mostly Asian teams participated. The size of the competition has slightly shrunk (six simulation teams, four of which were invited to the physical competition). The competition results, compared to last two years, suggest that the field has adopted new machine learning approaches while at the same time slightly converged to a few common practices. However, the regional nature of the physical participants suggests a challenge to promote wider participation all over the world and provide more resources to travel to the venue. In this article, we discuss the challenge, the approaches used by the three winning teams, and lessons learned to direct future research and competitions.

DRL-VO: Learning to Navigate Through Crowded Dynamic Scenes Using Velocity Obstacles

“…We further test our policy's ability to generalize to other environments and robot models. To do this, we entered our DRL-VO control policy in the ICRA 2022 BARN Challenge [7], where the goal is to navigate through unknown and highly constrained static environments, such as those shown in Fig. 13.…”

“…5) We demonstrate that our proposed control policy works effectively in highly constrained static environments and on different robot platforms without any retraining. To do this, we competed in the Benchmark Autonomous Robot Navigation (BARN) Challenge at the 2022 IEEE International Conference on Robotics and Automation (ICRA), where we placed 1st in the simulation competition and 3rd in the hardware competition [7]. 6) We make our code and our 3D human-robot interaction simulator available open source at: https://github.com/TempleRAIL/drl vo nav.…”

Arena-Rosnav 2.0: A Development and Benchmarking Platform for Robot Navigation in Highly Dynamic Environments

“…Of note, the aforementioned platforms are either limited to static environments, have a very limited amount of navigation approaches, or worlds and scenarios to test, and/or require specific hard-and software and a tedious setup proceeding, which hampers their widespread usage. The BARN challenge by Xiao et al [23] aspires to compare state-of-theart planners. The researchers provide an unified platform, which generates unknown environments to be mastered by the navigation approaches.…”

“…To develop an unified and simple-to-use API, we analyzed a large number of repositories of state-ofthe-art planners from other researchers, organizations, and challenges and provided an API that aspires to integrate these planners with as little adjustments as possible. In particular, we integrated all planners from the BARN leader-board of the 2022 BARN challenge [23] as well as other state-of-the-art learning-based approaches such as CADRL [14], CrowdNav [5], or the Cohan Planner [25]. The planners that were already present in the first version of our system, were outsourced to separate repositories.…”

“…This is in accordance with the BARN challenge, which might be helpful for outdoor navigation in unknown environments. Alongside that aspect, [29] LfLH [3] RLCA [30] MPC [31] Dragon [23] Crowdnav [5] Cohan [25] TRAIL [23] SARL [32] Arena [26] CADRL [14] Navrep [4] we also added SLAM approaches as a different and more realistic localization approach to the perfect localization and AMCL, which can be chosen by the user. The editors have a reworked user interface for easy and fast configuration and accessibility.…”