eduzhai > Applied Sciences > Engineering >

Benchmarking Metric Ground Navigation

  • king
  • (0) Download
  • 20210507
  • Save

... pages left unread,continue reading

Document pages: 6 pages

Abstract: Metric ground navigation addresses the problem of autonomously moving a robotfrom one point to another in an obstacle-occupied planar environment in acollision-free manner. It is one of the most fundamental capabilities ofintelligent mobile robots. This paper presents a standardized testbed with aset of environments and metrics to benchmark difficulty of different scenariosand performance of different systems of metric ground navigation. Currentbenchmarks focus on individual components of mobile robot navigation, such asperception and state estimation, but the navigation performance as a whole israrely measured in a systematic and standardized fashion. As a result,navigation systems are usually tested and compared in an ad hoc manner, such asin one or two manually chosen environments. The introduced benchmark provides ageneral testbed for ground robot navigation in a metric world. The Benchmarkfor Autonomous Robot Navigation (BARN) dataset includes 300 navigationenvironments, which are ordered by a set of difficulty metrics. Navigationperformance can be tested and compared in those environments in a systematicand objective fashion. This benchmark can be used to predict navigationdifficulty of a new environment, compare navigation systems, and potentiallyserve as a cost function and a curriculum for planning-based and learning-basednavigation systems. We have published our dataset and the source code togenerate datasets for different robot footprints atthis http URL.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...