Skip to content

Dev Branch for SocNavBench: Social Navigation Robot Algorithm Benchmarking Tool

Notifications You must be signed in to change notification settings

CMU-TBD/tbd_SocNavBenchmark

 
 

Repository files navigation

SocNavBench

Welcome to the Social Navigation Benchmark utility (SocNavBench), a codebase for benchmarking robot planning algorithms against various episodes of containing multi-agent environments, accompanying out paper "SocNavBench: A Grounded Simulation Testing Framework for Evaluating Social Navigation".

Fig1

We provide scenarios curated from real world data for social navigation algorithms to be tested and evaluated on. sim2realZara sim2realUniv

We also provide multiple curated maps that closely resemble the environments for the pedestrian datasets. thri-maps

How To's

Install

Guide for installation at INSTALLATION.md

Use

Guide for usage at USAGE.md

Acknowledgements

This work was funded under grants from the National Science Foundation (NSF IIS-1734361) and the National Institute on Disability, Independent Living, and Rehabilitation Research (NIDILRR 90DPGE0003).

This project is also built upon the Human Active Navigation (HumANav) and Learning-Based Waypoint Navigation (Visual-Navigation) codebases. Special thanks to Varun Tolani for helping us with his projects.

Releases

No releases published

Packages

No packages published

Languages

  • Python 93.7%
  • C++ 4.4%
  • Shell 1.7%
  • GLSL 0.2%