Stefano Carpin, Todor Stoyanov, Yashodhan Nevatia, Mike Lewis and Jijun Wang
Quantitative Assessments of USARSim Accuracy
Proc. Performance Metrics for Intelligent Systems (PerMIS) Workshop, 2006
Abstract: Effective robotic simulation depends on accurate modeling of physics and the environment as well as the robot, itself. This paper describes validation studies examining feature extraction, WaveLan radio performance, and human interaction no longer provide an adequate model of the process being for the USARSim robotic simulation. All four feature extrac- studied. Simulation requirements were far more relaxed for an tion algorithms showed strong correspondences between data earlier generation of robots that relied on planning and many collected in simulation and from real robots. In each case data extracted from a well lit scene produced a closer match to data extracted from a simulated image than to camera data from a poorly lit scene. The radio simulation also performed well the interaction between robot and environment. USARSim, by in validation showing levels of attenuation due to intervening contrast, provides detailed models of both the environment walls that were similar to signal strengths measured in the and the physics of interaction making accurate simulation for modeled environment. The human-robot interaction experiments showed close correspondence between simulator and robots in performance affected by robot model, control mode and task difficulty.
Paper: [PDF (1.4MB)]
  AUTHOR = {Carpin, S. and Stoyanov, T. and Nevatia, Y. and Lewis, M. and Wang, J.},
  TITLE = {Quantitative Assessments of USARSim Accuracy},
  BOOKTITLE = {In Proc. of Performance Metrics for Intelligent Systems (PerMIS) Workshop},
  YEAR = {2006},