Continuous Vehicle Slip Model Identification on Changing Terrains
Abstract
Ground robotics rely on accurate dynamic models for high performance control and estimation systems. To use odometry or predict the robots motion, an accurate model is needed for the vehicle's slip. For mobile robots, the mapping between inputs and resultant behavior depends critically on terrain conditions which vary significantly over time and space so cannot be pre-programmed. Integrated Perturbative Dynamics is used to successively identify systematic and stochastic models of vehicle slip. This is a real-time algorithm which works over arbitrary trajectories, with fast convergence along terrain boundaries, which allows for reliable operation of ground vehicles over extended periods of time in changing environments. Results are shown on a tracked surveillance robot as it drives over four distinct terrain types over 35 minutes. Fast convergence of the slip mode parameters is observed when the robot crosses terrain boundaries.
Real-Time Identification of Wheel Terrain Interaction Models for Enhanced Autonomous Vehicle Mobility
BibTeX
@workshop{Rogers-Marcovitz-2012-7543,author = {Forrest Rogers-Marcovitz and Neal Seegmiller and Alonzo Kelly},
title = {Continuous Vehicle Slip Model Identification on Changing Terrains},
booktitle = {Proceedings of RSS '12 Workshop on Long-term Operation of Autonomous Robotic Systems in Changing Environments},
year = {2012},
month = {July},
keywords = {field robotics, online learning, vehicle mobility modeling, vehicle slip, long term autonomy},
}