Experimental Study of Track-Soil Interactions of the Steering Performance of Tracked Robots over Soft Deformable Terrains

Abstract

The research of track-soil interaction modeling over soft deformable terrains is an important direction in terramechanics. Wong proposed a general theory of skidsteering tracked vehicle model by focusing on the trackterrain interaction of the bottom surface of tracks while neglecting the resistance and the bulldozing effect contributed by the laterally accumulated soil to the side of tracked vehicle. The phenomenon becomes nonnegligible and sometimes significant during small radius turn maneuvers, which are quite common in robot motion planning. In order to investigate this quite important and complicated interaction process and establish a complete track-terrain interaction model, we build an instrumented experimental platform. In this platform, the RGB-D information for deformable terrains can be measured and analyzed to obtain several important soil parameters in real-time, including the contours of the soils in contact with the supporting wheels, and the cross-section shapes of the accumulated soils. We apply image segmentation, mapping depth images, point cloud drawing and postprocessing to divide the tracked vehicle and the soil in point cloud images. We reconstruct the morphology of the soil accumulated on the side of vehicles during small radius steering maneuvers, and then obtain aforementioned parameters in order to explain the bulldozing effect.