Abstract

This paper derives a set of bulldozer dynamics to test a three-dimensional (3D) blade controller. This work was motivated by a desire to mimic how a human operator would control a bulldozer blade to build an autonomous system control module and by the lack of existing work on 3D bulldozer blade control. To that end, three-dimensional surface-to-track and surface-to-blade dynamics were derived to test the proposed 3D bulldozer blade controller in simulation. The track dynamics included considerations for track slip, while the blade dynamics accounted for non-symmetric mounds in front of the blade. The 3D bulldozer blade control effort was implemented using fuzzy logic to mimic a human operator's nonlinear modes of blade control, thereby following a Human Decision Making Model (HDMM). The inputs to the main portion of the control effort were the yaw, pitch, and roll errors between the blade's current orientation and the desired orientation. Accordingly, the outputs were the blade's yaw, pitch, and roll velocities. The pitch error, in particular, was tied to a separate fuzzy controller that outputted the desired cutting depth depending on the observed soil type, against tying back to the HDMM.