Skip to main content
eScholarship
Open Access Publications from the University of California

Design, simulation, and control of a vertically balancing treaded rover

  • Author(s): Grinberg, Matt R.
  • et al.
Abstract

An original design based on mechanical proficiency and algorithmic control of a continuous track rover enables it to perform self-balancing operations and large obstacle negotiations. While the tank-like treads are capable of overcoming basic terrain roughness, it is the rover's ability to manipulate its center of mass that allows for unorthodox vertical balance. An extended rotating boom, housing heavy on-board batteries, can be pivoted in order to effectively change the location of the center of mass. A sequence of control laws takes advantage of this movement using sensor information from potentiometers, accelerometers, and gyroscopes to properly upright and stabilizes the vehicle. In order to accurately estimate the rover's orientation, Kalman filtering techniques are applied to sensor measurements eliminating vibration and noise errors. A feedback gain is derived using a linear- quadratic regulator from a state-space linear model of the physical system and utilized to direct power to the motors. This enables both deliberate maneuvers and immediate response to impacts or other outside disturbances, making it possible for the rover to be agile and dexterous over various topographies and over-sized impediments. For processing, the NI single-board RIO allows for estimation and control computations to be implemented onboard in real time. Once programmed, the robot can function as a standalone unit capable of adjusting to its environment.

Main Content
Current View