Live Displacement Estimation for Rough Terrain Mobile Robot: BART LAB Rescue Robot

Abstract

This paper illustrates a sensorless based live displacement estimation (LDE) method for rescue robot manuevering in an unknown environment. Live displacement estimation is crucial at the instance when a heavy load is encountered as a disturbance on its path, which requires the pushing motion to alleviate the problem. This paper focuses on a pushing task, where a rescue robot pushes an object so that it can follow a path without sliding it away. Pushing is also mechanically unstable, and thus various control problems arise. Consequently, the successful completion of a pushing manipulation task requires planning. Therefore, for successful pushing operation of an object in an unknown environment it requires the precise estimation of the live displacement. LDE estimates the position and the estimated torque using the varying inertia observer (VIO), which is a variant of Disturbance Observer (DOB). The validity of the LDE was confirmed by a teleoperated rescue robot in various terrains.