Wheel Slip Detection Using Auto-Correlation

EasyChair Preprint 10056

Abstract

As the demands on exploration rovers increase, longer driving distances, higher efficiency, and increased maneuverability are needed. One aspect that will aid in these improvements is refining the speed and accuracy of detecting wheel-slip. This paper presents experimental results of statistical moments, wavelet detail coefficients, and a time-series autocorrelation analysis of signals obtained from a rover with slipping wheels. This paper considers wheel torques, forces, and motor currents. The experimental results demonstrate the signal stationarity of wheel forces and moments and thereby enables further time-domain signals analyses. After first showing that the signals are weakly stationary, the auto-correlation function was applied using a sliding window approach, creating a timeseries of coefficients. The preliminary results using autocorrelation coefficients to identify wheel slippage was not conclusive, however it demonstrates feasibility of the method. This technique does not require prior knowledge of the environment or terrain topology, which is an inherent assumption in other works

Keyphrases: Signals Analysis, slip detection, wheeled vehicle

@booklet{EasyChair:10056,
  author    = {Morgan May and Philip Ferguson},
  title     = {Wheel Slip Detection Using Auto-Correlation},
  howpublished = {EasyChair Preprint 10056},
  year      = {EasyChair, 2023}}