Distributed Robot Perception for Tracked Vehicles

Accurate estimation and compensation of slippage are critical challenges for tracked robots operating in unstruc- tured terrain. This work presents a novel real-time method for estimating longitudinal and lateral slippage in a tracked vehicle using the Iterative Re-weighted Least Squares (IRLS) algorithm. By combining data from proprio-ceptive and extero- ceptive sensors, the IRLS-based estimator is designed to predict slippages using only wheel speeds, GPS and IMU readings. We validate our approach in a simulated environment, comparing it to two other methods: a baseline with no slippage compensation and a prediction technique based on a exponential function approximation. Our estimation method adapts over time through data-driven updates and enables a slippage-aware controller [1] to achieve better trajectory tracking performance with respect to a fixed parameter estimator.