Adaptive Meta-Learning for Identification of Rover-Terrain Dynamics
S. Banerjee, J. Harrison, P. M. Furlong, M. Pavone
TL;DR
The paper addresses the challenge of safely navigating unknown terrain by online adaptation of rover-terrain dynamics. It introduces P-ALPaCA, a Parsimonious ALPaCA framework that augments a nominal linear-in-parameters rover model with meta-learned neural features, enabling joint inference of nominal and learned parameters. A regularization term enforcing orthogonality between nominal and learned features improves interpretability and parameter estimation while maintaining predictive accuracy. Through simulations on rocker-bogie rovers across compact and loose sand, the approach demonstrates accurate, probabilistic predictions and interpretable terrain parameter estimates, indicating potential for safer autonomous rover operations in uncertain environments.
Abstract
Rovers require knowledge of terrain to plan trajectories that maximize safety and efficiency. Terrain type classification relies on input from human operators or machine learning-based image classification algorithms. However, high level terrain classification is typically not sufficient to prevent incidents such as rovers becoming unexpectedly stuck in a sand trap; in these situations, online rover-terrain interaction data can be leveraged to accurately predict future dynamics and prevent further damage to the rover. This paper presents a meta-learning-based approach to adapt probabilistic predictions of rover dynamics by augmenting a nominal model affine in parameters with a Bayesian regression algorithm (P-ALPaCA). A regularization scheme is introduced to encourage orthogonality of nominal and learned features, leading to interpretable probabilistic estimates of terrain parameters in varying terrain conditions.