Surrogate-based excavation trajectory generation of unmanned excavator under uncertainty

Abstract

Generating a reliable excavation trajectory is the essential precondition for unmanned excavators to achieve safe and energy-efficient operation. Conventional trajectory planning methods often oversimplify the geometric path by using predefined curve types for parameterization and typically neglect environmental uncertainties. These simplifications can constrain the optimization space and reduce the practical executability of the generated trajectories, ultimately limiting excavation performance and adaptability in complex, variable working conditions. To address this challenge, this study presents a surrogate-based excavation trajectory generation method that considers soil parameters uncertainty. First, the distributions of unknown parameters in soil mechanics equation are determined by minimizing discrepancies between measured and estimated excavation forces. Next, a surrogate‐based excavation trajectory generation method is proposed to enhance performance by increasing path flexibility, and reliability constraints are incorporated to mitigate the effects of soil excavability on autonomous operation. Finally, the methodology’s effectiveness is systematically validated through numerical experiments.