Learning Skill-based Industrial Robot Tasks with User Priors

TL;DR

This paper presents a method combining user priors with Bayesian optimization to efficiently learn dexterous, contact-rich robot tasks, significantly reducing setup time and improving performance by leveraging prior knowledge.

Contribution

It introduces a novel approach that integrates user and transferred priors into Bayesian optimization for rapid, effective learning of industrial robot tasks, including multi-objective scenarios.

Findings

Operator priors accelerate Pareto front discovery.

Transferred priors improve real robot task learning.

Method outperforms baseline approaches in efficiency and quality.

Abstract

Robot skills systems are meant to reduce robot setup time for new manufacturing tasks. Yet, for dexterous, contact-rich tasks, it is often difficult to find the right skill parameters. One strategy is to learn these parameters by allowing the robot system to learn directly on the task. For a learning problem, a robot operator can typically specify the type and range of values of the parameters. Nevertheless, given their prior experience, robot operators should be able to help the learning process further by providing educated guesses about where in the parameter space potential optimal solutions could be found. Interestingly, such prior knowledge is not exploited in current robot learning frameworks. We introduce an approach that combines user priors and Bayesian optimization to allow fast optimization of robot industrial tasks at robot deployment time. We evaluate our method on three tasks that are learned in simulation as well as on two tasks that are learned directly on a real robot system. Additionally, we transfer knowledge from the corresponding simulation tasks by automatically constructing priors from well-performing configurations for learning on the real system. To handle potentially contradicting task objectives, the tasks are modeled as multi-objective problems. Our results show that operator priors, both user-specified and transferred, vastly accelerate the discovery of rich Pareto fronts, and typically produce final performance far superior to proposed baselines.