High-Dimensional Controller Tuning through Latent Representations

TL;DR

This paper introduces a machine learning-based method for automatic, efficient tuning of high-dimensional controller parameters, leveraging latent space representations and Bayesian optimization to improve generalization and reduce evaluation costs in complex robotic systems.

Contribution

It presents a novel approach combining latent space learning with Bayesian optimization for high-dimensional controller tuning, enabling better generalization and efficiency.

Findings

Effective tuning of high-dimensional controllers demonstrated on quadruped robots.

Significant reduction in the number of evaluations needed for tuning.

Successful transfer and generalization to new control tasks and robot dynamics.

Abstract

In this paper, we propose a method to automatically and efficiently tune high-dimensional vectors of controller parameters. The proposed method first learns a mapping from the high-dimensional controller parameter space to a lower dimensional space using a machine learning-based algorithm. This mapping is then utilized in an actor-critic framework using Bayesian optimization (BO). The proposed approach is applicable to complex systems (such as quadruped robots). In addition, the proposed approach also enables efficient generalization to different control tasks while also reducing the number of evaluations required while tuning the controller parameters. We evaluate our method on a legged locomotion application. We show the efficacy of the algorithm in tuning the high-dimensional controller parameters and also reducing the number of evaluations required for the tuning. Moreover, it is shown that the method is successful in generalizing to new tasks and is also transferable to other robot dynamics.