MuTT: A Multimodal Trajectory Transformer for Robot Skills

TL;DR

MuTT is a transformer-based model that predicts environment-aware robot skill executions by fusing vision and trajectory data, enabling efficient parameter optimization without real-world trials.

Contribution

We introduce MuTT, a novel multimodal transformer architecture that fuses vision and trajectory data for environment-aware robot skill prediction and optimization.

Findings

MuTT outperforms existing methods in predicting robot skill executions.

It enables environment-aware parameter optimization without real-world trials.

Demonstrates versatility across different robot skill representations.

Abstract

High-level robot skills represent an increasingly popular paradigm in robot programming. However, configuring the skills' parameters for a specific task remains a manual and time-consuming endeavor. Existing approaches for learning or optimizing these parameters often require numerous real-world executions or do not work in dynamic environments. To address these challenges, we propose MuTT, a novel encoder-decoder transformer architecture designed to predict environment-aware executions of robot skills by integrating vision, trajectory, and robot skill parameters. Notably, we pioneer the fusion of vision and trajectory, introducing a novel trajectory projection. Furthermore, we illustrate MuTT's efficacy as a predictor when combined with a model-based robot skill optimizer. This approach facilitates the optimization of robot skill parameters for the current environment, without the need for real-world executions during optimization. Designed for compatibility with any representation of robot skills, MuTT demonstrates its versatility across three comprehensive experiments, showcasing superior performance across two different skill representations.