English

MuTT: A Multimodal Trajectory Transformer for Robot Skills

Robotics 2024-08-23 v2 Machine Learning

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.

Keywords

Cite

@article{arxiv.2407.15660,
  title  = {MuTT: A Multimodal Trajectory Transformer for Robot Skills},
  author = {Claudius Kienle and Benjamin Alt and Onur Celik and Philipp Becker and Darko Katic and Rainer Jäkel and Gerhard Neumann},
  journal= {arXiv preprint arXiv:2407.15660},
  year   = {2024}
}
R2 v1 2026-06-28T17:49:33.214Z