Text2Interaction: Establishing Safe and Preferable Human-Robot Interaction

This is code used in "Text2Interaction: Establishing Safe and Preferable Human-Robot Interaction," presented at CoRL 2024.

For a brief overview of our work, please refer to our project page.

Further details can be found in our paper available on arXiv.

Without Text2Interaction

With Text2Interaction

Abstract

Adjusting robot behavior to human preferences can require intensive human feedback, preventing quick adaptation to new users and changing circumstances. Moreover, current approaches typically treat user preferences as a reward, which requires a manual balance between task success and user satisfaction. To integrate new user preferences in a zero-shot manner, our proposed Text2Interaction framework invokes large language models to generate a task plan, motion preferences as Python code, and parameters of a safe controller. By maximizing the combined probability of task completion and user satisfaction instead of a weighted sum of rewards, we can reliably find plans that fulfill both requirements. We find that 83% of users working with Text2Interaction agree that it integrates their preferences into the robot’s plan, and 94% prefer Text2Interaction over the baseline. Our ablation study shows that Text2Interaction aligns better with unseen preferences than other baselines while maintaining a high success rate.

Our work is split in two major parts:

1. Training skills, planning, and evaluation: We first use offline reinforcement learning to learn the robotic skills. We then plan an optimal robot plan that satisfies the user preferences and has a high success rate.
2. Generate preference functions: This code is used to generate the custom preference functions from user instructions using a large language model.

We describe the purpose, installation, steps to reproduce our results, and implementation details in greater detail in the README files of the two sub-repos. For your convenience, we added both code bases as git submodules to this repository.

Citation

Text2Interaction is offered under the MIT License agreement. If you find Text2Interaction useful, please consider citing our work:

@inproceedings{thumm_2024_Text2InteractionEstablishing,
  title = {Text2Interaction: Establishing Safe and Preferable Human-Robot Interaction},
  shorttitle = {Text2Interaction},
  booktitle = {8th Annual Conference on Robot Learning},
  author = {Thumm, Jakob and Agia, Christopher and Pavone, Marco and Althoff, Matthias},
  year = {2024},
  url = {https://openreview.net/forum?id=s0VNSnPeoA&referrer=%5BAuthor%20Console%5D(%2Fgroup%3Fid%3Drobot-learning.org%2FCoRL%2F2024%2FConference%2FAuthors%23your-submissions)},
  langid = {english},
}