CAPE: Corrective Actions from Precondition Errors using Large Language Models

Related Links:

• Webpage with Supplementary Materials
• Paper:
  • arXiv (best version)
  • Previous version (FMDM @ NeurIPS 2022)

NOTES

• In this paper, we introduce CAPE: an approach to address errors encountered by a robot or agent when executing a plan.
  • We use large language models (LLM) to perform planning and to infer the causes of errors.
  • We assume that these errors lie in preconditions that are not being met by the agent.
  • This work builds upon a submission made to the NeurIPS 2022 Workshop on Foundation Models for Decision Making .
• We showcase our method in simulation (AI2THOR) and with a real robot (Boston Dynamics Spot quadruped robot).
• This paper is to appear at ICRA 2024!

Citation

S. Sundara Raman, V. Cohen, I. Idrees, E. Rosen, R. Mooney, S. Tellex, and D. Paulius (2024). “CAPE: Corrective Actions from Precondition Errors using Large Language Models”. In: 2024 IEEE International Conference on Robotics and Automation (ICRA). IEEE.