This research explores the use of large language models (LLMs) for generating programs for service mobile robots. The authors introduce two contributions: CODEBOTLER, an open-source robot-agnostic tool for programming service mobile robots from natural language, and ROBOEVAL, a benchmark for evaluating the capacity of LLMs to generate programs for completing service robot tasks. The study also examines the modes of failure in LLM-generated robot programs, creating a taxonomy that highlights common pitfalls. The code and benchmark are publicly available.
Publication date: 21 Nov 2023
Project Page: https://amrl.cs.utexas.edu/codebotler/
Paper: https://arxiv.org/pdf/2311.11183
