IMR-LLM: Industrial Multi-Robot Task Planning and Program Generation using Large Language Models
Xiangyu Su, Juzhan Xu, Oliver van Kaick, Kai Xu, Ruizhen Hu
AI summary
Problem
Industrial multi-robot tasks involve rigid execution sequences and complex resource dependencies that exceed the reasoning limits of direct LLM prompting, while existing program generation methods often produce non-executable code due to environment-specific variations.
Approach
The framework uses LLMs to construct disjunctive graphs for high-level task planning, which are solved via deterministic heuristics, and leverages a unified operation process tree to guide the generation of executable, environment-adaptive low-level robot programs.
Key results
- Introduces IMR-LLM framework integrating LLMs with disjunctive graph scheduling and process tree-guided code generation
- Creates IMR-Bench, a multi-level complexity benchmark for industrial multi-robot task evaluation
- Achieves significant performance gains over existing methods across all evaluation metrics in simulation and real-world tests
- Enhances program executability and planning feasibility by decoupling constraint handling from LLM reasoning
Why it matters
Enables reliable deployment of LLMs in complex industrial automation by bridging high-level reasoning with executable, constraint-aware robot control.
Abstract
In modern industrial production, multiple robots often collaborate to complete complex manufacturing tasks. Large language models (LLMs), with their strong reasoning capabilities, have shown potential in coordinating robots for simple household and manipulation tasks. However, in in- dustrial scenarios, stricter sequential constraints and more complex dependencies within tasks present new challenges for LLMs. To address this, we propose IMR-LLM, a novel LLM-driven Industrial Multi-Robot task planning and program generation framework. Specifically, we utilize LLMs to assist in constructing disjunctive graphs and employ deterministic solving methods to obtain a feasible and efficient high-level task plan. Based on this, we use a process tree to guide LLMs to generate executable low-level programs. Additionally, we create IMR-Bench, a challenging benchmark that encompasses multi-robot industrial tasks across three levels of complexity. Experimental results indicate that our method significantly surpasses existing methods across all evaluation metrics.