ZeroCAP enables multi-robot systems to form complex patterns based on natural language instructions without prior training, combining LLMs with robotic coordination.

• Achieves zero-shot pattern formation where robots understand and execute spatial arrangements from text
• Incorporates context awareness allowing robots to adapt formations to environmental constraints
• Demonstrates practical applications in object manipulation and collaborative robotics
• Represents a significant advancement in human-robot interaction through natural language commands

This research bridges the gap between language understanding and physical robot coordination, enabling more intuitive control of robot teams for complex engineering tasks without specialized programming.

ZeroCAP: Zero-Shot Multi-Robot Context Aware Pattern Formation via Large Language Models