PhysVLM addresses the critical gap in robotic systems by enabling visual language models to understand physical reachability - a fundamental capability for practical deployment.

• Introduces Space-Physical Reachability Map (S-P Map) as a unified representation across diverse robotic platforms
• Enables robots to accurately determine what objects they can physically interact with
• Improves embodied reasoning tasks by reducing impractical or impossible instructions
• Demonstrates applications in manufacturing, security, and general robotics environments

This advancement enables more autonomous and reliable robotic systems in engineering applications, particularly in factory settings where physical manipulation is essential for productivity.

PhysVLM: Enabling Visual Language Models to Understand Robotic Physical Reachability