Give me scissors: Collision-Free Dual-Arm Surgical Assistive Robot for Instrument Delivery

A research team led by Xuejin Luo and Shiquan Sun has published a paper titled 'Give me scissors: Collision-Free Dual-Arm Surgical Assistive Robot for Instrument Delivery,' accepted for ICRA 2026. The work addresses a critical bottleneck in surgery: the repetitive, fatiguing task of scrub nurses handing instruments to surgeons. Unlike previous robotic systems limited to pre-defined, rigid paths, this new robot leverages a vision-language model (VLM) to interpret a surgeon's verbal command (e.g., 'Give me scissors') and autonomously generate the entire trajectory for a dual-arm robot to grasp and deliver the requested tool. This zero-shot capability is a significant leap in generalizability for assistive surgical robotics.

The system's core innovation is a unified quadratic programming framework that integrates a real-time 'obstacle minimum distance perception' method. This allows the dual-arm robot to dynamically avoid both environmental obstacles and self-collision while moving in the unpredictable, crowded space of an operating room. Extensive experimental validation demonstrated an 83.33% success rate in tool delivery while maintaining smooth, collision-free motion in all trials. The availability of the project page and source code suggests a push for reproducibility and further development. This research points toward a future where AI-driven robotic assistants can reliably take over logistical tasks in surgery, potentially reducing human error and allowing medical staff to focus on higher-level decision-making.