SurgVLA-Bench Evaluates VLA Models in Laparoscopic Surgical Robotics
June 30, 2026
SurgVLA-Bench provides a hierarchical task taxonomy and evaluation framework for vision-language-action models in laparoscopic surgery using the SurRoL simulation platform. Benchmarking shows autoregressive models like OpenVLA lead in semantic understanding, while flow matching models like pi-0 show different performance characteristics in surgical action execution.
HOW THIS AFFECTS YOU
●
builderThis provides a simulation-based framework for testing surgical robotics software before hardware deployment.
●
researcherYou can use this standardized taxonomy to compare VLA architectures in specialized robotic domains.