Benchmark RadarAI BENCHMARK PROFILE
TouchSafeBench
TouchSafeBench is a physics-grounded benchmark with 2,940 simulated episodes for evaluating collision grounding in vision-language models for safe human-robot collaboration.
- Released
- 2026-05-29
- Readiness
- Paper only
- Primary field
- Robotics & Autonomous Systems
Why it matters
Targets a critical safety capability in embodied AI, but current evidence lacks public availability and scoring details.
Motivation
Safe human--robot collaboration requires more than visual description: a monitor must determine whether the robot body is safely separated, already colliding with the scene or a person, or about to collide.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.