QERRA-HSR v0.1: A deterministic wrapper that hardens HRI safety boundaries in Webots
Watch TIAGo's state transitions from CLEAR to MONITOR to CRITICAL with keyboard triggers.
A new open-source project called QERRA-HSR v0.1 aims to harden human-robot interaction safety boundaries using a lightweight deterministic wrapper. Developed by a community contributor, the wrapper was tested in a Webots R2025a simulation with a PAL Robotics TIAGo humanoid robot. The setup includes a concrete laboratory floor with a red warning platform holding a rusty oil barrel as the hazard. A pedestrian stands nearby to trigger distress vectors.
The safety logic operates as a fast evaluation loop pinned inside a single Python controller. It defines three states: CLEAR (normal patrol speed), MONITOR (slowdown when bystander distress is detected), and CRITICAL (instant wheel velocity cut when the robot nears the hazard boundary). The states are triggered via keyboard—'D' for distress and 'H' for hazard—and log transitions in red console output. The demo shows the robot halting safely before breaching the perimeter, with no simulation lag. The project builds on earlier QERRA-v2 work and invites feedback on deterministic wrappers for HRI safety.
- QERRA-HSR v0.1 uses a deterministic, three-state safety loop (CLEAR, MONITOR, CRITICAL) running alongside the standard TIAGo controller.
- Tested in Webots R2025a with a PAL Robotics TIAGo humanoid; pressing 'D' triggers distress slowdown, 'H' cuts wheel velocity instantly at hazard boundary.
- The safety wrapper is lightweight and runs without lag, ensuring strict physical proximity enforcement without interfering with simulation steps.
Why It Matters
Deterministic safety wraps for humanoids could prevent accidents in real-world deployments by overriding risky behaviors before contact.