Safe Robotic Task Execution in Self-Driving Laboratories — Guest Lecture
Date:
Delivered a guest lecture to MSc students at the University of Liverpool on “Safe Robotic Task Execution in Self-Driving Laboratories”. The lecture covered safety considerations for mobile and manipulation robots in automated labs, perception-driven hazard detection, risk-aware planning, and strategies for fault detection, recovery, and human-robot coordination in shared lab spaces.
Key topics discussed (drawn from our recent work):
- PREVENT (Multimodal BTs): navigation and manipulation skills using a multimodal Behavior Tree approach for proactive risk evaluation and vigilant execution — hierarchical perception combining CNNs, vision-language modules, and sensors to detect hazards and trigger corrective actions (see PREVENT).
- LIRA (Localization, Inspection, Reasoning): an edge VLM-powered module for precise localization, automated inspection, and reasoning that enables fast, robust error detection and recovery in self-driving labs (see LIRA).