Symbolic Runtime Verification and Adaptive Decision-Making for Robot-Assisted Dressing

2025-04-22

AI Summary

This work addresses the challenges of real-time safety monitoring and adaptive decision-making in robot-assisted dressing. We propose a safety-controllable framework integrating runtime monitoring, formal verification, and Bayesian adaptive decision-making. Methodologically, we innovatively combine parametric discrete-time Markov chains (pDTMCs) with symbolic probabilistic model checking to enable dynamic Bayesian updating of safety constraints and interpretable, real-time Probabilistic Computation Tree Logic (PCTL) verification. The approach incorporates multi-objective optimization—balancing reachability, cost, and reward—and achieves millisecond-level detection and graded intervention for clothing entanglement risks in realistic human-robot interaction scenarios. Experimental results demonstrate 99.3% safety assurance, significantly enhanced decision interpretability, and—critically—the first integration of formal safety guarantees with online adaptivity in robot-assisted dressing.

Technology Category

Intelligent Robots: State EstimationReasoning under Uncertainty: Sequential Decision MakingNatural Language Processing: Safety and Robustness

Application Category

Responsible Web: Machine-in-the-loop, human agency and autonomyEconomics, Online Markets and Human Computation: LLM based quality controls for crowd workUser Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

We present a control framework for robot-assisted dressing that augments low-level hazard response with runtime monitoring and formal verification. A parametric discrete-time Markov chain (pDTMC) models the dressing process, while Bayesian inference dynamically updates this pDTMC's transition probabilities based on sensory and user feedback. Safety constraints from hazard analysis are expressed in probabilistic computation tree logic, and symbolically verified using a probabilistic model checker. We evaluate reachability, cost, and reward trade-offs for garment-snag mitigation and escalation, enabling real-time adaptation. Our approach provides a formal yet lightweight foundation for safety-aware, explainable robotic assistance.

Problem

• Develops a control framework for robot-assisted dressing with runtime monitoring
• Uses Bayesian inference to update Markov chain probabilities dynamically
• Ensures safety via symbolic verification of probabilistic logic constraints

Innovation

• Parametric Markov chain models dressing process
• Bayesian inference updates transition probabilities dynamically
• Symbolic verification ensures safety constraints compliance