Adaptive Physical Human�Robot Interaction Via a Passivity-Aware Model Predictive Variable Admittance Control
Dalia M. Mahfouz, Paolo Di Lillo, Omar M. Shehata, Elsayed Morgan, Filippo Arrichiello
AI summary
Problem
Existing variable admittance controllers for physical human-robot interaction often struggle to balance tracking accuracy, compliance, and safety under unpredictable human behaviors, frequently neglecting force directionality and lacking formal stability guarantees.
Approach
The authors develop a Model Predictive Variable Admittance (MPVA) controller that continuously optimizes stiffness and damping parameters using real-time interaction metrics, explicitly embedding passivity constraints to prevent unsafe energy injection.
Key results
- Real-time MPC adaptation of admittance parameters using force directionality and tracking error
- Experimental validation on a 7-DoF Kinova Jaco-2 robot across assistive and resistive modes
- Competitive tracking accuracy and reduced physical effort compared to fixed-gain and fuzzy baselines
- Minimal passivity violations ensuring safe energy exchange during physical contact
Why it matters
Enables safer, more intuitive collaborative robotics by providing a mathematically guaranteed, intent-aware compliance strategy for dynamic human-robot tasks.
Abstract
Physical Human–Robot Interaction (pHRI) re- quires control frameworks that balance accuracy, compliance, and safety under variable human behaviors. This paper pro- poses a novel Model Predictive Variable Admittance (MPVA) framework that integrates trajectory tracking, interaction force directionality, and passivity constraints into an online real-time optimization scheme. The proposed architecture is implemented on a 7-DoF Kinova Jaco-2 robot and validated experimentally through mixed assistive and resistive modes with multiple subjects performing pHRI tasks. Results supported by both objective metrics and subjective evaluation through a NASA TLX survey show that the MPVA achieves competitive track- ing accuracy, reducing physical effort with minimal passivity violations compared to other algorithmic baselines such as fixed-gain admittance and fuzzy-based adaptive admittance. This demonstrates safe and effective human-robot physical interaction across diverse modes.