Robots estimate and adapt to changing surface friction by combining sensing, modeling, and control so that contact forces remain safe and effective. Misestimating friction leads to slips, falls, damaged payloads, and hazards in human-shared environments. The Springer Handbook of Robotics edited by Oussama Khatib Stanford University and Bruno Siciliano University of Naples Federico II outlines the foundational role of force control and impedance strategies for handling contact variability.
Sensing and estimation
Accurate estimation begins with tactile sensing and force measurements. Researchers such as Antonio Bicchi University of Pisa have published extensively on tactile sensors and slip detection to infer local surface properties and contact stability. Proprioceptive data from joint encoders and force/torque sensors support online model identification that estimates the friction coefficient by comparing predicted contact forces from dynamics models with measured values. Vision contributes by providing texture and material cues that can be mapped to likely friction ranges using learning methods; such visual priors reduce uncertainty before contact but remain approximate and must be validated after touch.
Control and adaptation
Once friction is estimated, control schemes adjust behavior. Impedance control and force-control frameworks described by Oussama Khatib Stanford University and Bruno Siciliano University of Naples Federico II modulate stiffness and commanded forces to tolerate lower friction without losing task performance. For locomotion, teams led by Marco Hutter ETH Zurich demonstrated that legged robots can adapt gait parameters and contact timing in real time based on traction estimates to avoid falls on loose, wet, or icy ground. Machine learning policies and adaptive controllers can blend model-based estimates with experience, using online updates to cope with noisy sensors and unmodeled effects.
Relevance extends beyond engineering: environmental conditions such as snow, mud, or oil films change friction abruptly and carry cultural and territorial implications for deployment in urban, agricultural, or industrial settings. In human-robot collaboration contexts, conservative force limits and rapid slip detection protect people and property. Consequences of poor adaptation include operational failure and safety incidents, while robust estimation and adaptive control increase reliability and acceptance of robots in diverse workspaces. Together, sensing, estimation, and control form a layered approach that enables robots to sense slipperiness, infer its causes, and adjust behavior to maintain performance and safety.