Adaptive Motion Planning via Contact-Based Intent Inference for Human-Robot Collaboration

TL;DR

This work tackles safe and efficient human-robot collaboration under partial perception by proposing a contact-informed adaptive motion-planning framework. It combines an optimization-based force estimator from joint-torque data with a torque-based contact detector, link-level localization, and a constrained optimization procedure to infer human intent from contact forces and locations, expressed via Fc and contact point s on link ell. The inferred intent then drives an online motion planner that deforms the planned trajectory using a horizon-conditioned, smooth deformation scheme, enabling rapid, continuous corrections without sustained kinesthetic guidance. validated on a 7-DOF manipulator, the approach yields accurate force estimation (mean torque error ~0.665 N·m) and enables reliable obstacle avoidance through brief human contacts, reducing operator burden while improving safety and efficiency in HRC; future work will personalize the planner’s responsiveness through learning-based parameter adaptation.

Abstract

Human-robot collaboration (HRC) requires robots to adapt their motions to human intent to ensure safe and efficient cooperation in shared spaces. Although large language models (LLMs) provide high-level reasoning for inferring human intent, their application to reliable motion planning in HRC remains challenging. Physical human-robot interaction (pHRI) is intuitive but often relies on continuous kinesthetic guidance, which imposes burdens on operators. To address these challenges, a contact-informed adaptive motion-planning framework is introduced to infer human intent directly from physical contact and employ the inferred intent for online motion correction in HRC. First, an optimization-based force estimation method is proposed to infer human-intended contact forces and locations from joint torque measurements and a robot dynamics model, thereby reducing cost and installation complexity while enabling whole-body sensitivity. Then, a torque-based contact detection mechanism with link-level localization is introduced to reduce the optimization search space and to enable real-time estimation. Subsequently, a contact-informed adaptive motion planner is developed to infer human intent from contacts and to replan robot motion online, while maintaining smoothness and adapting to human corrections. Finally, experiments on a 7-DOF manipulator are conducted to demonstrate the accuracy of the proposed force estimation method and the effectiveness of the contact-informed adaptive motion planner under perception uncertainty in HRC.

Figures (5)

• Figure 1: Human-robot collaboration in a shared space. Although the ultimate goals are aligned, the robot’s planned path does not account for the human arm as an obstacle, while the human perceives the collision risk and attempts to correct the robot’s motion through physical interaction to ensure safety.
• Figure 2: Contact is detected using torque measurements together with the robot dynamics model. Once a contact event is identified, the corresponding link is determined to reduce the search space for contact localization. Constrained optimization is then applied to the identified link to estimate the precise contact force and position. Finally, the estimated contact information is utilized to infer human intent and to adapt the robot motion accordingly.
• Figure 3: Force estimation results. (a) Experimental setup with human-applied pushes, pulls, and taps at different links during circular end-effector motion. (b) Estimated force directions and magnitudes along the circular path, with surfaces illustrating variations within each contact episode. (c) Time histories of three-dimensional force estimates. (d) Comparison of estimated and measured joint torques across seven joints.
• Figure 4: Experimental results of single-contact correction. (a) Snapshots and (b) corrected path with the original path, contact point, and estimated contact force in the first experiment; (c) and (d) present the corresponding results in the second experiment.
• Figure 5: Experimental results of multi-contact correction. (a) Main view and (b) left view of the corrected path with the original path, contact points, and estimated contact forces in the third experiment.