Demonstrating a Control Framework for Physical Human-Robot Interaction Toward Industrial Applications

TL;DR

The paper addresses the challenge of delivering safe, reliable physical human-robot interaction (pHRI) with industrial-grade performance for Industry 5.0. It introduces an open-source torque-based control framework built on a second-order quadratic programming (QP) formulation that enforces strict kinematic and self-collision safety while supporting null-space, full-body, and dual compliance modes on a Kinova Gen3, integrated through mc_rtc. A novel low-level torque controller and a dual-compliance strategy (without admittance control) enable precise torque tracking without sacrificing compliance. Experimental results demonstrate competitive tracking accuracy compared with position control and improved safety and interaction performance, supporting industrial deployment potential. The framework emphasizes reproducibility and industrial applicability, offering real-time mode switching, robust safety constraints, and open-source tooling for broader adoption.

Abstract

Physical Human-Robot Interaction (pHRI) is critical for implementing Industry 5.0, which focuses on human-centric approaches. However, few studies explore the practical alignment of pHRI to industrial-grade performance. This paper introduces a versatile control framework designed to bridge this gap by incorporating the torque-based control modes: compliance control, null-space compliance, and dual compliance, all in static and dynamic scenarios. Thanks to our second-order Quadratic Programming (QP) formulation, strict kinematic and collision constraints are integrated into the system as safety features, and a weighted hierarchy guarantees singularity-robust task tracking performance. The framework is implemented on a Kinova Gen3 collaborative robot (cobot) equipped with a Bota force/torque sensor. A DualShock 4 game controller is attached to the robot's end-effector to demonstrate the framework's capabilities. This setup enables seamless dynamic switching between the modes, and real-time adjustments of parameters, such as transitioning between position and torque control or selecting a more robust custom-developed low-level torque controller over the default one. Built on the open-source robotic control software mc_rtc, our framework ensures reproducibility for both research and industrial deployment, this framework demonstrates a step toward industrial-grade performance and repeatability, showcasing its potential as a robust pHRI control system for industrial environments.

Figures (7)

• Figure 1: The different compliant control modes for physical human-robot interaction (pHRI) demonstrated in this paper. Top left: Null-space compliance, where the main task (e.g., the pose of the end-effector) is stiff, and compliance is restricted to the null-space of the task. Top right: Full-body compliance, where both the main task and the null-space comply. Bottom: Dual compliance, where the main task becomes compliant only when interacting with the relevant part of the body involved in the kinematic task, and remaining stationary otherwise, while the robot remains compliant in the null-space.
• Figure 2: Overview of the proposed control framework. The light-gray block represents the tasks and constraints in the QP. The light-purple block denotes the low-level torque control, ensuring desired joint acceleration and torque tracking by controlling motor currents. The light-green block corresponds to the robot hardware, including actuators and sensors. The yellow block represents the computation of external forces, which are fed into the QP.
• Figure 3: Friction identification data for Kinova Gen3. In shaded grey the actual data, in green the Coulomb friction part, in yellow the viscous friction part and we added a potential estimation of the static friction.
• Figure 4: Comparison of the norm of the joint torque tracking error for a static scenario (top) and a dynamic scenario (bottom). Both subplots compare our torque controller (blue) and Kinova’s high-velocity torque controller (orange). The y-axis is presented in a logarithmic scale to highlight differences in the error magnitude.
• Figure 5: Position error over drop segments for torque and position control strategies. The top subplot represents the position error during torque control, while the bottom subplot represents the position error during position control. Each line corresponds to an individual drop segment. The position error is measured in millimeters (mm) and is plotted as a function of time, with each segment synchronized from the F/T sensor's force measurement. The data illustrate the differences in error behavior between the two control strategies, highlighting the capability of torque control to perform at least as well as position control.