Haptic Shoulder for Rendering Biomechanically Accurate Joint Limits for Human-Robot Physical Interactions

TL;DR

The paper tackles safe, scalable testing for human-robot physical interaction (pHRI) without human subjects by introducing SHULDRD, a low-cost, anatomically faithful shoulder replica that renders real-time force feedback. It combines a reach-cone based biomechanical model to enforce biomechanically accurate joint limits, a non-linear tendon-inspired haptic model, and an open-source hardware/software platform for real-time pHRI planning and learning. Through experiments, SHULDRD demonstrates superior configuration-space reach ($71.25\%$) compared to the human shoulder ($51.5\%$), realistic coupling of joint limits, and tendon-force rendering with a non-linear model that closely matches human tendon behavior (RMS error $0.0795$). The work delivers a practical, repeatable, and accessible testing platform that can accelerate safe robot autonomy and learning in pHRI, reducing reliance on human trials while enabling large-scale data collection.

Abstract

Human-robot physical interaction (pHRI) is a rapidly evolving research field with significant implications for physical therapy, search and rescue, and telemedicine. However, a major challenge lies in accurately understanding human constraints and safety in human-robot physical experiments without an IRB and physical human experiments. Concerns regarding human studies include safety concerns, repeatability, and scalability of the number and diversity of participants. This paper examines whether a physical approximation can serve as a stand-in for human subjects to enhance robot autonomy for physical assistance. This paper introduces the SHULDRD (Shoulder Haptic Universal Limb Dynamic Repositioning Device), an economical and anatomically similar device designed for real-time testing and deployment of pHRI planning tasks onto robots in the real world. SHULDRD replicates human shoulder motion, providing crucial force feedback and safety data. The device's open-source CAD and software facilitate easy construction and use, ensuring broad accessibility for researchers. By providing a flexible platform able to emulate infinite human subjects, ensure repeatable trials, and provide quantitative metrics to assess the effectiveness of the robotic intervention, SHULDRD aims to improve the safety and efficacy of human-robot physical interactions.

Figures (9)

• Figure 1: SHULDRD used in a human-robot physical interaction experiment where SHULDRD simulates a human shoulder. The image on the upper left shows the shoulder position relative to its complex spherical joint limits. The graph on the lower left shows the activation of the actuators rendering the elastic forces mimicking those produced by human tendons when the arm goes beyond its reachable space.
• Figure 2: The image to the left shows an exact match for the kinematics of the shoulder joint, where $\theta$ and $\phi$ represent the flexion/extension (forward / backward) and abduction/adduction (out to the side / into the body). The third DOF, humeral rotation, is shown but not highlighted. The image to the right shows a serial linkage system over the Franka Panda arm. In the center is depicted a sphere with a radius equivalent to the length of the upper arm and its center represents the center of the shoulder joint. Depicted in blue are the arcs that the shoulder can trace. The green dotted lines are the predicted linearized motion from a serial linkage design.
• Figure 3: A reach cone starts as 4 angles defining the maximum ROM in each direction. These 4 angles shown in the top 4 images define points on the unit sphere: $[p_1, p_2, p_3, p_4] \in \boldsymbol{P}$, where $\boldsymbol{P}$ is the set of points on the unit sphere defining the joint limits. The second row of images shows different reach cones at different humeral orientations
• Figure 4: In part (A) "Pre-Calculations the figure shows the addition of the visible, $V$, in cyan placed in the reach cone to designate the internal region. The visible point is then used to divide the internal space into wedges one of which is depicted in to its three surfaces of red, blue, and green. These surfaces are saved as surface normals $B_i$ and $S_i$ depicted in blue and red. Section (B) shows the calculation to define whether a shoulder orientation belongs in the reach cone. Section (C) shows the final analysis needed for when an orientation is found outside the cone and needs to be pushed back towards the cone in the appropriate direction.
• Figure 5: Full SHULDRD device mounted onto a mannequin. The device is oriented to align its zero positions with anatomical zero. The single joint center is highlighted in yellow. The axes depict the 3 axes of rotation: red for flexion, green for abduction, and light blue for humeral rotation. As is seen in the image all 3 axes intersect at the orange joint center.