SafePR: Unified Approach for Safe Parallel Robots by Contact Detection and Reaction with Redundancy Resolution

TL;DR

SafePR presents a unified, real-time framework for safe interaction with parallel robots by detecting, classifying, isolating, and reacting to human-contact events using built-in sensors and a generalized-momentum observer. It combines neural-network based contact-body and clamping classification with particle filtering for second-link localization, all integrated with online redundancy resolution across velocity, acceleration, and torque control modes, and with inequality constraints to avoid type-II singularities and self-collisions. The approach is validated on a planar 3-RRR PR with 72 real-world collision/clamping experiments, achieving contact detection within 10–100 ms, reactions within 25–275 ms, and contact forces below ISO/TS 15066 thresholds, while preserving feasible kinematic and dynamic behavior. SafePR’s open-source software and methodology demonstrate that high-speed PRs can safely operate in human environments without added hardware, marking a significant step toward practical HRC with PRs and paving the way for six-DoF extensions and model-predictive enhancements.

Abstract

Fast and safe motion is crucial for the successful deployment of physically interactive robots. Parallel robots (PRs) offer the potential for higher speeds while maintaining the same energy limits due to their low moving masses. However, they require methods for contact detection and reaction while avoiding singularities and self-collisions. We address this issue and present SafePR - a unified approach for the detection and localization, including the distinction between collision and clamping to perform a reaction that is safe for humans and feasible for PRs. Our approach uses information from the encoders and motor currents to estimate forces via a generalized-momentum observer. Neural networks and particle filters classify and localize the contacts. We introduce reactions with redundancy resolution to avoid self-collisions and type-II singularities. Our approach detected and terminated 72 real-world collision and clamping contacts with end-effector speeds of up to 1.5 m/s, each within 25-275 ms. The forces were below the thresholds from ISO/TS 15066. By using built-in sensors, SafePR enables safe interaction with already assembled PRs without the need for new hardware components.

Figures (29)

• Figure 1: (a) The considered parallel robot with contact scenarios: (b) chain clamping, collision at the (c) platform, (d) first and (e) second link. This work addresses the scenarios that (f) a reaction to a contact leads to (g) self-collision or (h) type-II singularity and, thus, an increase in the risk of injury to humans or damage to the robot.
• Figure 2: Contact detection and reaction with SafePR: Contact type (clamping, collision), location and forces are estimated based on built-in sensors for real-time reactions in the form of a structure opening or retraction movements while fulfilling limitations regarding self-collisions and type-II singularities.
• Figure 3: (a) The 3-RRR parallel robot from Mohammad.2023 with (b) a contact at $\boldsymbol{x}_\mathrm{c}$
• Figure 4: Kinetostatic analysis from Mohammad.2023_IsolLoc. (a) External force $\boldsymbol{f}_{\mathrm{ext,mP}}$ on the platform with the estimate $\hat{\boldsymbol{F}}_\mathrm{ext,mP}{=}(\hat{\boldsymbol{f}}_{\mathrm{ext,mP}}^\mathrm{T}, \hat{\boldsymbol{m}}_{\mathrm{ext,mP}}^\mathrm{T})^\mathrm{T}$, the minimum lever $\boldsymbol{r}_\mathrm{mP,LoA}$, the line of action $\boldsymbol{r}_\mathrm{LoA}(\lambda)$ and intersection points at $\lambda_1,\lambda_2$. The minimum distance $d_{\min,i}$ is between $\boldsymbol{r}_\mathrm{LoA}(\lambda)$ and the coupling point $\boldsymbol{r}_{\mathrm{cJ}i}$. (b) Link forces $\boldsymbol{F}_{\mathrm{ext,link1/2}}$ with their projections $\boldsymbol{F}_{\mathrm{ext,mP1/2}}$ on the platform coordinates. $\boldsymbol{F}_\mathrm{ext,mP2}$ and $\boldsymbol{r}_{\mathrm{pJ}i,\mathrm{cJ}i}$ include the angle $\alpha_i$.
• Figure 5: Retraction movement from Mohammad.2023_Reaction