Force-Safe Environment Maps and Real-Time Detection for Soft Robot Manipulators

TL;DR

This work tackles force safety in soft robot manipulation by shifting the safety criterion from purely geometric avoidance to bounds on environmental contact forces. It introduces force-safe environment maps by translating task-space force-unsafe deformation regions (FODR) into configuration-space representations via forward kinematics under piecewise constant curvature (PCC) kinematics, and further builds a real-time detection pipeline. A key contribution is the construction of a continuous force-unsafe set in configuration space using $\alpha$-shape reconstruction from discretized occupancy data, enabling fast, real-time queries during planning and control. The approach is validated in both simulation and hardware on a planar two-segment soft robot, showing accurate force safety detection with runtimes suitable for real-time operation and establishing a foundation for force-safe motion planning in delicate, deformable environments. The framework paves the way for robust, contact-aware manipulation where safety is guaranteed by controlling contact forces rather than merely avoiding contact.

Abstract

Soft robot manipulators have the potential for deployment in delicate environments to perform complex manipulation tasks. However, existing obstacle detection and avoidance methods do not consider limits on the forces that manipulators may exert upon contact with delicate obstacles. This work introduces a framework that maps force safety criteria from task space (i.e. positions along the robot's body) to configuration space (i.e. the robot's joint angles) and enables real-time force safety detection. We incorporate limits on allowable environmental contact forces for given task-space obstacles, and map them into configuration space (C-space) through the manipulator's forward kinematics. This formulation ensures that configurations classified as safe are provably below the maximum force thresholds, thereby allowing us to determine force-safe configurations of the soft robot manipulator in real-time. We validate our approach in simulation and hardware experiments on a two-segment pneumatic soft robot manipulator. Results demonstrate that the proposed method accurately detects force safety during interactions with deformable obstacles, thereby laying the foundation for real-time safe planning of soft manipulators in delicate, cluttered environments.

Figures (12)

• Figure 1: Soft robot manipulator motions in hardware and simulation are overlaid on top of each other, with colors indicating force safety: green for safe and red for unsafe configurations. (a) shows a timelapse of motion toward the first obstacle, and (b) toward the second obstacle.
• Figure 2: Overview of our force-safe environment maps and real-time detection approach. (1) illustrates the deformation of a polygonal environment along its inward normal direction to define the force-unsafe deformation region (2). (3) shows inflation of the FODR obstacles by the manipulator’s thickness, allowing us to consider only the backbone for subsequent computations. (4) demonstrates our method for mapping FODR poses to configuration space and (5) shows computation of alpha shape regions for efficient force safety queries for multiple robot configurations.
• Figure 3: We define the original obstacle as polytope $\mathcal{N}$ and the force-unsafe deformation region as $\mathcal{P}$. Robot poses illustrate (i) collision-free (no contact) configuration, (ii) contact within force-safe limits, and (iii) unsafe contact exceeding $F^{max}$.
• Figure 4: Piecewise Constant Curvature (PCC) kinematics of the soft robot manipulator Webster2010Design. (a) 3D schematic of a single segment, parameterized by arc length $\ell$, curvature $\kappa$, and thickness $r$, with radius of curvature $R = 1/\kappa$. (b) Planar view showing the joint angle and backbone configuration of a single segment.
• Figure 5: $\mathrm{FODR}$ inflation using Minkowski sums. (Left) The $\mathrm{FODR}$ (light gray) and soft robot manipulator are shown. (Right) The $\mathrm{FODR}$ boundary is inflated by the manipulator’s thickness (dark gray), resulting in a grown obstacle that accounts for the robot’s thickness

Theorems & Definitions (4)

• Definition 1
• Definition 2
• Theorem 1: Force-Safe Configuration Detection
• proof