Self-Sensing for Proprioception and Contact Detection in Soft Robots Using Shape Memory Alloy Artificial Muscles

TL;DR

Soft robots lack intrinsic proprioception and safe contact sensing without sacrificing compliance. This work shows SMA-based self-sensing using electrical resistance $R$ and temperature $T$ to infer pose under no-contact, and extends to detecting external contact by comparing predicted and true pose; a simple polynomial regression suffices. The authors demonstrate no-contact SMA force predictions with average error $0.118\,\

Abstract

Estimating a soft robot's pose and applied forces, also called proprioception, is crucial for safe interaction of the robot with its environment. However, most solutions for soft robot proprioception use dedicated sensors, particularly for external forces, which introduce design trade-offs, rigidity, and risk of failure. This work presents an approach for pose estimation and contact detection for soft robots actuated by shape memory alloy (SMA) artificial muscles, using no dedicated force sensors. Our framework uses the unique material properties of SMAs to self-sense their internal stress, via offboard measurements of their electrical resistance and in-situ temperature readings, in an existing fully-soft limb design. We demonstrate that a simple polynomial regression model on these measurements is sufficient to predict the robot's pose, under no-contact conditions. Then, we show that if an additional measurement of the true pose is available (e.g. from an already-in-place bending sensor), it is possible to predict a binary contact/no-contact using multiple combinations of self-sensing signals. Our hardware tests verify our hypothesis via a contact detection test with a human operator. This proof-of-concept validates that self-sensing signals in soft SMA-actuated soft robots can be used for proprioception and contact detection, and suggests a direction for integrating proprioception into soft robots without design compromises. Future work could employ machine learning for enhanced accuracy.

Figures (7)

• Figure 1: Our approach, using self-sensing of stress in a soft robot's artificial muscle actuators, can detect when external contact has occurred. No additional sensors are added to the soft limb's design.
• Figure 2: Architecture overview. Our existing limb design uses (a) computer vision for ground truth measurements of the robot's pose, where the limb includes (b) a silicone body actuated by shape memory alloy wire coils with a thermocouple affixed at the base. (c) Our data-driven approach includes our prior work in temperature-safe feedback control of these thermal SMA wires, with a data collection and sensing setup. (d) Our free-motion contact model uses a simple beam-bending approximation to map pose to actuator stress.
• Figure 3: Both linear and quadratic regression models for no-contact actuator force self-sensing demonstrate that resistance deconflicts temperature hysteresis, given a split between hot (orange) and cold (blue) operating ranges per the SMA's critical transition temperature.
• Figure 4: A snapshot of no-contact actuator force predictions shows a realistic trend using only self-sensing of resitance and in-situ temperature readings.
• Figure 5: External force estimation with self-sensing signals shows that including electrical resistance $R$ in addition to temperature and pose ($T,\theta$) has minimal improvement. However, we notice that the $R,\theta$ model predicts nonzero forces at similar times as those with temperature, prompting the question: can $R$ substitute $T$ for contact detection?