Self-Sensing Feedback Control of an Electrohydraulic Robotic Shoulder

TL;DR

A bio-inspired antagonistic robotic shoulder with two degrees of freedom powered by self-sensing hydraulically amplified self-healing electrostatic actuators and closed-loop controllable robotic manipulators based on an inherent self-sensing capability of electrohydraulic actuators is presented.

Abstract

The human shoulder, with its glenohumeral joint, tendons, ligaments, and muscles, allows for the execution of complex tasks with precision and efficiency. However, current robotic shoulder designs lack the compliance and compactness inherent in their biological counterparts. A major limitation of these designs is their reliance on external sensors like rotary encoders, which restrict mechanical joint design and introduce bulk to the system. To address this constraint, we present a bio-inspired antagonistic robotic shoulder with two degrees of freedom powered by self-sensing hydraulically amplified self-healing electrostatic actuators. Our artificial muscle design decouples the high-voltage electrostatic actuation from the pair of low-voltage self-sensing electrodes. This approach allows for proprioceptive feedback control of trajectories in the task space while eliminating the necessity for any additional sensors. We assess the platform's efficacy by comparing it to a feedback control based on position data provided by a motion capture system. The study demonstrates closed-loop controllable robotic manipulators based on an inherent self-sensing capability of electrohydraulic actuators. The proposed architecture can serve as a basis for complex musculoskeletal joint arrangements.

Figures (8)

• Figure 1: (A) Bio-inspired robotic shoulder powered by two pairs of Peano-HASEL actuators with our self-sensing control capabilities. (B) Self-sensing HASEL design separating low voltage from high voltage electrodes for capacitive self-sensing. (C) Close-up of the ball-and-socket joint mechanism which allows for two-DoF motions: Tendons are attached to the output link of the ball joint, allowing the end effector to pivot up to $80°$ along the $x$ and $y$ axes.
• Figure 2: (A) HASEL design featuring five actuation pouches, each measuring $w =$ 50mm by $h =$ 20mm. Placed on top is the self-sensing pouch sized $w =$ 50mm by $h =$ 30mm. This sensing pouch detects the voltage difference across its low-voltage electrodes. When the pouch is inactive, the sensing electrodes are closely spaced, resulting in a high capacitance and a corresponding low voltage drop. When fully actuated at displacement $\Delta q$, the capacitance of the low-voltage electrodes is at its lowest, resulting in the highest possible voltage drop as the pouch reaches its maximum displacement. (B) Electrical self-sensing circuit: A 10V peak-to-peak 2kHz sinusoidal signal is generated. The signal is then amplified by a non-inverting operational amplifier circuit (OP07CD, Texas Instruments) with a gain of $1:2$. The signal path consists of two resistors of equal value (1MΩ) as well as the sensing electrodes of the Peano-HASEL, all connected in series. An instrumentation amplifier (INA821, Texas Instruments) monitors the voltage drop over the sensing electrodes and passes the output through a band-pass filter ($f_c =$ 2) to the data acquisition (DAQ) system.
• Figure 3: Ball joint design and inverse kinematic model.
• Figure 4: Open-loop range for the $\phi$ angle. A 5.5 3 sinusoidal signal is applied to $H_{\phi,1}$ and $H_{\phi,2}$ with a half-period phase shift.
• Figure 5: (A) Working principle of the system: By contracting and relaxing the artificial muscle pairs, the manipulator can reach the desired pitch and roll angles. (B) Proposed PID feedback control architecture for $i = \{ \phi_1, \phi_2, \theta_1, \theta_2\}$. The controller supports two configurations: 1) Self-sensing feedback control based on estimations $\mathbf{x_e}$ without external sensors. 2) Feedback control using motion capture position data $\mathbf{x}$ (providing ground truth data). This data serves as a benchmark while also training and validating the prediction model.