Compact LED-Based Displacement Sensing for Robot Fingers

TL;DR

This work presents a compact LED-based displacement sensor designed for robot fingers to infer net contact forces and torques via relative plate motion in a 6-DOF elastomeric flexure. By employing LEDs as both emitters and receivers in an LED-LED network, the sensor achieves high signal-to-noise and small displacement sensitivity in a finger-sized package, with on-board electronics and no amplification required. A ResNet-inspired time-series architecture maps 24 LED-receiver signals to six force/torque components, achieving mean errors around $0.05$–$0.07$ N for x/y forces and $2.6$–$2.9$ N·mm for torques, with $R^2$ values near 0.99 for the primary axes; z-axis performance is weaker due to smaller true forces and hysteresis. The sensor demonstrates robust, low-cost, and easily integrated sensing suitable for manipulation tasks, offering a practical path toward ubiquitous finger-level sensing and data-driven manipulation policies. Future work targets hysteresis reduction, mechanical improvements, and deeper integration into anthropomorphic hands.

Abstract

In this paper, we introduce a sensor designed for integration in robot fingers, where it can provide information on the displacements induced by external contact. Our sensor uses LEDs to sense the displacement between two plates connected by a transparent elastomer; when a force is applied to the finger, the elastomer displaces and the LED signals change. We show that using LEDs as both light emitters an receivers in this context provides high sensitivity, allowing such an emitter and receiver pairs to detect very small displacements. We characterize the standalone performance of the sensor by testing the ability of a supervised learning model to predict complete force and torque data from its raw signals, and obtain a mean error between 0.05 and 0.07 N across the three directions of force applied to the finger. Our method allows for finger-size packaging with no amplification electronics, low cost manufacturing, easy integration into a complete hand, and high overload shear forces and bending torques, suggesting future applicability to complete manipulation tasks.

Figures (9)

• Figure 1: Our LED-based sensor, standalone (left) and mounted at the base of a robot finger (right). Our sensor uses LEDs mounted on two plates connected by a transparent elastomer to sense displacement between the plates that corresponds to the applied forces and torques. This method allows for a compact, fully integrated package at a low manufacturing cost.
• Figure 2: A diagram of our LED-to-LED sensor concept. We align two PCBs with LEDs directly above each other. We attach these circuit boards to rigid plates, we then suspend both of these plates in a transparent, elastic material. Assuming the bottom plate is fixed, a force applied to the top plate will displace the plate and change how to emitter LED shines on the receiving LEDs. We can then map this change to the contact wrench that induced the displacement.
• Figure 3: To analyze single emitter-receiver pairs, we used a precise linear motor to displace an emitting LED while recording the response from both a receiving LED and a photodiode. Left: sharper cone response of the LED compared to the photodiode. Right: zooming in on response across a .1mm displacement, we see significantly higher SNR when using the LED as receiver versus the photodiode.
• Figure 4: Our complete sensor is composed of two custom PCBs that both contain LED emitters and clusters of LED receivers. The PCBs are fastened to rigid, 3-D printed plates. The plates are then fixed above each other in a mold that we then fill with polydimethylsiloxane (PDMS). We use a single JST connector to communicate between the top and bottom board, and an additional JST connector on the bottom board to communicate with external electronics. The rigid plates are fitted with threaded holes to mount external hardware. Our sensor is 27 mm in diameter, 20mm in height, and weighs 16.4g.
• Figure 5: Mechanical hysteresis testing over three consecutive loading and unloading cycles. We fixed our sensor on a reference F/T sensor and used a linear probe to displace the top plate by 1 mm.