Real-Time 3-D Force Measurements Using Vision-Based Flexible Force Sensor

This paper presents the design and application of a vision-based flexible three-dimensional (3D) force sensor, specifically designed for force measurement in the field of instrument measurement. The sensor utilizes visual information, employing highly elastic silicone as the contact surface and continuous marker patterns as the visual data source. By integrating a vision system, the sensor captures image data and relies on a novel deep learning model to achieve precise measurements of three-dimensional forces. Experimental evaluation was conducted to assess the sensor’s performance, demonstrating its exceptional fitting performance with a minimum mean squared error of 0.15N. Moreover, the flexible sensor exhibits excellent real-time decoupling capabilities for 3D forces at a frequency rate of 30 Hz for three orthogonal force components. we embedded the sensor into a mechanical gripper, and successfully achieved self-adaptive force grasping, verifying its effectiveness in accurately measuring 3D forces in the instrument measurement environment. The proposed vision-based flexible 3D force sensor presents an innovative solution for force sensing applications in robotics and related fields. Its compact design, high fitting performance, and real-time decoupling capabilities make it a valuable tool for enhancing the accuracy and versatility of force sensing systems in various practical instrument measurement scenarios.