Computer Vision Technique to Analyze Deformation in a Compliant-Based Visual Force Feedback Device

A vision-based method was developed to automate the measurement of deformation in compliant mechanisms. Traditional force feedback systems involve excessive complexity, cost, or incompatibility with soft and miniaturized robotic applications. Literature supports vision-based techniques for non-intrusive force estimation. The proposed method introduces Distance-to-critical-point (Dcp) as a geometric metric for deformation tracking. Dcp was hypothesized to exhibit linear correlation with pinch force and to allow accurate extraction using computer vision methods. Experiments included varied jaw openings and object stiffness levels. Accuracy of Dcp measurements obtained by computer vision was compared to manual measurements using mean absolute error, and paired t-tests. The findings support that the computer vision method shows no statistical difference from manual measurement. Results revealed strong linear correlations between Dcp and pinch force in all test scenarios. Thus, computer vision method established Dcp as a consistent visual proxy for force estimation in compliant, teleoperated grippers. Vision-based deformation metrics provide sensor-less force feedback in soft robotics and enable scalable solutions for precision teleoperation.A vision-based method was developed to automate the measurement of deformation in compliant mechanisms. Traditional force feedback systems involve excessive complexity, cost, or incompatibility with soft and miniaturized robotic applications. Literature supports vision-based techniques for non-intrusive force estimation. The proposed method introduces Distance-to-critical-point (Dcp) as a geometric metric for deformation tracking. Dcp was hypothesized to exhibit linear correlation with pinch force and to allow accurate extraction using computer vision methods. Experiments included varied jaw openings and object stiffness levels. Accuracy of Dcp measurements obtained by computer vision was compared to manual measurements using mean absolute error, and paired t-tests. The findings support that the computer vision method shows no statistical difference from manual measurement. Results revealed strong linear correlations between Dcp and pinch force in all test scenarios. Thus, computer vision method established Dcp as a consistent visual proxy for force estimation in compliant, teleoperated grippers. Vision-based deformation metrics provide sensor-less force feedback in soft robotics and enable scalable solutions for precision teleoperation.