Design And Development Of A Machine Learning-Based Gesture-Controlled 3D-Printed Robotic Hand Using Computer Vision

Abstract

This paper introduces a 3D-printed humanoid robotic hand made from a bio-based plastic material, PLA (Polylactic Acid), controlled through human-machine collaboration utilizing machine learning and computer vision algorithms. The system is designed to replicate real-time actions through gesture recognition, a noninvasive technique that enables a wide range of real-time interactions. A Kinect sensor is employed due to its multiple features, allowing for precise and versatile user interaction. The paper suggests using a non-invasive 3D sensor to regulate manipulation in real time. The Kinect sensor can find objects in three dimensions and recognise gestures in a realistic way. Using skeletal data from the Kinect, the status of the user’s right arm is captured and sent to the 3D-printed robotic hand. Mimicking human movements in this way is an effective method for controlling robotic operations. The system is calibrated to recognize specific gestures, accurately identifying the user’s hand state and angles, streamlining the process and enhancing overall efficiency.