Apprentice Arm

Kurapati Sasi Srija , Sidagam Anirudh , Sadhanala Himasri , P.V.S Harsha Vardhan , Vemula Bhaskar Ajay

Keywords: Hand Gesture Recognition, 6-DOF Robotic Arm, Human-Robot Interaction (HRI), MediaPipe Hands, Computer Vision, Machine Learning, K-Means Clustering, Smart Presets, Exponential Moving Average (EMA), Deadband Filtering, Arduino UNO, PCA9685 Servo Driver, Real-Time Control System, Flask Web Application, Three.js Visualization.

Apprentice Arm is a 6-Degree-of-Freedom, hand-gesture controlled, real-time. (6-DOF) robotic arm system based on the use of computer vision, machine learning, and embedded systems to allow contactless, intuitive robotic manipulation. The system will live video on a regular USB webcam, and will extract 21 three sided video. dimensional hand landmarks with MediaPipe Hands deep-learning of Google. pipeline, and maps the spatial geometry of the hand of the user into six. independent servo motor angles Base rotation,Shoulder lift, Elbow bend, Wrist Pitch, Wrist Roll and Gripper open/close-in real time. The software layer is developed using a Python Flask back-end which integrates. Hand tracking using MediaPipe, Exponential Moving Average (EMA) signal. deadband filtering, smoothing and serial communication into one. cohesive application. The backend is annotated MJPEG video which is being streamed to a modern. web dashboard (HTML5/CSS3/JavaScript) that features a live camera feed, an interactive Three.js 3D digital twin of the robotic arm, six manual override. sliders, a complete task recorder and player engine and an ML-based Smart. Presets module which opts K-Means clustering (scikit-learn) to automatically. find and propose the most common arm positions based on recorded. task history. Hardware On the hardware side, an Arduino UNO microcontroller receives commands regarding the angles. over a 9600-baud UART serial connection and drives a 16 channel I2C PWM, the PCA9685. servo driver. The Base, Shoulder, and have three MG996R high-torque servos. Elbow joints: There are three SG90 micro-servos that drive the Wrist Pitch, Wrist. Roll, and Gripper. The servos use two 3.7 V lithium-ion as a power source. batteries (in series) (~7.4 V), reduced to a regulated (~5.5 V). with an LM2695 DC-to-DC buck converter and then to the V+ of PCA9685. servo power rail. The Apprentice Arm has a gesture to servo end-to-end latency of. around 60 ms, a hand-tracking frequency of 30 FPS, a refresh of the dashboard. frequency of 12.5 Hz, and 3D frame rate of over 60 FPS. The system is developed to be modular, reproducible and extensible, which is why it is suitable. in the case of academic research, educational demonstrations, and hobbyist robotics. HRI projects.

Volume 14 Issue 4, April-2026

Pages : g475-g483