A robotic hand that mimics a human hand opening and closing in real time via camera hand tracking, sending finger states over Bluetooth to an Arduino UNO driving servos.
An academic project titled "Design and Implementation of a Robotic Hand Mimicking Human Hand Motion Using Artificial Intelligence and Computer Vision", carried out by a team of students with technical assistance from Techno Enjaz. The system captures the user's hand with a camera, locates the finger joints and infers the state of each finger (extended or folded), then sends this state wirelessly over Bluetooth HC-05 to an Arduino UNO board, which drives the servo motors of the robotic hand to mimic the same motion.
The project is an applied example of Human–Robot Interaction and gesture control without gloves or sensors on the hand. The prototype successfully mimicked opening and closing the fingers in real time, and the tests showed that the wireless link was more efficient and reliable than a wired link for controlling the hand.
| Item | Details |
|---|---|
| Project type | Prototype of a robotic hand controlled by hand gestures |
| Field | Robotics, computer vision, pose estimation, embedded systems |
| Project status | Implemented prototype tested in practice |
| Techno Enjaz role | Technical support and assistance to the students during project development |
| Core technologies | Computer-vision hand tracking and pose estimation, Arduino UNO, servo motors, HC-05, DC-DC step-down regulator |
| Outputs | Robotic hand, finger-state detection, five-value control array, wireless control system for six servo motors |
The human hand is a highly complex system with many degrees of freedom: the fingers, wrist, and palm can work together or independently. Conventional robots usually rely on fixed, pre-programmed instructions, so they cannot interact naturally with changing tasks, and it is hard to build a traditional mathematical model that reproduces every hand movement.
The project proposes using computer vision and AI instead: a camera captures the hand's motion and algorithms turn it into control commands, so a robotic hand can be driven by natural hand movement rather than fixed programming. The project sees potential for this idea in smart prosthetics, assistive devices for people with disabilities, and remote control in hazardous environments.
The project was carried out academically by the students, with technical assistance and support from Techno Enjaz during development, covering the integration of the computer-vision components with embedded control and the development of the operating experience. It is presented among the office's work as an example of its support for robotics and gesture-control projects.
The project's flowchart describes a loop that repeats for every video frame:
The book devotes a full chapter to pose estimation and Google's MediaPipe library, including the MediaPipe Hands model, which locates 21 key points on the hand covering the joints, fingertips, and wrist, and works with simple cameras such as a laptop webcam. The book explains its stages: detecting the hand on the first frame or when it is lost, cropping the hand region, then estimating the 21 landmarks, using the previous frame to speed up tracking.
The book also reviews the BlazePose body-pose model and Face Mesh face tracking, alongside the basics of digital image processing, machine learning, deep learning, and convolutional neural networks as the project's theoretical foundation.
| Component | Role in the system |
|---|---|
| Camera | Captures the user's hand in real time |
| Arduino UNO (ATmega328P microcontroller) | Main controller: receives the finger states and generates PWM signals for the motors |
| Six servo motors (SERVO1 to SERVO6) | Move the parts of the robotic hand to set angles |
| HC-05 Bluetooth module | Short-range wireless link over UART between the computer and the controller |
| DC-DC step-down regulator | Lowers the supply voltage to a suitable, stable level for the board and motors |
| Power source | Battery or adapter powering the system |
The book lists the HC-05's specifications as Bluetooth V2.0 in the 2.4–2.485 GHz band, with a typical range of 10 to 100 meters and an operating voltage of 3.3 to 5 V.
According to the practical results reported in the book:
These are descriptive results from a practical demonstration; the book does not report quantitative measurements such as finger-state recognition accuracy, response time in milliseconds, or the number of trials and users.
According to the future outlook in the project:
If you are working on a motion-tracking robotic hand and need technical help developing the idea or building a prototype, you can contact the Techno Enjaz office to discuss your project's requirements and the appropriate scope of support.
The camera captures each frame and converts it to RGB, then a hand-detection algorithm locates the hand's key points and infers each finger's state. A five-value array is built, where 1 means an extended finger and 0 a folded one, and is sent over Bluetooth to the Arduino UNO, which drives the servos so the hand mimics the same state.
The prototype includes a camera to capture the hand, an Arduino UNO with the ATmega328P microcontroller, six servo motors to move the hand's parts, an HC-05 Bluetooth module for wireless UART communication, and a DC-DC step-down regulator with a power source.
The project relies on computer-vision hand pose estimation, and the book explains Google's MediaPipe Hands model, which locates 21 key points on the hand covering the joints, fingertips and wrist and works with simple cameras. Each finger's state is inferred from these points to generate control commands.
In the practical demonstration the hand mimicked opening and closing the fingers instantly and smoothly, and tests showed the wireless link was more efficient and reliable than a wired one for controlling it. The book reports no quantitative measurements such as recognition accuracy or response time.
The current version handles only two states per finger, grasping objects was not documented, and tracking accuracy is affected by lighting and occlusion. The project suggests improving detection with deep learning, using stepper or encoder-equipped motors, adding touch, force and depth sensors, and developing the hand into a full robotic arm.
Reach out to Techno Enjaz engineering team to discuss technical feasibility, prototype hardware, and development scope.
Discover more engineering prototypes and academic implementations
A remotely controlled ground robot prototype combining Arduino UNO, three wireless links (Bluetooth, Wi-Fi, 433 MHz RF) and a camera whose images are analyzed with YOLO.
A prototype that measures a tank's water-level percentage with an HC-SR04 sensor and Arduino Nano, showing it on an LCD and sending it via HC-05 Bluetooth to a phone app.
A sequential control system for a chemical mixing station using a Delta PLC and Ladder logic, with level sensors, temperature monitoring and a DOPSoft HMI, tested in simulation.
