Shape-Based-hand-Gesture-Recognition

Srishti 2020

Abstract

The aim of the project is to create hand bot which can replicate the movements of hands of the person who is interacting with the bot, with the help of image processing. This project is a combination of live motion detection and gesture identification.

Motivation

The motivation is to create a bot that can copy the movements of an original hand. With further motivations this bot can be upgraded to an artificial hand. This model is just to understand the coordination between the human and robotic actions. This bot aims in maximising of the patterns of the human hand, to replicate it.

Components

2x Arduino
2x NRF module
5x Servomotor
Laptop (for image processing)
5x 10k Resistors
9V Batteries
2x Breadboard
Jumpers
Braid wire (for the stifness of the fingures of the bot)
Copper wire

Workflow

Mechanical aspect of the design

Structure

The bot has 5 servo motors connected to separate copper strings which is joined to the fingures of the bot. When the arduino of the bot recieves the signals, the corresponding set of motor is switched on amd hence the copper string is pulled, thus, bending the corresponding fingure. When the srint is released the braid wire in the rear part of the fingure brings the fingure into its original state.

Electrical aspect of the design

Microcontroler

Two Arduino Uno are used, one in the receiver segment and other in the bot.

Sensor

The laptop camera works as a sensor in the project, for image identification.

Actuators

Five 90 degree servo motors are used in the bot for the comprassion and relaxation of the fingures.

Other components

2 nRF24L01 modules are used for the data transmission from the laptop to the bot.
10k resistors are attacher to the motors to reduce the effective voltage over the motors.

Cost Structure

Components	Cost(INR) per unit	Cost(INR)
Arduino Uno(x2)	499	998
nRF24L01 module(x2)	81	162
Servo motor(x5)	149	745
9V Battery	25	25
Total Cost	---	1930

Applications

For conrolling the movements of the robotic hands in factories for more precise works.
For upgrading the home automation technology, with the help of hand gesture recognition.

Limitations

the intermidiate state between straight and complete bend fingure cannot be taken into consideration.
the bot can hold items with less weight only.

Future Improvements

better length analysis will help in judging the semi-bend state,i.e., including the length from the palm point.
Image processing can be replaced by ultra sensitive flex sensors and small stepper motors for better precision of the motion.

anu-123-gif / shape-based-hand-gesture-recognition Goto Github PK