Image Recognition Robotic Car

CZ3004 Multidisciplinary Design Project
2022.07 @ Nanyang Technological University
The project is created for NTU computer science module CZ3004.

Project Purpose

On a 2 meters by 2 meters map, 5 obstacles are randomly placed on the map.

There is 1 distinct symbol pasted on one side of each obstacle.

Goal is to develop a robotic car that can find the shortest path to scan the sign on the obstacle,

and accurately recognise the symbol.

Presentation Video Link: YouTube Video Link

The ultimate objective

1. Send the obstacles information from controller app on Android tablet to robotic car.
   
2. The car automatically locates the 5 obstacles, find the shortest Hamiltonian path to traverse to all obstacles.
   
3. When the car reaches each obstacles, activates the pi-camera to scan the image and detect the symbol.
   
4. After detection is done, the car send back the image id to the controller app on Android tablet.

Table of contents

This repo only consists of 2 components:

1. algo:

Algorithm to plan a path to traverse all 5 obstacles placed on the map.

How to use:
Refer to "main.py" for example of usage

2. image:

Image recognition to detect the symbol pasted on the obstacles.
Machine Learning Model - TensorFlow Lite
Computer Vision Tool - OpenCV

How to use:
On Rpi, install tflite_runtime, opencv. Then, refer to "main.py" for example of usage