Vision-based robot indoor navigation

Teo, Zhin Hang (2022) Vision-based robot indoor navigation. Final Year Project, UTAR.

Preview

PDF Download (5Mb) | Preview

Abstract

This report discusses a vision-based indoor navigation robot training system. The need for manufacturing workers has been increasing throughout the years. Many labours are required in businesses to transfer items and components all around. Many companies still use manual labour for part delivery within the factory, where the completion progress and time depend highly on the workers. With that said, any possible threat may stop the business's operation, making a loss to the company. Indoor navigation robots can close the gap in manual labour. This project aims to develop a vision-based navigation robot using OpenCV and C++ programming language in an indoor environment. The objective of this project is to design, develop and simulate programming code to perform image processing and path planning, integrate programming code with a microcontroller wirelessly, and test and evaluate the performance of the navigation robot. The computer and ESP32 board are the central processing unit for this project to execute path planning and motor command analysis. NodeRed links both processing units together. An algorithm is developed in the computer to achieve image processing, user input, path planning, simulation, and writing of output files. A different algorithm is created in the microcontroller to derive and perform the data delivered from the computer. A navigation robot is built to test the workability and efficiency of the algorithms. In general, the algorithm can provide the nearest path to navigate around the environment without manual assistance. The user will only have to give the program relevant coordinate information and running mode. To conclude, a vision-based indoor navigation robot training system is effectively established. Implementing this system could effectively help increase product delivery within the environment. This system supports one end coordinate or multiple saved coordinates, where a supply chain operation structure can be implemented above it to administer all the operations fully.

Actions (login required)