Vision-based Robot Navigation System for Medical Setting

Lim, Wei Cheei (2020) Vision-based Robot Navigation System for Medical Setting. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Lim Wei Cheei Full Text.pdf Restricted to Registered users only Download (4MB)

Abstract

Robotic technology had been advanced since the early 2000s and had been a popular research. The aim of constant development of this technology is to aid human beings from task which human are limited to. This research is conducted to design a vision-based robot navigation algorithm which target medical setting environment. The algorithm is designed to resolve wayfinding problem occurring in hospital facilities by navigating patients or visitor to the desired destination within the shortest path available. The objective of this research is to design a visionbased algorithm which are able to traverse in the ground floor of a medical setting environment. A ratio of 1:439.5 of Hospital University Kebangasaan Malaysia structural model will be use as tool to validate the algorithm. The designed algorithm uses vision sensor as main input to produce data by undergoing image processing which further use in self-localization and path planning. A mobile robot is designed using on-board processor, Raspberry Pi 3 model B+ equipped with Raspberry Pi 8MP camera will be implemented on 2-wheel differential mobile robot chassis mounted on both side of wheels. Using Python 3 programming language, the algorithm is designed using various open source library files such as OpenCV, A*star path planning. This research proposed an algorithm which allowed the mobile robot to self-localize by recognizing the colour feature in the surrounding environment. Besides, the algorithm also allows the mobile robot to maneuver to the destination through a shortest path computed by path planning algorithm. Experiments and testing of the algorithm will be carried out using computer simulation and implementation of physical mobile robot in the scaled-down hospital structural model. Both simulation and physical experiment are conducted on A*star planning and A*star path planning with vision localisation. Both type of experiment is carried out with every possible combination of starting point and ending point with 5 attempt each. The both experiments are conducted and the success rate of the mobile robot completing the destination path is tabulated. Success rate of mobile robot completing destination path obtained from simulation experiments are successful with 100% passing rate. However, the success rate of mobile robot manoeuvring in physical structural model is significantly low with average passing rate only 30%. Hardware design of the mobile robot and the uncontrollable environment of testing field are the major drawback of the success rate of this designed algorithm. Summing up the research, the designed algorithm is successful.