Artificial intelligence (AI) algorithms for virtual reality (VR) golf

• Main Author: Looi, Denzel Xiongkai
• Other Authors: Lin Feng
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2020
• Subjects: Engineering::Computer science and engineering
• Online Access: https://hdl.handle.net/10356/144797
• Institution: Nanyang Technological University

In Asia, Singapore has the second highest prevalence of stroke. Over the past 10 years, the crude incidence rate increased by 22%, while the fatality rate dropped by 2.4%, indicating an increase in stroke survivors. Stroke survivors tend to have certain disabilities, such as paralysis or problems controlling movement, sensory disturbances including pain, problems with thinking and memory. It has been shown that exercise therapy aid in patients’ recovery and that golf training could be one form of exercise. Due to the disabilities of these patients, handling a club could be difficult, therefore creating a virtual reality (VR) golf game that uses a gaming controller that is lighter than a club would be preferred for rehabilitation and in the long run cheaper as compared to actual golfing. The VR golf game would be implemented using Unity game engine to build and run on an Oculus Quest (Gen 1). VR golf game would include a full 18-hole golf course as well as a driving range with both chipping and putting green. To evaluate the performance of a patient, three mini-games were included, Go/No-Go, Block Tapping and Mental Rotation. Each of these mini-games measures a single mental capability of a patient. If possible, implement an AI algorithm to analyse a player’s swing to indicate just how well an in-game swing was. The implementation of the golf game only covered the core aspect of golf, such as hitting the ball and switching clubs. Additional features were developed into the game, these features could only be implemented in VR and not in real life, examples are ball tracing for the golfer to see where the ball went, teleporting from hole to hole, as well as summoning of club and golf balls. However, the game did lack the realism of golfing, to name a few aspects would be the sounds heard in real life, wild animals, hitting of a ball, walking on different types of land as well as the haptic feedback from controllers. The physics of the game was not 100% realistic as well as the low fidelity of the game environment. At the end of this project, the initial AI Algorithm idea was never implemented due to time constraints. The game itself only covered the essence of a golf game. In more recent updates, Oculus Quest is capable of hand tracking, which would eliminate the need for controllers all together. This meant that patients would only require only the Oculus Quest headset to start their rehabilitation. In the market, there are currently other VR headsets that do full-body tracking. With this, it could improve the realism of the game in the future. I hope that this game can be of use to any post-stroke patients for their continuous rehabilitation.