Autonomous shuttlecock collecting robot
Badminton is one of the favorite sports among the society in this century. The badminton players range from male to female and youths to elderlies. Due to its popularity, a lot of badminton equipment are invented to increase the efficiency of the training. Besides the basic equipment such as racquet...
Saved in:
Main Author: | |
---|---|
Other Authors: | |
Format: | Final Year Project |
Language: | English |
Published: |
Nanyang Technological University
2021
|
Subjects: | |
Online Access: | https://hdl.handle.net/10356/150420 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Institution: | Nanyang Technological University |
Language: | English |
Summary: | Badminton is one of the favorite sports among the society in this century. The badminton players range from male to female and youths to elderlies. Due to its popularity, a lot of badminton equipment are invented to increase the efficiency of the training. Besides the basic equipment such as racquet and shuttlecock, there are also badminton equipment such as badminton equipment bag, badminton headband, wrist band, badminton shoes, badminton shorts and shuttlecock launcher available in the market. Shuttlecock launcher is popular for badminton training, however there is nothing available to collect the shuttlecocks currently. Therefore, the purpose of this project is to design and develop an autonomous shuttlecock collecting robot. With this robot, the training efficiency of the badminton players can be increased significantly as the players can concentrate of their training since the shuttlecock collecting task can be done by the robot. The report will discuss on the software system of the robot which focus on the shuttlecock detection by deep learning model, optimization of the deep learning model in order to implement on microcontroller and the navigation system of the robot. In this project, SSD model trained by TensorFlow is able to achieve an inference speed of 15fps when performing shuttlecock recognition on a laptop without GPU. The SSD model is then optimised by Intel OpenVino and result in a significant increase of inference speed (max 35fps) during the shuttlecock recognition on a resources constraint microcontroller, Raspberry Pi 4 Model B. At the end of the project, an autonomous shuttlecock collecting robot is successfully developed and it is able to detect shuttlecocks by camera and navigate to the location of the shuttlecocks for collecting purpose. |
---|