![The dataset proposed by the researchers captures badminton players' movements and responses, aiding AI-driven coaching assistants to improve stroke quality for all skill levels. Credit: SeungJun Kim at Gwangju Institute of Science and Technology (GIST) GIST-MIT CSAIL researchers develop a biomechanical dataset for badminton performance analysis](https://scx1.b-cdn.net/csz/news/800a/2024/gist-mit-csail-researc.jpg)
In sports activities coaching, apply is the important thing, however with the ability to emulate the methods {of professional} athletes can take a participant’s efficiency to the subsequent stage. AI-based customized sports activities teaching assistants could make this a actuality by using revealed datasets. With cameras and sensors strategically positioned on the athlete’s physique, these methods can monitor every thing, together with joint motion patterns, muscle activation ranges, and gaze actions.
Utilizing this knowledge, customized suggestions is supplied on participant approach, together with enchancment suggestions. Athletes can entry this suggestions anytime, and wherever, making these methods versatile for athletes in any respect ranges.
In a examine revealed within the journal Scientific Information on April 5, 2024, researchers led by Affiliate Professor SeungJun Kim from the Gwangju Institute of Science and Expertise (GIST), South Korea, in collaboration with researchers from Massachusetts Institute of Expertise (MIT), CSAIL, U.S., have developed a MultiSenseBadminton dataset for AI-driven badminton coaching.
“Badminton may gain advantage from these numerous sensors, however there’s a shortage of complete badminton motion datasets for evaluation and coaching suggestions,” says Ph.D. candidate Minwoo Seong, the primary creator of the examine.
The examine took inspiration from MIT’s ActionSense mission, which used wearable sensors to trace on a regular basis kitchen duties similar to peeling, slicing greens, and opening jars. Seong collaborated with MIT’s group, together with MIT CSAIL postdoc researcher Joseph DelPreto and MIT CSAIL Director and MIT EECS Professor Daniela Rus and Wojciech Matusik. Collectively, they developed the MultiSenseBadminton dataset, capturing actions and physiological responses of badminton gamers.
This dataset, formed with insights from skilled badminton coaches, goals to reinforce the standard of forehand clear and backhand drive strokes. For this, the researchers collected 23 hours of swing movement knowledge from 25 gamers with various ranges of coaching expertise.
Through the examine, gamers have been tasked with repeatedly executing forehand clear and backhand drive photographs whereas sensors recorded their actions and responses. These included inertial measurement models (IMU) sensors to trace joint actions, electromyography (EMG) sensors to observe muscle indicators, insole sensors for foot stress, and a digicam to file each physique actions and shuttlecock positions.
With a complete of seven,763 knowledge factors collected, every swing was meticulously labeled primarily based on stroke kind, participant’s ability stage, shuttlecock touchdown place, impression location relative to the participant, and sound upon impression. The dataset was then validated utilizing a machine studying mannequin, guaranteeing its suitability for coaching AI fashions to guage stroke high quality and provide suggestions.
“The MultiSenseBadminton dataset can be utilized to construct AI-based schooling and coaching methods for racket sports activities gamers. By analyzing the disparities in movement and sensor knowledge amongst totally different ranges of gamers and creating AI-generated motion trajectories, the dataset will be utilized to customized movement guides for every stage of gamers,” says Seong.
The gathered knowledge can improve coaching by means of haptic vibration or electrical muscle stimulation, selling higher movement and refining swing methods. Moreover, participant monitoring knowledge, like that within the MultiSenseBadminton dataset, may gasoline digital actuality video games or coaching simulations, making sports activities coaching extra accessible and reasonably priced, probably remodeling how individuals train.
Extra info:
Minwoo Seong et al, MultiSenseBadminton: Wearable Sensor–Based mostly Biomechanical Dataset for Analysis of Badminton Efficiency, Scientific Information (2024). DOI: 10.1038/s41597-024-03144-z
Quotation:
Researchers develop a biomechanical dataset for badminton efficiency evaluation (2024, Might 6)
retrieved 7 Might 2024
from https://techxplore.com/information/2024-05-biomechanical-dataset-badminton-analysis.html
This doc is topic to copyright. Other than any honest dealing for the aim of personal examine or analysis, no
half could also be reproduced with out the written permission. The content material is supplied for info functions solely.