Presentation in Ubicomp/ISWC 2022

Hi, I am Chengshuo Xia. I have presented three reports in 2022 Ubicomp/ISWC at Atlanta, USA from 9.11 - 9.15, including 1 IMWUT paper and 2 ISWC short papers. Related papers’ name are:

• IMWUT paper: VoLearn: A Cross-Modal Operable Motion-Learning System Combined with Virtual Avatar and Auditory Feedback

• ISWC short paper: Knee Osteoarthritis Classification System Examination on Wearable Daily-Use IMU Layout

• ISWC short paper: Virtual IMU Data Augmentation by Spring-Joint Model for Motion Exercises Recognition without Using Real Data

Research Summary

All three reports are related to the motion and exercise research, including the motion training system development, motion disease screening and the novel sensing technique for human motion. I would introduce each report separately. 

VoLearn

VoLearn is a motion editing and guidance system that allows a 2D motion video to be converted into a 3D motion. By using the system’s user interface, the user could customize the motion regarding its amplitude and speed. It also allows combining the two separate motions together to form a new one. After a customized motion is designed, the system would analyze the motion and generate a recommended sensor placement to track the user’s real motion. In the real world, the users study the motion by wearing their smartphone on the placement where the system recommends. The smartphone-based application would detect the user’s moving data and compare it with designed/reference motion data to assist the user study/train the motion. The smartphone will generate the auditory feedback to the users to enable the users to learn the motion in terms of its position and speed.

KneeOA

This paper investigated the IMU-based machine learning system to diagnose knee OA disease. As the IMU would become more tiny and light, it can be embedded into the daily wearable object easily and more body limb’s kinematic data can be captured. The paper mentioned 13 sensor placement on the body and using the machine learning technique to build the screening system. The paper evaluated the best sensor wearing placement with 1 to 3 sensors. 

Proposed System

VirtualAugmentation

Recently, using the virtual sensor data to build the machine learning system has attracted a lot of attention. This paper proposed a method that uses the physical simulation method to generate various virtual acceleration distributions for one motion. It aims to use a limited length of 3D motion to generate more virtual acceleration data. Reduce the development cost. Thus, improve the flexibility of the exercise recognition system to identify categories.

Overview of the designed virtual IMU data augmen- tation method and its corresponding usage.

The feedback you got after the presentation

For my three presentations, each has two questions from the audience. The most lively discussion was the first presentation, “Virtual Augmentation”. A famous professor from Georgia Technology College asked me about the design motivation of this work and compared it with other popular work. And after the presentation, the other student from Georgia Tech came to discuss with me about the virtual IMU data development and application. And the VoLearn’s received the questions regarding the technique details and application extending. I discussed deeply with a lecture from University College London about using IMU to detect the exercise and other interesting issues in training a user, such as the users normally cheating the system that they actually did not complete the motion. The KneeOA presentation received the questions of data augmentation usage and the early stage disease detection. I also talked with a Ph.D student from The Hong Kong University of Science and Technology who cooperated with a famous hospital’s rehabilitation department in China. We discussed any future direction of rehabilitation and exercise training.

Thoughts on Participation in the conference

This is my first time attending a top-tier conference since COVID-19 in person. Before this time, I had been joining other conferences online. Actually, I really liked the in-person conference. Because it gives me a lot of opportunities to talk with other excellent researchers deeply. And it is also important to exchange ideas and discuss the direction.

[1] Chengshuo Xia, Xinrui Fang, Riku Arakawa, and Yuta Sugiura. 2022. VoLearn: A Cross-Modal Operable Motion-Learning System Combined with Virtual Avatar and Auditory Feedback. Proc. ACM Interact. Mob. Wearable Ubiquitous Technol (PACM IMWUT) (Ubicomp) 6, 2, Article 81 (June 2022)

[2] Chengshuo Xia, Tsubasa Maruyama, Haruki Toda, Mitsunori Tada, Koji Fujita, and Yuta Sugiura. 2022. Knee Osteoarthritis Classification System Examination on Wearable Daily-Use IMU Layout. The 2022 International Symposium on Wearable Computers (ISWC ’22), September 11–15, 2022, Cambridge, United Kingdom. ACM, New York, NY, USA

[3] Chengshuo Xia and Yuta Sugiura. 2022. Virtual IMU Data Augmentation by Spring-Joint Model for Motion Exercises Recognition without Using Real Data. The 2022 International Symposium on Wearable Computers (ISWC ’22), September 11–15, 2022, Cambridge, United Kingdom. ACM, New York, NY, USA

Presentation Slide

Presentation Movie