Home Publications Research Grants Inventions & Patents Awards Additional Research Activities Faculties & Institutions Research Areas

Combining Pose Estimation and inertial measurement tracking data for a biomechanical simulation of river wave surfing using optimal control

Weiß A, Koelewijn A, Masmoudi I, Lluch È (2024)

---

Publication Type: Conference contribution

Publication year: 2024

Event location: Glasgow/Scotland

Open Access Link: https://www.ecss.mobi/DATA/CONGRESSES/GLASGOW_2024/DOCUMENTS/2024_BOA_Web.pdf

Abstract

 INTRODUCTION: 

River wave surfing is gaining traction as a land-based alternative to traditional ocean surfing. However, its complexity and dynamics present challenges in sensor selection for biomechanical analysis, which is valuable for optimizing performance and preventing injuries. We have developed an inertial measurement unit (IMU)-based simulation approach for gait. We started to apply that to three-dimensional motions, but it is not working robustly yet (1). Recent advancements in human pose estimation however show precise spatial tracking but are prone to occlusions and changing lighting conditions and are bound to the capture volume (2). Inspired by (3), we propose integrating data from IMUs and pose estimation techniques within an optimal control framework to provide insights into both kinematics and kinetics in river wave surfing. 

METHODS: 

One participant was instructed to surf freely in a river wave. He was equipped with ten IMUs, which were aligned to the corresponding body parts via calibration movements (1). Additionally, his movements were recorded from four viewpoints with RGB cameras synchronized with the IMUs. Nine single surf cycles were identified using the gyroscope data. From the camera recordings, we predicted planar joint angles using RTMPose (4), and from that three-dimensional joint angles through MotionBERT (5). We then created surfing simulations by solving optimal control problems (1, 6) on a 23-degree-of-freedom biomechanical model with 92 muscle-tendon units (1). In this optimal control problem, we tracked the joint angles from the camera images as well as the raw accelerometer and gyroscope data from the IMUs. The interaction between the board and water was adjusted from skiing simulations (6). We then analyzed the kinematic and kinetic differences in the hip and knee between front (FT) and back turns (BT) and between the front and rear stance leg within each surf cycle. 

RESULTS: 

Comparing front and back turns, we found only minor differences in peak hip (BT: 57.2°, FT: 56.4°) and knee (BT: 64.5°, FT: 65.8°) angles and hip (BT: 25.5 Nm, FT: 27.9 Nm) and knee (BT: 44.5 Nm, FT: 51.4 Nm) moments. Comparing the front and rear leg, the peak knee moment was on average 61.8% higher on the rear leg (87.8 Nm) over 9 turns; peak rear leg hip moment was only 4.5% higher (32.6 Nm). The absolute peak front leg joint angle was 18.5% higher for hip flexion (58.7°), and 8.2% higher for front leg knee flexion (69.8°). 

CONCLUSION: 

Our results provide the first biomechanical simulations of river wave surfing. We found that the peak joint angles of the front leg are higher, while the joint moments are on average increased in the rear leg joints. That indicates that strength is more important for the rear leg while flexibility is an important ability for the front leg. 

Authors with CRIS profile

Involved external institutions

How to cite

APA:

Weiß, A., Koelewijn, A., Masmoudi, I., & Lluch, È. (2024). Combining Pose Estimation and inertial measurement tracking data for a biomechanical simulation of river wave surfing using optimal control. In Proceedings of the 29th Annual Congress of the EUROPEAN COLLEGE OF SPORT SCIENCE. Glasgow/Scotland.

MLA:

Weiß, Alexander, et al. "Combining Pose Estimation and inertial measurement tracking data for a biomechanical simulation of river wave surfing using optimal control." Proceedings of the 29th Annual Congress of the EUROPEAN COLLEGE OF SPORT SCIENCE, Glasgow/Scotland 2024.

BibTeX: Download