Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
A machine learning approach to compensate the drift in IMU sensors
Stable walking is one of the most fundamental dynamic movements to ensure the quality of lifestyle. However, the functionality of these systems deteriorates as people age, leading to a higher risk of falling. Fall-related injuries such as hip fracture, traumatic brain, and upper limb worsen the quality of lifestyle and put immense pressure on personal and social levels (1% of total annual healthcare expenditure; 25billion euros in Europe).
Diagnostic tools to monitor movement deficits and abnormal motor patterns reliably and objectively during daily activities outside of clinical settings are mostly missing. Hence, the interpretation of data is limited. The advanced development of inertial measurement units (IMU) provided an alternative measuring paradigm. However, the characteristic of IMU causes drift in the system, which gets accumulated during calculation, thus giving wrong results. This study aims to develop a robust drift-compensation algorithm to derive spatial gait parameters.
Keywords: Wearables, Balance, Neuromotor diseases, Movement deficits, Balance, Sensor fusion, Tracking, Human motion, Machine learning, IMU
Stable walking is one of the most fundamental dynamic movements to ensure the quality of lifestyle. The walking mechanism is often overlooked, despite the complexity of the regulation from multi-sensory systems; central nervous, peripheral nervous, and musculoskeletal system. However, the functionality of these systems deteriorates as people age, leading to a higher risk of falling. Fall-related injuries such as hip fracture, traumatic brain, and upper limb worsen the quality of lifestyle and put immense pressure on personal and social levels (1% of total annual healthcare expenditure; 25billion euros in Europe).
Spatio-temporal parameters extensively have been used as biomarkers to assess/monitor the quality of movement. However, these parameters are primarily available with expensive and labour-intensive indoor settings equipment such as an optical motion capture system. Diagnostic tools to monitor movement deficits and abnormal motor patterns reliably and objectively during daily activities outside of clinical settings are mostly missing. Hence, the interpretation of data is limited. The advanced development of inertial measurement units (IMU) provided an alternative measuring paradigm, requiring much less labour and data acquisition in different settings. Although it can provide comparable results/parameters to the optical motion capture systems, some parameters remain challenging due to the characteristic of IMU.
The study will primarily investigate building a drift-compensated sensor fusion algorithm to gain more accurate kinematic data yielding spatial parameters such as step width and step length. The work will be validated by an optical motion capture system (VICON).
Stable walking is one of the most fundamental dynamic movements to ensure the quality of lifestyle. The walking mechanism is often overlooked, despite the complexity of the regulation from multi-sensory systems; central nervous, peripheral nervous, and musculoskeletal system. However, the functionality of these systems deteriorates as people age, leading to a higher risk of falling. Fall-related injuries such as hip fracture, traumatic brain, and upper limb worsen the quality of lifestyle and put immense pressure on personal and social levels (1% of total annual healthcare expenditure; 25billion euros in Europe). Spatio-temporal parameters extensively have been used as biomarkers to assess/monitor the quality of movement. However, these parameters are primarily available with expensive and labour-intensive indoor settings equipment such as an optical motion capture system. Diagnostic tools to monitor movement deficits and abnormal motor patterns reliably and objectively during daily activities outside of clinical settings are mostly missing. Hence, the interpretation of data is limited. The advanced development of inertial measurement units (IMU) provided an alternative measuring paradigm, requiring much less labour and data acquisition in different settings. Although it can provide comparable results/parameters to the optical motion capture systems, some parameters remain challenging due to the characteristic of IMU.
The study will primarily investigate building a drift-compensated sensor fusion algorithm to gain more accurate kinematic data yielding spatial parameters such as step width and step length. The work will be validated by an optical motion capture system (VICON).
This project aims to build a drift-compensated fusion algorithm, in which the trajectory will be validated to the VICON.
Tasks:
- literature research (10%)
- Sensor fusion algorithm (40%)
- Data analysis and statistics (30%)
- Report/presentation (20%)
This project aims to build a drift-compensated fusion algorithm, in which the trajectory will be validated to the VICON.
Tasks: - literature research (10%) - Sensor fusion algorithm (40%) - Data analysis and statistics (30%) - Report/presentation (20%)
Yong Kuk Kim: yong.kim@hest.ethz.ch
Manuel Kaufmann: manuel.kaufmann@inf.ethz.ch
Yong Kuk Kim: yong.kim@hest.ethz.ch Manuel Kaufmann: manuel.kaufmann@inf.ethz.ch