From motion to trusted measurement
Turning everyday sensors into reliable, decision-grade signals
Embedded Files
SiMuR
Multisensor Systems and Robotics
The Multisensor Systems and Robotics (SiMuR) group works on the design and characterization of wearable multisensor systems for the analysis of human movement, with the aim of turning movement into a signal that can be measured continuously and reliably, beyond the laboratory.
Our work focuses on two areas: reliable human movement measurement and the interpretation of real-world activity through models that generalize across people, conditions, and devices.
Our main application domain is active and healthy ageing, where we use objective measures of physical activity, gait and mobility to assess autonomy in daily life. Our methods also address rehabilitation, sports, occupational health and human–robot interaction, turning raw sensor data into reliable, actionable information.
Two Main Research Areas
Two Main Research Areas
Portable Systems for Reliable Human Motion Measurement and Estimation
Portable Systems for Reliable Human Motion Measurement and Estimation
We develop technologies to estimate key movement variables—including segment orientation, joint angles, gait events, contact times, speed, distance travelled and changes in direction—from wearable sensors.
The goal is to achieve accurate, traceable and uncertainty-aware measurements using portable, affordable and minimally intrusive instrumentation, reducing the need for laboratory-based mocap systems.
Our work combines sensor calibration, signal processing, biomechanical modelling and experimental validation, with optical mocap used as a reference when required in the MovLab motion-capture laboratory.
Learning-Based Human Motion Analysis in Real-World Conditions
Learning-Based Human Motion Analysis in Real-World Conditions
We develop learning-based methods to extract meaningful information from human movement data collected during free-living activities and under real-world conditions.
The goal is to develop robust and transferable models for human movement analysis that can operate reliably across subjects, conditions and sensing technologies, supporting applications outside the laboratory.
Our work explores representation learning, domain adaptation and learning from partially annotated data, while ensuring that uncertainty and limitations are explicitly characterised.
Representative Projects
Representative Projects
Sponsors
Sponsors
Page updated
Report abuse