Techniques for establishing biomechanical model through motion capture
Techniques for establishing a biomechanical model of a user performing motion are described. A plurality of sensor modules corresponding to a set of designated body parts (e.g., arms or legs) of the user generate sensing signals when the user performs a pose. Sensing data including accelerometers and/or gyroscopes data generated from the sensing signals is analyzed to detect a medio-lateral direction of each of the designated body parts to infer the pose from the captured motion of the user.
1. A method for establishing a biomechanical model through motion capture, the method comprising:
obtaining a set of sensing data in a computing device, the sensing data generated via a plurality of sensor modules, the sensor modules respectively corresponding to designated body parts of a user, each of the sensor modules including one or more of inertial sensors generating a sensing signal representing a movement of one of the designated body parts, wherein one of the sensor modules is designed as a hub module, others of the sensor modules are designated as satellite modules, the sensing data from the satellite modules are received in the hub module, and combined with the sensing data generated within the hub module, transported to the computing device;
estimating in the computing device respective sensor-to-segment (STS) quaternions of the designated body parts from the sensing data;
estimating a medio-lateral direction of each of the designated body parts; and
finalizing the medio-lateral direction using some or all of the sensing data from the sensor modules.
2. The method as claim in claim 1 , wherein the inertial sensors are either motion sensors including accelerometers or gyroscopes.
3. The method as claim in claim 2 , wherein said estimating respective STS quaternions comprises: producing an estimate of a reaction of the each of the designated body parts to a gravity detected from a sensor reference frame.
4. The method as claim in claim 3 , wherein the medio-lateral direction of the each of the designated body parts is estimated with respect to the sensor reference frame.
5. The method as claim in claim 4 , wherein the medio-lateral direction is assigned a priori according to a physical orientation of an inertial sensor with respect to a corresponding designated body part of the user.
6. The method as claim in claim 4 , wherein said producing an estimate of a reaction of each of the designated body parts to a gravity comprises: measuring and averaging accelerometer data during a time a predefined pose is performed by the user.
7. The method as claim in claim 6 , wherein said producing an estimate of a reaction of each of the designated body parts to a gravity further comprises: normalizing gyroscope data to unity.
8. The method as claim in claim 4 , wherein said estimating of a medio-lateral direction of the each of the designated body parts further comprises: refining the medio-lateral direction based on computation of a cost function over a pool of candidate directions.
9. The method as claim in claim 8 , wherein the pool of directions lies in a horizontal plane and are equally distributed in a range of +/−60 degrees with respect the medio-lateral direction initially estimated.
10. The method as claim in claim 9 , wherein the horizontal plane is computed based on the accelerometer data during the time the predefined pose is performed by the user.
11. The method as claim in claim 10 , wherein the cost function is based on assumptions that functional motions are performed within a sagittal plane with null pronation.
12. The method as claim in claim 1 , further comprising:
approximating a pose in a sport with a quasi-static condition lasting for a predefined time;
computing respective attitudes of the designated body parts; and
inferring the pose in real time.
13. The method as claim in claim 12 , wherein said inferring the pose in real time comprises: detecting a model of the pose based on some or all of accelerometers data or attitudes.
14. The method as claim in claim 13 , wherein the model comprises specific tolerance bounds for each of the designated body parts.
15. The method as claim in claim 13 , wherein said inferring the pose in real time further comprises: detecting absence of significant motion of some or all of the designated body parts over a pre-defined time window.
16. The method as claim in claim 1 , wherein each of the satellite modules includes a microcontroller, at least an inertial sensor and a transceiver for intercommunication with the hub module, the hub module includes a microcontroller, at least an inertial sensor and a transceiver for intercommunication with the satellite modules respectively and another transceiver for communicating wirelessly with the computing device, each of the sensor modules produces sensing data at a predefined frequency when the user makes moves.