A wearable motion capture device able to detect dynamic motion of human limbs

Limb motion capture is essential in human motion-recognition, motor-function assessment and dexterous human-robot interaction for assistive robots. Due to highly dynamic nature of limb activities, conventional inertial methods of limb motion capture suffer from serious drift and instability problems. Here, a motion capture method with integral-free velocity detection is proposed and a wearable device is developed by incorporating micro tri-axis flow sensors with micro tri-axis inertial sensors. The device allows accurate measurement of three-dimensional motion velocity, acceleration, and attitude angle of human limbs in daily activities, strenuous, and prolonged exercises. Additionally, we verify an intra-limb coordination relationship exists between thigh and shank in human walking and running, and establish a neural network model for it. Using the intra-limb coordination model, dynamic motion capture of human lower limbs including thigh and shank is tactfully implemented by a single shank-worn device, which simplifies the capture device and reduces cost. Experiments in strenuous activities and long-time running validate excellent performance and robustness of the wearable device in dynamic motion recognition and reconstruction of human limbs.


S-5
Supplementary Fig. 4| (Subject 4) Results of lower limb motion capture in the sagittal plane by using single device worn on the shank and determining the thigh motion from the shank motion by the trained neural network model of intra-limb coordination. The pitch angles of thigh θt and shank θs are used to represent the corresponding elevation angles in the sagittal plane respectively. The joint angle of knee β in the sagittal plane is derived by subtracting θs from θt. The error of β (Eβ) represents the estimation performance of both θt and β, including the measurement error of the wearable device and the model error of the neural network. Three repeated experiments are conducted continuously.

Supplementary Fig. 5| Maximum deflection of shank.
Deflection of shank is the angle between the shank and the vertical and is represented by |θs| here. Maximum deflection of shank reveals the ability of lifting one's heel and can be used as an indicator of fatigue. For each subject, the maximum deflection of shank |θs| peaks in each repeated experiment of lower limb motion capture is evaluated by averaging peak values of |θs| during the last 25 s running on a treadmill at velocity 10 km/h.

S-7
Supplementary Fig. 6| Comparison between knee flexions of subject with knee injury and those of healthy ones. a, Knee flexion ability of four subjects while walking on a treadmill at velocity 5 km/h. b, Knee flexion ability of four subjects while running on a treadmill at velocity 10 km/h. People with knee injury (Subject 4) suffer from weakened knee flexion ability and thus present smaller maximum knee angle than heathy people (Subject 1, 2, and 3) when walking and running due to hurt or pathological knee constraints. Maximum knee angle b peaks is used as an indicator of knee flexion ability, which is evaluated by averaging peak values of b during the last 25 s walking at velocity 5 km/h and running at velocity 10 km/h in three repeated experiments of lower limb motion capture. Error bars here represent standard deviation (SD).

Supplementary Fig. 7| The complexity and metabolic penalty of lower limb motion capture by different methods. a-b,
The number of motion capture device and corresponding bilateral added mass by using conventional motion capture method (a) and our proposed method (b). c, The estimated metabolic penalty per kilogram of added mass for each segment, based on coefficients from the literature 25,58,59 . The setup for lower limb motion capture by using different methods and the corresponding metabolic penalty are shown in Supplementary Table 3 The conditioning circuit is composed of three constant temperature difference (CTD) circuits. Using the CTD circuit of the first pair of hot film (denoted as R !" ) and cold film (denoted as R #" ) as an example, R !" and R #" are connected into a Wheatstone bridge with two resistors (R $" and R %" ), the differential voltage is amplified and then fed back to the bridge to implement the CTD. The resistor R &%" is used to regulate the temperature difference between hot film and ambient. The bridge top voltage is further converted to digital signal by an analog/digital converter (ADC) for sampling. Outputs of the three CTD circuit are used to figure out the planar motion velocity vb.

S-12
Supplementary Fig. 11| Optimization of hidden neurons. The RMSE of θt and γt is used as the performance index of hidden neuron optimization. The smaller the RMSE is, the better performance the network has. The RMSE decreases with neuron number and gradually reaches to steady state. The RMSE keeps nearly constant with more than 30 hidden neurons. Network training is conducted repeatedly three times with the same hidden neurons. Error bars here represent standard deviation (SD).