A Wearable-Sensor System with AI Technology for Real-Time Biomechanical Feedback Training in Hammer Throw

Abstract

Developing real-time biomechanical feedback systems for in-field applications will transfer human motor skills' learning/training from subjective (experience-based) to objective (science-based). The translation will greatly improve the efficiency of human motor skills' learning and training. Such a translation is especially indispensable for the hammer-throw training which still relies on coaches' experience/observation and has not seen a new world record since 1986. Therefore, we developed a wearable wireless sensor system combining with artificial intelligence for real-time biomechanical feedback training in hammer throw. A framework was devised for developing such practical wearable systems. A printed circuit board was designed to miniaturize the size of the wearable device, where an Arduino microcontroller, an XBee wireless communication module, an embedded load cell and two micro inertial measurement units (IMUs) could be inserted/connected onto the board. The load cell was for measuring the wire tension, while the two IMUs were for determining the vertical displacements of the wrists and the hip. After calibration, the device returned a mean relative error of 0.87% for the load cell and the accuracy of 6% for the IMUs. Further, two deep neural network models were built to estimate selected joint angles of upper and lower limbs related to limb coordination based on the IMUs' measurements. The estimation errors for both models were within an acceptable range, i.e., approximately ±12° and ±4°, respectively, demonstrating strong correlation existed between the limb coordination and the IMUs' measurements. The results of the current study suggest a remarkable novelty: the difficulty-to-measure human motor skills, especially in those sports involving high speed and complex motor skills, can be tracked by wearable sensors with neglect movement constraints to the athletes. Therefore, the application of artificial intelligence in a wearable system has shown great potential of establishing real-time biomechanical feedback training in various sports. To our best knowledge, this is the first practical research of combing wearables and machine learning to provide biomechanical feedback in hammer throw. Hopefully, more wearable biomechanical feedback systems integrating artificial intelligence would be developed in the future.

Keywords: deep learning; hammer throw; real-time biomechanical feedback; wearable devices; wireless sensor systems.

Figure 1

A reconfigurable WSN structure based on Arduino platform (The figure is adapted from the 1st author’s Ph.D. thesis [6]).

Figure 2

The general modus operandi/framework for developing wearable-sensor systems to help establish real-time biomechanical feedback training in field (The figure is adapted from the 1st author’s Ph.D. thesis [6]).

Figure 3

System architecture: the system consists of a sensor node and a receiver node. The sensor node device, including an Arduino microcontroller, an XBee wireless communication module, an on-board IMU (inserted onto the PCB inside the device), an attachable IMU (wrapped with red tape), an embedded micro load cell (along with a customized easy-to-release connector), etc., is worn on an athlete’s waist with a belt, which is used for collecting data in field tests and sending the data to the receiver node. The receiver node has another XBee module connected to an end device (e.g., laptop), which is used for receiving, processing, and analyzing data.

Figure 4

A wearable-sensor system including its hardware and software development is designed for real-time biomechanical feedback training in hammer throw. AI technology is also applied to estimate selected joint angles based on a local motion feature existing among upper and lower limbs during the hammer-throw movements. An optoelectronic motion capture system is used as a referencing system to provide reliable kinematic data.

Figure 5

The hardware design of the wearable device: (a) The layout of the PCB; (b) The logic diagram of the PCB design.

Figure 6

The deep neural network structure of the simplified model.

Figure 7

A typical result of the DNN modeling: (a) An example of training the complete model for 1000 epochs; (b) The simplified model’s prediction error; (c) The complete model’s prediction error; (d) The scatter plot of using the complete model to predict the left and right hip flexion/extension angles; (e) The scatter plot of using the complete model to predict the left and right knee flexion/extension angles; (f) The scatter plot of using the complete model to predict the left and right ankle flexion/extension angles.

Figure 7

A typical result of the DNN modeling: (a) An example of training the complete model for 1000 epochs; (b) The simplified model’s prediction error; (c) The complete model’s prediction error; (d) The scatter plot of using the complete model to predict the left and right hip flexion/extension angles; (e) The scatter plot of using the complete model to predict the left and right knee flexion/extension angles; (f) The scatter plot of using the complete model to predict the left and right ankle flexion/extension angles.

Figure 8

The two patterns of tension development during throwing: gradually increasing during body rotation (left in the figure, Peak 5–8) and suddenly increasing toward the end (right in the figure) (the figure is adapted from the 1st author’s PhD thesis [6]).