Using Wearable Sensors to Estimate Vertical Ground Reaction Force Based on a Transformer

In this paper, we present a new method to estimate ground reaction forces (GRF) from wearable sensors for variety of real-world situations. We address the drawbacks using force plates with limited activity range and high cost in previous work. use transformer encoder as feature extractor extract temporal spatial features more efficiently. Using Mean Absolute Percentage Error (MAPE) evaluation criterion, experimental results show that average error predicted values improved by 32% compared RNN architecture 25% LSTM architecture. Finally, Gate_MSE solve problem large peak GRF prediction. Meanwhile, paper explores effect number or modes on