Indirect Force Measurement via Hamiltonian Monte Carlo-Based Probabilistic Model of FSR Sensor

Force-sensitive resistor (FSR) sensors have been playing a central role because of their endless applications in the robotics and healthcare sectors. They can be used for detecting abnormalities in respiration systems, contact force estimation, and human–machine interaction. Traditionally, the dynamic of the FSRs is modeled analytically to capture the change in resistance. However, these traditional approaches face a critical challenge because of the presence of uncertainty in measurement, violation of saturation limits, bias, and hysteresis. To address these issues, a Gaussian process regressor (GPR)-based probabilistic model via Hamiltonian Monte Carlo (HMC) called Gaussian process Monte Carlo (GPMC) is proposed for capturing the dynamics of the FSRs. In addition, the two models are developed for breaking the force measurement into the low and high regions with a range of ~0.2–10 and 10–98.8 N, respectively. To demonstrate the uncertainty in force measurement, a 95% prediction interval is drawn for the developed models. The experimental analysis advent that the presented model for the force measurement outperforms other state-of-the-art methods in terms of mean squared error (MSE), mean absolute error (MAE), and root mean square error (RMSE). It was noted that the presented model achieves an accuracy in the sequence of [MSE, MAE, RMSE] as [8.3063, 2.0763, 2.8820] and [0.0755, 0.2132, 0.2748] for high and low regions, respectively.