TY - GEN
T1 - A Predictive Model of Human Movements based on Model Predictive Control for Human-Robot Interaction
AU - Gillam, Aeden G.
AU - Razavian, Reza Sharif
N1 - Publisher Copyright:
© 2024 IEEE.
PY - 2024
Y1 - 2024
N2 - Predicting human movements is vital to safely control robots that physically interact with humans. However, predictive neuromuscular models that are fast enough for realtime control applications have proven challenging, due to the complexity of the neural and musculoskeletal systems. Nonlinear optimization-based prediction of movements in a musculoskeletal model is prohibitively slow. On the other hand, highly simplified models based on linear control theory cannot handle complexities of the human musculoskeletal system. Model Predictive Control (MPC) can potentially fill the gap between these two modeling extremes, by taking into account physiological nonlinearities, constraints, and redundancies while keeping computations fast through its receding horizon formulation. This study presents a new predictive model for the human movements based on MPC, which can control activity of four muscles acting on an inertia in a two-dimensional space to generate movements. The MPC results are compared to that of the prominent human motor control model in the neuroscience literature, which is based on linear quadratic regulator. The predicted movements are similar between the two controllers and are qualitatively similar to human behavior. MPC achieves these results while satisfying physiological constraints on muscle activities and ranges of motion - features that are not present in the existing models. These results demonstrate promise and potential for MPC controllers to accurately predict human neuro-muscular activities for the next generation controllers for human-robot interaction.
AB - Predicting human movements is vital to safely control robots that physically interact with humans. However, predictive neuromuscular models that are fast enough for realtime control applications have proven challenging, due to the complexity of the neural and musculoskeletal systems. Nonlinear optimization-based prediction of movements in a musculoskeletal model is prohibitively slow. On the other hand, highly simplified models based on linear control theory cannot handle complexities of the human musculoskeletal system. Model Predictive Control (MPC) can potentially fill the gap between these two modeling extremes, by taking into account physiological nonlinearities, constraints, and redundancies while keeping computations fast through its receding horizon formulation. This study presents a new predictive model for the human movements based on MPC, which can control activity of four muscles acting on an inertia in a two-dimensional space to generate movements. The MPC results are compared to that of the prominent human motor control model in the neuroscience literature, which is based on linear quadratic regulator. The predicted movements are similar between the two controllers and are qualitatively similar to human behavior. MPC achieves these results while satisfying physiological constraints on muscle activities and ranges of motion - features that are not present in the existing models. These results demonstrate promise and potential for MPC controllers to accurately predict human neuro-muscular activities for the next generation controllers for human-robot interaction.
UR - http://www.scopus.com/inward/record.url?scp=85203250941&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85203250941&partnerID=8YFLogxK
U2 - 10.1109/AIM55361.2024.10637233
DO - 10.1109/AIM55361.2024.10637233
M3 - Conference contribution
AN - SCOPUS:85203250941
T3 - IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM
SP - 82
EP - 87
BT - 2024 IEEE International Conference on Advanced Intelligent Mechatronics, AIM 2024
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2024 IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM 2024
Y2 - 15 July 2024 through 19 July 2024
ER -