Fatigue-PINN: Physics-Informed Fatigue-Driven Motion Modulation and Synthesis

TL;DR

Fatigue-PINN tackles the lack of fatigue-aware data-driven motion synthesis by introducing an end-to-end physics-informed architecture that operates in torque space. It encodes OpenSim-derived joint kinematics into torques via an Inverse Dynamics BiLSTM, modulates fatigue with a PINN adaptation of the Three-Compartment Controller (3CC-$\lambda$), and decodes fatigued torques back into joint angles through a Forward Dynamics BiLSTM, enabling fatigue-consistent animation without fatigued motion capture data. The key contributions include joint-specific fatigue configurations, physics-informed losses to constrain learning, and a semi-dynamics encoder–decoder pipeline compatible with angle-based animation systems. The framework demonstrates realistic fatigue effects in open-type motions and high predictive accuracy, highlighting practical impact for animation realism, ergonomic design, and fatigue-aware biomechanical analysis, while outlining limitations for contact-rich tasks and suggesting future work with contact modeling and diffusion-based enhancements.

Abstract

Fatigue modeling is essential for motion synthesis tasks to model human motions under fatigued conditions and biomechanical engineering applications, such as investigating the variations in movement patterns and posture due to fatigue, defining injury risk mitigation and prevention strategies, formulating fatigue minimization schemes and creating improved ergonomic designs. Nevertheless, employing data-driven methods for synthesizing the impact of fatigue on motion, receives little to no attention in the literature. In this work, we present Fatigue-PINN, a deep learning framework based on Physics-Informed Neural Networks, for modeling fatigued human movements, while providing joint-specific fatigue configurations for adaptation and mitigation of motion artifacts on a joint level, resulting in more realistic animations. To account for muscle fatigue, we simulate the fatigue-induced fluctuations in the maximum exerted joint torques by leveraging a PINN adaptation of the Three-Compartment Controller model to exploit physics-domain knowledge for improving accuracy. This model also introduces parametric motion alignment with respect to joint-specific fatigue, hence avoiding sharp frame transitions. Our results indicate that Fatigue-PINN accurately simulates the effects of externally perceived fatigue on open-type human movements being consistent with findings from real-world experimental fatigue studies. Since fatigue is incorporated in torque space, Fatigue-PINN provides an end-to-end encoder-decoder-like architecture, to ensure transforming joint angles to joint torques and vice-versa, thus, being compatible with motion synthesis frameworks operating on joint angles.

Figures (9)

• Figure 1: General overview of our Fatigue-PINN framework.
• Figure 2: 3CC. Figure reproduced from Frey-Law2012
• Figure 3: BiLSTM Inverse/Forward Dynamics Model. The feedback loop of each BiLSTM layer is marked with green arrows, whereas its backward sequence is indicated with red arrows.
• Figure 4: The impact of different levels of fatigue on punching (Frame 290, i.e. the moment of hit), throwing (Frame 240, at throw), and waving (Frame 100) motions. The levels of fatigue are defined w.r.t Residual Capacity as arises from Equation (\ref{['eq:3CC-lambda']}). For instance, $30\%$ fatigue $\rightarrow RC(t) = 70\%$, etc.
• Figure 5: Effects of fatigue on punching motion. At frame $100$ the character is at guard stance, and frame $290$ exhibits the moment of impact of the left hand. The discrepancies in hand positioning are marked with yellow dotted lines, while the distances between fatigued and non-fatigued hand positions are marked with red arrows.