Periodic Reporting for period 1 - AgileSim (Contributions of muscle mechanics and sensorimotor control to agile locomotion)
Periodo di rendicontazione: 2024-05-01 al 2026-04-30
Humans and animals navigate complex terrain seemingly effortless. This is in stark contrast with even the most performant robots, illustrating that walking over complex terrains is by no means trivial. Our neuromusculoskeletal system is equipped with mechanisms
that allow us to recover from unexpected perturbations. Two key mechanisms are muscle intrinsic mechanics and sensory-driven feedback control. Immediate changes in muscle force upon a perturbation allow the body to respond fast to sudden perturbations
through quick-acting muscle mechanical responses. Feedback responses, slower due to transmission delays, are also critical to stability as they are more flexible whereas muscle mechanical responses are determined by feedforward control and muscle
properties. We do not yet know how these pathways interact to help us maintain agility and robustness, in the presence of external perturbations, or in the case of sensory loss. I aim to gain novel insights into how muscle mechanics and sensory feedback allow agile
locomotion across conditions. Here, I aimed unravel fundamental principles governing relative contributions of these mechanisms, using a blended experimental- and computational approach. I already collected a unique experimental dataset that I combined with
physics-based simulations. I aimed to use novel approaches to predict locomotion patterns and feedforward and feedback control by optimizing performance criteria in presence of sensorimotor noise without relying on experimental data. Validation of simulation to
experimental data allows us to evaluate which performance criteria and muscle properties drive observed interactions between muscle mechanics tuned by feedforward and feedback control. As many neurological disorders impair stable locomotion,
fundamental insights obtained through my project have potential to inform treatments. Lastly, novel insights in locomotor neuromechanics inspires designs of legged robots and prosthetics to assist during locomotion
that allow us to recover from unexpected perturbations. Two key mechanisms are muscle intrinsic mechanics and sensory-driven feedback control. Immediate changes in muscle force upon a perturbation allow the body to respond fast to sudden perturbations
through quick-acting muscle mechanical responses. Feedback responses, slower due to transmission delays, are also critical to stability as they are more flexible whereas muscle mechanical responses are determined by feedforward control and muscle
properties. We do not yet know how these pathways interact to help us maintain agility and robustness, in the presence of external perturbations, or in the case of sensory loss. I aim to gain novel insights into how muscle mechanics and sensory feedback allow agile
locomotion across conditions. Here, I aimed unravel fundamental principles governing relative contributions of these mechanisms, using a blended experimental- and computational approach. I already collected a unique experimental dataset that I combined with
physics-based simulations. I aimed to use novel approaches to predict locomotion patterns and feedforward and feedback control by optimizing performance criteria in presence of sensorimotor noise without relying on experimental data. Validation of simulation to
experimental data allows us to evaluate which performance criteria and muscle properties drive observed interactions between muscle mechanics tuned by feedforward and feedback control. As many neurological disorders impair stable locomotion,
fundamental insights obtained through my project have potential to inform treatments. Lastly, novel insights in locomotor neuromechanics inspires designs of legged robots and prosthetics to assist during locomotion
During the 9-months I worked on this MSCA PF, I have remodeled the existing one-legged muscle model of the guinea fowl to a bipedal model, including all major muscles of the limb, resulting in 48 actuators per limb (conform workpackage 1). We decreased the number of via points and wrapping surfaces to increase computational efficiency, while remaining as true as possible to anatomical reality. We estimated tendon slack lengths by optimizing for efficiency (minimizing activation squared) across a range of motions. This was crucial as tendon slack lengths determine the muscle’s operating range but cannot be directly measured experimentally. We performed an inverse kinematic analysis followed by an inverse dynamic analysis based on marker trajectories and ground reaction forces. We estimated muscle-tendon lengths and moment arms from kinematics using the model. Inverse dynamic moments, muscle-tendon lengths and moment arms were input to dynamic optimization to estimate tendon slack lengths as well as the muscle states (activations, fiber lengths, and muscle forces). I have also initiated predictive simulations, but those are currently being validated. Both the deterministic as well as the predictive simulations will be continued while I am working in my new job (faculty position).
The existing model of the guinea fowl could not move (was static, not dynamic) and therefore could not be used for dynamic muscle analysis and truly be used as a tool for muscle analysis. Since muscles are the only actuators in our movements, yet experimental assessment of muscle forces and lengths during movement are limited to a few muscles due to feasibility of experiments. Computational musculoskeletal models and simulation offer a unique opportunity to estimate muscle states (i.e. length, force, activation) during movement. Here, we further developed an existing muscle model of the guinea fowl (Numida meleagris), a common animal model to study agile locomotion, to enable physiologically plausible simulations of the coordinated action of all muscles in dynamic movements. We used the model for inverse analyses of joint torques and muscle forces. and evaluated model outcomes based on experimental data. We are now using predictive simulation methods to explore the causal relationship between muscle properties and movement, especially during agile locomotor tasks.
The model is a result in itself and aside from the fact that we anticipate publishing the model this summer, other researchers already have indicated that they would like to work with our model, indicating the novelty and the necessity of the development of this model.
The model is a result in itself and aside from the fact that we anticipate publishing the model this summer, other researchers already have indicated that they would like to work with our model, indicating the novelty and the necessity of the development of this model.