Fossilgaitsim: A simulation framework for gait of extinct quadruped animals

Abstract

Ancient animal locomotion has recently been investigated using biomechanicalmodelling and simulation studies, using techniques used mostly for humanmotion. We aim to investigate if these techniques can be used to study notonly the gait of extinct animals, but also their morphometry. In biomechanicalmodelling and simulation studies, gait simulations are created by solvingoptimal control problems. The optimal control problem is solved to find anoptimal gait cycle for a dynamics model that represents the animal. Theobjective of the optimization is to minimize energy expenditure, similar to theoptimization used for movement planning by the central nervous system. Thedynamics model and its parameters are derived from body fossils, but someof the resulting model parameter estimates are inaccurate, e.g., those relatedto mass and inertia. Therefore, we aim to use simulations to investigate thesemorphometric parameters, by finding the dynamics model and parameters forwhich the optimal control simulation best fits the trackways.Here, we introduce a MATLAB framework that can be used to create opticalcontrol simulations of extinct animals' gait. The framework can be used todesign different dynamics models that represent an extinct animal andinvestigate how well their respective simulations match trackways that arerelated to the animal. Different skeletal models can be created by varying themodel parameters and the degrees of freedom, e.g., by adding or removingsegments in the tail and the head, and by using different ranges of motion forthe degrees of freedom. We aim present the framework and show someexemplary simulations that were creating with this framework.So far, we have used the framework to create a simulation of a quadrupedanimal with 3 degrees of freedom in each leg, and a trunk segment. Theoptimal control problem was solved by minimizing the squared torque, whileconstraining the gait cycle to be periodic. We solved for a walking gait at 0.5m/s, using a standing simulation as initial guess. The resulting simulationdisplayed gait behaviour typical for quadruped walking. Next, we aim to showthe gait variations that can be obtained by varying the model’s morphology. Inthe future, we aim to use this framework to investigate footprint patterns suchas Ameghinichnus patagonicus, produced by an early mammaliaform ofsome sort, in Middle Jurassic outcrops in Patagonia, Argentina.