We created physiologically accurate models of internal muscular actuation in octopus arms. Introducing muscle stored energy function to bridge a gap between the fields of biomechanics and nonlinear elasticity theory (Cosserat rods), our modeling approach is first of its kind.
We used the energy shaping control methodology with our control-oriented muscle models to obtain continuum muscle controls to generate task-specific (e.g., reaching, grasping) three-dimensional motions. This procedure facilitates an understanding of the complex muscular arm dynamics in terms of its energy landscape, which is shaped by muscle actuations. Furthermore, using the energy shaping control strategy as a foundation, we explored the synergy between two levels of decision-making in octopuses -- in the brain and in the arm.
Among various stereo-typical arm movement patterns, bend propagation seems to be the most prominent one, where octopuses create a bend near the base of the arm and propagate the bend along the arm by traveling waves of muscle actuation. The questions of how (sensorimotor control) and why (optimality) octopuses engage in such maneuvers are not settled. We have used optimal control theory, reduced-order modeling, and sensory feedback control to shed light on bend propagation.
We proposed a novel algorithm (based on the geometry of a soft arm) that estimates all six modes of arm deformation from recorded video frames. The estimation formulation is "dual" to the energy-based control strategies. We successfully implemented the algorithm on octopus arms and on soft robotic arms in a laboratory setting.