SUMMARY

A key step in understanding natural motion is producing physical, well understood test platforms with a dynamic models closely resembling biological muscle. These test beds can then serve as a basis for better understanding the interrelated nature of the nervous system and muscles, for kinematics/dynamics experiments to understand balance and synergies, and for building full-strength, safe muscles for prosthetics, rehabilitation, human force amplification, and humanoid robotics. This work presents biologically-inspired cellular actuator arrays consisting of many small flexible actuation cells interconnected using varied topologies to yield muscle-like properties. The fingerprint method is introduced to compactly represent the complex networks and an analysis method based on graph theoretic modeling provides a systematic approach to generating dynamic equations of motion and static properties directly from the fingerprint. Given two simple requirements, the method can represent the array dynamics for any internal cell dynamics, separating the complexity of the cell from the complexity of the array. Forward-loop control methods, such as stochastic broadcast control, are explored to distribute control signals across the redundant actuators, and physical actuator array responses are used to validate theoretic results. Finally, existing linear actuation technologies are fit to the internal cell requirements to aid in real world implementation.