Bonhyun Ku with Advisor A. Banerjee
The spine and surrounding muscles allow vertebrates to be flexible, dynamic, and balanced. Geared motors have been used to create distributed actuation to mimic the spine in robots. While minimizing weight, the gear train trades off speed for torque to meet various design criteria. An alternative gearless, bio-inspired electromechanical actuation was introduced to maximize the specific torque for a chosen angular flexibility in a module. In contrast to a geared electrical motor that trades off speed for torque, this actuator trades off range of motion for torque. Creating this trade-off makes the actuator attractive for a distributed actuation system, such as a spine, which inherently requires a limited range of motion as opposed to the full revolution.
Because modules are mechanically coupled in the spine, it is important to control spine dynamics, which are modeled as an open chain. Given the limited angular flexibility of a module, the mass matrix of the spine can be approximated as a constant matrix regardless of joint positions. The proposed distributed actuator’s torque characteristic is a highly nonlinear function of position and coil current. This nonlinearity is further amplified due to magnetic core saturation and the airgap fringing effect. A dynamic model of the spine and the actuator’s precise torque-to-current conversion are necessary to create the control framework. A torque controller is proposed, since a position controller with an inner-loop current controller is inherently unstable. This research is supported by the National Science Foundation.