| Distributed manipulation is the process by which many relatively simple actuators produce complex motion in a manipulated object. The particular distributed manipulation technology studied in this thesis is the actuator array. Actuator arrays are planar arrangements of two-degree-of-freedom actuators that cooperatively translate and orient objects. Actuator arrays have potential advantages over traditional manipulation techniques in that they can manipulate objects of various sizes and shapes without hardware changes, are fault-tolerant due to their hyper-redundant architecture, and can manipulate multiple objects simultaneously.; The primary contributions of this thesis are the derivation and stability analysis of closed-loop control laws for actuator arrays using no-slip contact between actuators and manipulated object, and the design, construction, and verification of a prototype actuator array system. Both regulating and trajectory tracking control are considered, and the stability analyses consider the possibility of stick-slip contact between the array and object. Actuator arrays are inherently hybrid systems, with forces changing discontinuously as the object changes sets of supporting actuators and as the contact state at each actuator switches between slipping and no-slip contact. This thesis extends the Multi-variable Circle Criterion, a standard tool in nonlinear system analysis, to analyze systems with such discontinuous forcing functions. It also uses direct solution of the Kalman-Yakubovich-Popov equations to derive error bounds for trajectory tracking controllers. The design, construction, and operation of the prototype actuator array system, the Advanced Distributed Manipulator System (ADMS), are discussed in detail. Experiments validate theoretical results and show controller performance with vision feedback. |
