Intelligent robotic learning systems : a fuzzy-neural architecture and real-time implementations

The main focus of this thesis is on the development of intelligent robotic systems which are capable of learning a wide array of intelligent behaviors. The framework utilized in this development is that of nonclassical control architectures to realize real-time and primarily reactive system performance.

A Neural integrated Fuzzy conTroller (NiF-T), which integrates the fuzzy logic representation of human knowledge with the learning capability of neural networks, is developed for nonlinear dynamic control problems. The NiF-T architecture comprises three distinct parts: (1) Fuzzy logic Membership Functions (FMF), (2) Rule Neural Network (RNN), and (3) Output-Refinement Neural Net- work (ORNN). FMF are utilized to fuzzify input parameters. RNN interpolates the fuzzy rule set; after defuzzification, the output is used to train ORNN. The weights of the ORNN can be adjusted on-line to fine-tune the controller.

NiF-T can be applied for a wide range of sensor-driven robotics applications, which are characterized by high noise levels and nonlinear behavior, and where system models are unavailable or are unreliable. In this thesis, real-time imple- mentations of Ball Balancing Beam (BBB), autonomous mobile robot navigation, and mobile robot convoying behavior utilizing the NiF-T are realized. The NiF-T successfully balances balls with one-third of the required 27 rules. With learning capability, these systems approach their goals more frequently, in general. The robot, SMAR-T, successfully and reliably hugs the wall, locks onto the hall center, and convoys a leader at any specified distance.