Biologically inspired computational structures and processes for autonomous agents and robots

Recent years have seen a proliferation of intelligent agent applications; from robots for space exploration to software agents for information filtering zmd electronic commerce on the Internet. Although the scope of these agent applications have blossomed tremendously since the advent of compact, affordable computing (and the recent emergence of the World Wide Web), the design of such agents for specific applications remains a daunting engineering problem. Rather than approach the design of artificial agents from a purely engineering standpoint, this dissertation views animals eis biological agents, and considers artificial analogs of biological structures and processes in the design of effective agent behaviors. In particular, it explores behaviors generated by artificieil neurzd structures appropriately shaped by the processes of evolution and spatiad learning. The first part of this dissertation deals with the evolution of au:tificial neural controllers for a box-pushing robot task. We show that evolution discovers high fitness structiures us­ ing little domeun-specific knowledge, even in feedback-impoverished environments. Through a careful analysis of the evolved designs we also show how evolution exploits the envirormientcd constrednts sind properties to produce designs of superior adaptive value. By modifying the task constraints in controlled ways, we also show the ability of evolution to quickly adapt to these changes and exploit them to obtciin significant performance gains. We also use evolution to design the sensory systems of the box-pushing robots, particularly the number, placement, and remges of their sensors. We find that evolution automaticeiUy disceirds unnecessary sen­ sors retaining only the ones that appear to significantly affect the performcmce of the robot. This optimization of design across multiple dimensions (performaince, number of sensors, size of neural controller, etc.) is implicitly achieved by the evolutioneiry algorithm without any external pressure (e.g., penality on the use of more sensors or neurocontroUer units). When used in the design of robots with limited battery capacities, evolution produces energy-efficient robot designs that use minimai numbers of components and yet perform reasonably well. The performance as well as the complexity of robot designs increase when the robots have access to a spatial learning mechanism that allows them to leam, remember, and navigate to power sources in the environment. The second part of this dissertation develops a computational characterization of the hippocampal formation which is known to play a significant role in animal spatial learning. The model is based on neuroscientific cind behavioral data, £ind learns place maps based on inter­ actions of sensory and dead-reckoning information streams. Using an estimation mechanism known as Kalman filtering, the model explicitly deals with uncertainties in the two information streams, edlowing the robot to effectively leam and localize even in the presence sensing and motion errors. Additionally, the model has mechanisms to handle perceptual alieising problems (where multiple places in the environment appear sensorily identical), incrementcdly learn and integrate local place maps, and leam and remember multiple goal locations in the environment. We show a number of properties of this spatial learning model including computationai replica­ tion of several behavioral experiments performed with rodents. Not only does this model ma.lfp significant contributions to robot localization, but also offers a number of predictions and sug­ gestions that can be validated (or refuted) through systematic netirobiological and behavioral experiments with animals.