| Active sensing can be defined as a sensing process that requires the expenditure of energy. Often this energy is used to move sensors relative to the environment. Importantly, active sensing is used in both biological and engineered systems, making it ripe ground for bio-inspired engineering. In this thesis I investigate active sensing in a model system (the rat vibrissal system) to uncover principles of biological active sensing that could apply - either directly or by analogy - to engineered systems. The central question is: how do biological systems choose movements that efficiently gather information during perception?;The rat vibrissal (whisker) system is a premier biological system for investigating active sensing. Using this system, I explore five principles of active sensing. First, the vibrissal system is a distributed sensor network composed of modular sensors, allowing for greater reliability and scalability compared to single sensors. Second, the whisker system is a scanning system and I demonstrate an adaptive scan-path mechanism based on velocity control. This mechanism enables sampling at different resolutions during a single sampling cycle. Third, the vibrissal system represents a hierarchical sensor in which movements at each stage get progressively faster and sensors become increasingly lower in mass. I present data showing how predictive, or look-ahead" spatial mechanisms enhance search behavior. Fourth, I developed instrumentation to measure the small, low-force whisker-object contacts and illuminate the rats non-uniform sampling strategy. This non-uniform sampling is characterized by periodic fixations at distinct spatial locations and jitter around these fixations. This multiple length-scale sampling strategy is analogous to optimal search strategies, which minimize search time. Finally, I show evidence that temporal jitter around fixations may prime neural circuits to certain incoming data, a potentially useful strategy for engineered systems with limited computational resources. These principles are discussed in the context of engineered systems, neural processing, sensorimotor control, and active sensing.;In summary, this thesis describes five primary contributions to the vibrissal field, which also inform the broader field of active sensing. These contributions further open the rat vibrissal system as a biological model of active sensing that can be used for bio-inspired engineering. |
