| An immune cell's ability to interpret small chemical gradients in its environment and migrate to their sources is central to its ability to clear infection and cellular damage. The signals the cell must sense are often weak and noisy and are present in the spatially complex environment of dense tissue. Thus, the sensor/actuator system that the cell must implement to sense these signals and robustly migrate through this milieu to its target is quite sophisticated. Indeed; the molecular basis for this behavior, chemotaxis, involves possible hundreds of proteins and small molecules and thousands of reactions. While each molecule and interaction probably plays some role in optimizing or error-proofing chemotaxis, the fundamental principles of the control of the cells guidance system is likely to be much simpler. Embedded in this morass of biochemistry is some core set of signal paths and feedback loops that implement the core control law for the cell.; Conceptually, there are three operations the cell must accomplish, signal detection and localization, signal amplification/noise filtering, and motion towards the signal. This thesis, first, quantitatively characterizes the first two of these processes using theory and experiment and, second, shows how the control laws used by immune cells may be abstracted and used to construct a, robust, collaborative control system for robotic signal tracking. The former project discovered that cell membrane/receptor distribution plays a significant role in signal detection and amplification; a feature missed by previous analyses, and that the biochemical pathway is likely implementing a linear amplifier rather than a nonlinear one because cell shape played little role in signal amplification. Along the way, novel techniques for quantitative cellular image analysis from multiple fluorescent channels were developed and measurements of the temporal response of the amplifier system were made. The results of the latter project show that an implementation of the fundamental control principles of an immune cell as an algorithm for controlling an autonomous robots solves an optimal sensor fusion problem, has an excellent robustness/sensitivity performance, and has a computationally fast implementation.; Overall, this thesis represents a tight coupling of engineering analysis to quantitatively understand cellular control processes and to translation control laws arrived at through millions of years of evolution into useful algorithms for the control of artificial systems. |
