A pipelined reconfigurable architecture for visual-based navigation

This paper is concerned with a pipelined reconfigurable architecture for implementing image processing algorithms useful for robotic navigation. The computational complexity and real-time application of these algorithms suggest a custom hardware approach, but the large quantity of different algorithms useful for different tasks suggest a software reconfigurable approach. A logic reconfigurable based architecture combines both properties: hardware speed and software flexibility. Moreover, the large quantity of data to be processed using conventional Cartesian image processing algorithms, combined with autonomous vehicle constraints (power consumption, size and weight) makes the use of reconfigurable existent machines impossible. The log-polar vision reduces the amount of data to be processed and simplifies the computation of several interesting algorithms, allowing a reconfigurable approach which is scalable, powerful and small enough for an autonomous robot