Hammod, Maen, Chaomin Luo, Mohan Krishnan, and Mark Paulik
With the rapid development of robotic systems and their related applications, the required computational power for these systems has become increasingly difficult to achieve. Robots are expected to interface with a wide range of sensors such as cameras, LADAR, GPS…etc., process the acquired data, drive autonomously using a complex navigation and goal selections algorithms, and finally perform their planned task.
This research investigates different parallel computing architectures to achieve the required computational power in a robotic system at a minimal cost. The new Intel multi-core processor architecture along with parallel computing software is used to distribute the work on the different processor cores and achieve a high computational speed-up factor. In addition, a multi-computer cluster architecture is used to fulfill all the needed tasks and at a minimal cluster communication overhead. This research demonstrates the computational enhancements for image processing test benches adapted for a parallel computing architecture.