Reconfigurable Hardware Architecture for Vision-Based Driving Systems

Abstract

Vision-based driving systems are embedded systems that commonly have to consider real time constraints. Proposed systems use general purpose computers to execute designed algorithms based on their processing power and reduced designing time. These computers and other sensors are considerably power consuming and considerably increase the final cost of proposed solutions. Design of specific hardware systems that can be easily embed in autonomous vehicles without compromising power consumption, processing time and system final cost is an interesting solution. This work presents a hardware structure that can be used to increase the performance of autonomous vehicles visual navigation systems. Proposed structure was test with hardware version of image filters that are used in pre-processing steps of visual navigation algorithms. Results showed that proposed hardware achieved acceptable frame rate with a good image size compared to other proposed hardware systems.