Self-Driving Car Using Image Processing

Autonomous cars represent a new mode of transportation using hardware and software to facilitate autonomous driving without human intervention. This is a design document for the creation of a low-cost autonomous vehicle prototype based on the Raspberry Pi 3 as the core processor. The system combines lane detection, traffic light identification, and traffic sign recognition into a single multi-threaded pipeline that facilitates real-time multi-task perception. Experimental testing demonstrates that the prototype attains a mean accuracy of 87%, a processing rate of 13–14 frames per second, and a mean decision time of approximately 105 ms in outdoor and indoor environments. In contrast to earlier Raspberry Pi-based prototypes that executed tasks sequentially or merely followed lanes, the system exhibits the novelty of stable multi-task perception with limited resources. The findings indicate the potential and limitations of embedded systems in autonomous navigation, with possible future applications in transportation, industrial automation, and surveillance.