Chapter 6, Design and Development of AI Neuromorphic to Control the Autonomous Driving System

One of the most defining features of AI is autonomous driving. By using energy-efficient computing frameworks based on spiking neural networks, neuromorphic (brain-inspired) control has the potential to make a substantial contribution to autonomous behaviour. Neuromorphic versions of four well-known independent driving controllers—Stanley, Pursuit, PID, and MPC—were investigated in this study utilizing a physics-aware simulation framework. The models’ performance was compared with that of traditional CPU-based implementations and conducted thorough evaluations using a wide range of intrinsic characteristics. Provide instructions for constructing control-oriented neuromorphic structures and highlight the significance of the tuning parameters and neural resources that make them tick. The findings indicate that a small number of neurons—100 to 1,000—would be sufficient for the majority of models to achieve peak performance. As was proposed with the MPC controller, they similarly emphasize the implication of hybrid conventional and neuromorphic systems. In this case, the MPC controller, also shows how important it is to combine conventional and neuromorphic architectures. Mostly at higher speeds, where they incline to deteriorate quicker than in traditional enterprises, this research also shows the limits of neuromorphic implementations.