Intelligent Systems Context-Aware Adaptive Reinforcement Learning Approach for Autonomous Vehicles

Autonomous vehicles are replacing humans in the driver's seat. It requires intelligent systems. Therefore, it is crucial to have cognitive and context-aware Adaptive Reinforcement Learning for highly sensitive, safety, and security-driven autonomous systems architecture design. The system must know any context as part of the machine learning outcome. Therefore, it is important to bring the dynamic, safety-driven algorithms and their hardware components together to co-work based on context-aware adaptive concepts. The context part of the Reinforcement Learning system is used as input parameters to the Context Aware Machine Learning decision support mechanism that inference controls the operations in an autonomous system. Therefore, it is essential to have the system adaptive in nature and dynamic at runtime to update the learning-based output, and it can be further inference based on context. This paper explores the context specification allowing it to change dynamically as the hardware applies the algorithm. A context that changes dynamically thus provides a modified set of input parameters to the operations control mechanism so that the forward-looking behavior of the autonomous system can be driven adaptively. Our research results prove that the hardware dependency to adopt such a cognitive and context-aware adaptive machine learning algorithm for context-aware autonomous driving systems is important.