SELF-EVOLVING NEURO-SYMBOLIC ROBOTIC ARCHITECTURES FOR AUTONOMOUS ADAPTATION IN DYNAMIC HUMAN-CENTRIC ENVIRONMENTS: INTEGRATING CONTINUAL LEARNING AND SCENE REASONING FOR REAL-TIME COGNITIVE COLLABORATION IN SERVICE AND ASSISTIVE ROBOTS

Robotic systems operating in dynamic, human-centric environments require a high degree of adaptability, cognitive reasoning, and collaboration to be effective and safe. This paper proposes a self-evolving neuro-symbolic architecture for service and assistive robots, integrating continual learning with symbolic scene reasoning to achieve real-time cognitive collaboration. Unlike traditional robotic architectures, our model allows robots to adapt continually through interactions, refining both perceptual understanding and task-level planning. Drawing on recent advances in neurosymbolic AI, continual learning frameworks, and cognitive robotics, we demonstrate how the architecture supports autonomous adaptation in unstructured environments such as homes, hospitals, and care facilities. The system architecture is validated in simulation and physical deployment scenarios, showing improved performance in real-time decision-making, task success rate, and human-robot interaction fidelity.