Topic: Evolving Intelligence for Continual Learning and Adaptive Autonomous Systems

Autonomous intelligent systems are increasingly expected to operate in dynamic, uncertain, and open-ended environments where data distributions, tasks, and operating conditions continuously change. Unlike conventional AI systems that are typically trained offline under relatively fixed assumptions, next-generation autonomous systems need to learn continuously from streaming data, adapt to evolving environments, and acquire and update knowledge throughout their operational lifetime. These challenges have stimulated growing interest in evolving intelligence, continual learning, online adaptation, and lifelong autonomous learning.

Evolving intelligence provides a promising paradigm for developing intelligent systems that can dynamically update their structures, parameters, representations, and knowledge as new information becomes available. Recent advances in evolving neural networks, spiking neural networks, neuro-fuzzy systems, online learning, adaptive learning, and streaming data analytics have created new opportunities for building AI systems capable of learning and adapting with reduced reliance on repeated offline retraining. Such approaches are particularly relevant to real-world applications where data arrive continuously, environments are non-stationary, and computational resources may be constrained.

At the same time, the development of embodied intelligence and autonomous systems is creating new demands for continual and adaptive learning. Robots and other autonomous agents must integrate perception, learning, reasoning, decision-making, and control while interacting with physical environments. Enabling these systems to continuously acquire new skills, adapt to changing environments, recover from unexpected situations, and transfer previously acquired knowledge remains a fundamental challenge for intelligent robotics. The integration of evolving intelligence with embodied learning, sensorimotor intelligence, adaptive control, and autonomous decision-making may provide new foundations for more resilient and flexible autonomous systems.

This Special Topic aims to bring together recent advances in evolving intelligence, continual and lifelong learning, adaptive AI, and autonomous systems, with an emphasis on both fundamental methodologies and real-world applications. Contributions are particularly encouraged on learning from streaming and non-stationary data, evolving neural and computational intelligence, adaptive and self-evolving models, spiking and neuromorphic approaches for continual learning, explainable and trustworthy evolving intelligence, and their integration into embodied and autonomous systems. The topic welcomes theoretical studies, novel algorithms, system-level developments, and application-oriented research that advances the transition from static AI models toward continuously learning and adaptive autonomous intelligence.

Subtopics but not limited to:
● Evolving intelligence and evolving computational systems;
● Continual, lifelong, and online learning;
● Learning from streaming and non-stationary data;
● Adaptive and self-evolving neural networks;
● Spiking neural networks for continual and adaptive learning;
● Neuromorphic intelligence and efficient adaptive computing;
● Neuro-fuzzy and hybrid evolving intelligence;
● Knowledge acquisition, updating, and transfer in lifelong learning;
● Explainable and trustworthy evolving AI;
● Adaptive perception and sensorimotor learning;
● Continual learning for embodied intelligence;
● Adaptive planning, decision-making, and control;
● Skill learning and adaptation for autonomous robots;
● Multi-agent and distributed adaptive intelligence;
● Edge and resource-efficient continual learning;
● Real-world applications of evolving intelligence in autonomous systems.