Special Session Ⅰ

Evolutionary Learning and Transfer Optimization for Intelligent Robotic Systems

Chair		Co-Chairs
Songbai Liu Shenzhen University, China		Qiuzhen Lin Shenzhen University, China		Kay Chen Tan The Hong Kong Polytechnic University, China

Keywords: Evolutionary Transfer Optimization, Learnable Evolutionary Algorithm, Complex Optimization Problems, Data-Driven Optimization, Adaptive Robotic Systems, Autonomous Intelligent Systems

Special Session Information: 

Intelligent robotic systems are inherently plagued by a spectrum of complex optimization problems. These include multi-objective motion planning balancing energy consumption, speed, and stability; expensive tuning of manipulator control parameters relying on physical simulations or real-world experiments; dynamic path planning adapting to changing environments or task goals in unstructured settings; multi-task learning that requires acquiring multiple related skills with knowledge sharing; and large-scale policy search in high-dimensional spaces of sensorimotor control. The complexity of these problems poses significant challenges to the computational efficiency, solution quality, and generalization capability of traditional optimization algorithms.

This special session focuses on leveraging evolutionary transfer optimization and learning-assisted evolutionary algorithms to address these challenges. We encourage research on transfer mechanisms that utilize prior knowledge, simulation data, or cross-task similarity to accelerate the optimization process; learning paradigms that employ meta-learning, surrogate models, and neural networks to guide the search direction and reduce evaluation cost; and the development of intelligent evolutionary algorithms capable of adapting to dynamic environments and automatically decomposing complex tasks. We invite high-quality contributions in algorithm design, theoretical analysis, and validation on real-world robotic platforms, aiming to collectively advance computational paradigms for solving core optimization problems in robotics.

Topics of interest include but are not limited to:

• Development and analysis of new evolutionary transfer optimization algorithms

• The development and analysis of new data-driven and learning-assisted evolutionary algorithms

• Fast Online Evolution and Meta-Learning Adaptation Mechanisms for Dynamic Environments

• Large-Scale Evolutionary Algorithms and Their Distributed Computing Frameworks for High-Dimensional State-Action Space Optimization

• Knowledge Representation, Transfer, and Evolution in Robotic Multi-task and Lifelong Learning

• Applications of evolutionary algorithms in machine learning and deep learning

• Real-world optimization problems solved using evolutionary algorithms

• Theoretical studies on the convergence and performance of evolutionary algorithms

• Evolutionary optimization in robotics and autonomous systems