A5080060340- Advanced Multi-Objective Optimization Algorithms
- Metaheuristic Optimization Algorithms Research
- Evolutionary Algorithms and Applications
- Robot Manipulation and Learning
- Manufacturing Process and Optimization
- Optimal Experimental Design Methods
- Scheduling and Optimization Algorithms
- Prosthetics and Rehabilitation Robotics
- Muscle activation and electromyography studies
- Stroke Rehabilitation and Recovery
- Teleoperation and Haptic Systems
- Optimization and Search Problems
- Advanced machining processes and optimization
- Assembly Line Balancing Optimization
- Machine Learning and Data Classification
- Reinforcement Learning in Robotics
South China University of Technology
2021-2024
Download This Paper Open PDF in Browser Add to My Library Share: Permalink Using these links will ensure access this page indefinitely Copy URL DOI
Download This Paper Open PDF in Browser Add to My Library Share: Permalink Using these links will ensure access this page indefinitely Copy URL DOI
Introduction Robotic assembly tasks require precise manipulation and coordination, often necessitating advanced learning techniques to achieve efficient effective performance. While residual reinforcement with a base policy has shown promise in this domain, existing approaches rely on hand-designed full-state features policies or extensive demonstrations, limiting their applicability semi-structured environments. Methods In study, we propose an innovative Object-Embodiment-Centric Imitation...
Surrogate-assisted evolutionary algorithms have made significant advancements in addressing expensive optimization problems. However, existing research primarily focuses on low-dimensional multi/many-objective optimization. This paper introduces a surrogate-assisted algorithm with feature extraction framework to address high-dimensional approach includes two key insights. Firstly, enhance the exploration capabilities, proposed employs sub-region search strategy define promising decision...
Surrogate-assisted multiobjective evolutionary algorithms (SA-MOEAs) have made significant progress in solving expensive multi-objective/many-objective optimization problems. However, most of them work well low-dimensional settings but often struggle high-dimensional ones. The main reason lies the fact that some techniques used SA-MOEAs, such as Kriging model, are not applicable for exploring search space. As a result, this research investigates frameworks incorporating dimensionality...
Assembly is an essential part of complex product manufacturing. Due to the complexity assembly operations and rapidly changing market demands, it a labor-intensive industry with insufficient automation improvement. Force control methods are developed allow industrial robots do Contact-Rich manipulation force sensors fixed at end-effect. Deep Reinforcement Learning (DRL) provides method learn skills by trial error. Nevertheless, performance framework combining DRL sensitive hyper-parameters,...
This paper proposes a framework for learning and control in precise peg-in-hole tasks using commercial off-the-shelf (COTS) robot system. It addresses the limitations of relying solely on model-based or learning-based methods, which are difficult to quickly adapt new tasks, by introducing partial model uncertainty handling. Firstly, this presents learn residual fine policy with coarse operation reward shaping reinforcement enhance interpretability incorporating expert knowledge. Secondly,...