Editorial: Evolutionary computation-based machine learning and its applications for multi-robot systems
Year: 2023
Type: editorial
Abstract: Editorial: Evolutionary computation-based machine learning and its applications for multi-robot systemsWith the explosive of communication technology, the optimization and application of multi-robot systems becomes a challenging issue. From the perspective of algorithm design, classical integer programming, Markov decision and auction algorithms are difficult to solve complex problems and still require human expert knowledge. Evolutionary computation-based machine learning (ECML) methods have shown merits in dealing with complex optimization problems. Especially, ECML also has a strong search ability, which can decrease the computation cost of cluster nodes analytics to a large extent. Therefore, this issue aims to provide a platform for such a forum to focus our minds on the ECML methods and their application for multi-robot systems.Qi Chen and Yanan Qin et al, design a model predictive control (MPC) tracking algorithm based on quantum particle swarm optimization (QPSO) for cable-driven continuum robots (CDCRs) to realize effective trajectory tracking under constraints. In order to make kinematic analysis of a CDCR, the forward and inverse mapping among actuation space, joint space and work space is analyzed by using the piecewise constant curvature method and the homogeneous coordinate transformation. To improve the performance of conventional MPC for complex tracking tasks, QPSO is adopted in the rolling optimization of MPC for its global optimization performance, robustness and fast convergence. Finally, QPSO ensures an optimal feasible control value at every moment, which makes the robot track the target trajectory more stably and accurately.Jie Li and Shuang Gao et al, focus on unmanned aerial vehicles (UAVs) to provide emergency rescue services for the affected people. And, the study can improve the efficiency of emergency rescue. For the problem of scattered distribution of people in the affected areas, the researchers propose a global optimal target poi clustering model based on particle swarm optimization (PSO) to centralize the population. Then, the researchers design a green energy consumption calculation model to solve the problem of limited energy consumption of UAV. Considering the distance between each UAV and target-POI, the power of each UAV and the priority of target-POI, the appropriate UAV is selected to perform flight tasks on target-POI to ensure that all UAVs have lower energy consumption while completing flight tasks. Finally, the researchers design the Trans-UTPA multi-UAV cooperative path planning algorithm. The strategy learning of Trans-UTPA has no action space limitation and can be multi-task parallel, which improves the efficiency and generalization of sample processing.Kai Yu and Zixi Jia et al, propose a novel multimodal matching feedback self-tuning model, which is a novel neural network called Vital Information Matching Feedback Self-tuning Network (VIM-Net). The VIM-Net is mainly composed of two matching feedback modules, a visual matching feedback module (V-mat), a trajectory matching feedback module (T-mat). Specifically, V-mat matches the target information of visual recognition with the entity information extracted by the command; T-mat matches the serialized trajectory feature with the direction of movement of the command. Ablation experiments and comparative experiments are conducted on the proposed model using the Matterport3D simulator and the Room-to-Room (R2R) benchmark dataset, and the final navigation effect is shown in detail. The results prove that the model proposed in this paper is indeed effective on the task.Minhao Wan and Dehui Zhao et al, design a new camouflage pattern generation method. The Max Pooling theory combined with the discrete Laplacian differential operator is applied and the Max Pooling-Laplacian algorithm is proposed to compress and enhance the target background. In order to improve the accuracy and speed of camouflage pattern generation; combined with the K-means clustering principle, the background pixel primitives are used as the processing object, and the sample data are computed iteratively to obtain the camouflage pattern blended with the background. The results of the evaluation using color similarity and shape similarity show that the new camouflage pattern generation method achieves camouflage pattern design for different backgrounds and achieves good results.To sum up, this research topic has published 4 papers, which include listed findings from academicians and industry personnel in the related domain. These findings are helpful to make readers better to comprehend and learn the recent knowledge in the related domain.
Source: Frontiers in Neurorobotics
Author: Lianbo Ma
Institution: Northeastern University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $3,295
OpenAlex ID: W4366827992