当前位置: X-MOL 学术Simul. Model. Pract. Theory › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Distributed evolutionary learning control for mobile robot navigation based on virtual and physical agents
Simulation Modelling Practice and Theory ( IF 4.2 ) Pub Date : 2019-12-19 , DOI: 10.1016/j.simpat.2019.102058
Hiram Ponce , Ernesto Moya-Albor , Lourdes Martínez-Villaseñor , Jorge Brieva

This paper presents a distributed evolutionary learning control based on social wound treatment for mobile robot navigation using an integrated multi-robot system comprised of simulated and physical robots. To do so, this work proposes an extension of the population-based metaheuristic wound treatment optimization (WTO) method into a distributed scheme. In addition, this distributed WTO method is implemented on the multi-robot system allowing them to experience the environment in their own and communicate their findings, resulting in an emergence intelligence. We implemented our proposal using the combination of five simulated robots with one physical robot for tuning a navigation controller to move freely in a workspace. Results showed that the solution found by this multi-robot system aims using the output controller in the physical robot for successfully achieving the goal to move the robot around a U-maze, without applying any transfer learning approach. We consider this proposal useful in evolutionary robotics, and of great importance to decrease the gap related to transfer knowledge in robotics from simulation to reality.



中文翻译:

基于虚拟和物理代理的移动机器人导航分布式进化学习控制

本文提出了一种基于社交伤口处理的分布式进化学习控制,该方法使用由模拟和物理机器人组成的集成多机器人系统对移动机器人进行导航。为此,这项工作提出将基于人群的元启发式伤口处理优化(WTO)方法扩展到分布式方案中。此外,这种分布式WTO方法是在多机器人系统上实现的,使他们可以亲身体验环境并交流发现结果,从而获得了应急情报。我们通过结合使用五个模拟机器人和一个物理机器人来实现我们的建议,以调整导航控制器以在工作区中自由移动。结果表明,此多机器人系统找到的解决方案旨在通过使用物理机器人中的输出控制器来成功实现绕U型迷宫移动机器人的目标,而无需应用任何转移学习方法。我们认为该建议对进化型机器人学很有用,并且对于减小与将机器人学知识从仿真转移到现实有关的差距非常重要。

更新日期:2019-12-19
down
wechat
bug