Applying DCOP_MST to a Team of Mobile Robots with Directional Sensing Abilities
AAMAS '16: Proceedings of the 2016 International Conference on Autonomous Agents & Multiagent Systems(2016)
摘要
DCOP_MST is an extension of the DCOP framework for representing and solving dynamic multi-agent applications that include teams of mobile sensing agents. Local search algorithms, enhanced with exploration methods were recently found to produce high quality solutions for DCOP_MST in software simulations. Applying DCOP_MST to robots with directed sensors (e.g., cameras) requires addressing limitations, which were not part of the original design of the DCOP_MST model, e.g., limited angle of the field of vision, collisions between robots etc. In this paper we contribute to the ongoing effort of applying DCOPs to real world applications by addressing the challenges one faces when applying the DCOP_MST model to a team of mobile sensing robots with directed sensors. We integrate the required adjustments into a new model, DCOP_MSTR, which is the modified version of DCOP_MST for such a real world robot application with directed sensors. The proposed revised model was implemented and evaluated both in software simulations and on a team of robots carrying cameras. Our evaluation of existing algorithms revealed the need to combine actions that change the location of a robot with actions that change its sensing direction in order to achieve effective exploration when solving DCOP_MSTR
更多查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络