Memory Centric Power Allocation for Multi-Agent Embodied Question Answering
技术动态
来源：arXiv cs.RO发布时间待核实
arXiv:2604.17810v2 Announce Type: replace Abstract: This paper considers multi-agent embodied question answering (MA-EQA), which enables robot teams to answer queries based on their long-horizon observations. In contrast to existing edge resource management methods that optimize sensing, communication, or computation performance metrics, MA-EQA focuses on the quality of aggregated memory.