Edge Intelligence Enabled Heterogeneous Multi-Robot Networks: Hybrid Framework, Communication Issues, and Potential Solutions
Silan Li, Shiyue He, Yu Zhang, Xiaotong Shi, Guojun He, Tao Jiang
Abstract
Heterogeneous multi-robot systems have been increasingly employed to collaboratively execute tasks in harsh and unknown environments. High-quality collaboration requires robots to efficiently communicate with each other and have ample intelligence to deal with uncertainties and emergencies. However, these demands are still difficult to be satisfied under existing single-mode networking architectures and onboard computing paradigms. Therefore, this article first develops a hybrid edge-aided framework for multi-robot collaboration, where the centralized and decentralized modes are elegantly combined to support a robust and efficient network, while an edge intelligence platform with powerful servers and base stations is employed to provide abundant intelligence. Under this framework, we discuss three critical communication issues that keep robots coordinated and guarantee efficient information sharing. The issues include network construction among heterogenous robots, bandwidth-efficient communication protocol design, and link transmission adaption. For each issue, potential solutions are provided and the challenges that need to be further explored are also discussed.