Cohesive Velocity Transitions in Robotic Platoons Using Nesterov-type Accelerated Delayed Self Reinforcement (A-DSR)

Abstract

This article addresses the rapid transfer of velocity-transition information through a robotic platoon, without the need for centralized communication. Such rapid propagation of information through a network is important since it can lead to cohesive maneuvers, which in turn aids in performance, e.g., to maintain smaller headway between the agents in the platoon. A challenge, however, is that in absence of centralized communication, each agent uses neighbor-based update laws, which tend to be diffusive in nature and lead to loss of cohesion in the overall response. The main contribution of this work is to use a Nesterov-type, accelerated-gradient-based approach to improve both the convergence and cohesion of a platoon's response. An advantage is that the modified update law can be implemented using a delayed-self-reinforcement (DSR) method that does not require new information from the network nor any modification of the network connectivity. Additionally, the article presents conditions for stability as well as experimental results to illustrate the performance improvements.