MA-DV2F: A Multi-Agent Navigation Framework using Dynamic Velocity Vector Field

TL;DR

MA-DV2F is a multi-agent navigation framework that dynamically generates velocity vector fields for each vehicle, ensuring safe, efficient, and accurate navigation in complex environments with multiple agents.

Contribution

The paper introduces MA-DV2F, a novel framework that dynamically updates velocity vector fields for multi-agent navigation, improving safety and efficiency over existing methods.

Findings

Outperforms existing methods in safety and accuracy

Efficiently scales to large numbers of vehicles

Reduces collision risk through dynamic velocity updates

Abstract

In this paper we propose MA-DV2F: Multi-Agent Dynamic Velocity Vector Field. It is a framework for simultaneously controlling a group of vehicles in challenging environments. DV2F is generated for each vehicle independently and provides a map of reference orientation and speed that a vehicle must attain at any point on the navigation grid such that it safely reaches its target. The field is dynamically updated depending on the speed and proximity of the ego-vehicle to other agents. This dynamic adaptation of the velocity vector field allows prevention of imminent collisions. Experimental results show that MA-DV2F outperforms concurrent methods in terms of safety, computational efficiency and accuracy in reaching the target when scaling to a large number of vehicles. Project page for this work can be found here: https://yininghase.github.io/MA-DV2F/