Optimal Multilayered Motion Planning for Multiple Differential Drive Mobile Robots with Hierarchical Prioritization (OM-MP)

TL;DR

This paper introduces a hierarchical, multilayered motion planning framework for multiple differential drive robots, transforming MAPF into MATP, with global and local optimization strategies to improve efficiency, scalability, and task optimality in complex environments.

Contribution

It proposes a novel hierarchical optimization scheme combining global and local planners, employing AGS and CBFs, to enhance multi-robot trajectory planning performance over existing methods.

Findings

Outperforms state-of-the-art algorithms like CBS and ECBS in capacity and scalability.

Achieves faster solution times with improved trajectory feasibility.

Demonstrates superior task optimality through simulations and experiments.

Abstract

We present a novel framework for addressing the challenges of multi-Agent planning and formation control within intricate and dynamic environments. This framework transforms the Multi-Agent Path Finding (MAPF) problem into a Multi-Agent Trajectory Planning (MATP) problem. Unlike traditional MAPF solutions, our multilayer optimization scheme consists of a global planner optimization solver, which is dedicated to determining concise global paths for each individual robot, and a local planner with an embedded optimization solver aimed at ensuring the feasibility of local robot trajectories. By implementing a hierarchical prioritization strategy, we enhance robots' efficiency and approximate the global optimal solution. Specifically, within the global planner, we employ the Augmented Graph Search (AGS) algorithm, which significantly improves the speed of solutions. Meanwhile, within the local planner optimization solver, we utilize Control Barrier functions (CBFs) and introduced an oblique cylindrical obstacle bounding box based on the time axis for obstacle avoidance and construct a single-robot locally aware-communication circle to ensure the simplicity, speed, and accuracy of locally optimized solutions. Additionally, we integrate the weight and priority of path traces to prevent deadlocks in limiting scenarios. Compared to the other state-of-the-art methods, including CBS, ECBS and other derivative algorithms, our proposed method demonstrates superior performance in terms of capacity, flexible scalability and overall task optimality in theory, as validated through simulations and experiments.