Distributed Non-Convex Model Predictive Control for Non-Cooperative Collision Avoidance of Networked Differential Drive Mobile Robots

Run Mao,Huafeng Dai

doi:10.1109/access.2021.3134696

Run Mao, Huafeng Dai

Open Access

PDF Available

https://doi.org/10.1109/access.2021.3134696

Copy DOI

Export

Save

Cite

Journal: IEEE Access	Publication Date: Jan 1, 2022
Citations: 5	License type: CC BY 4.0

Affiliation: Chengdu University

Abstract
Highlights/Summary
Full-Text PDF
Similar Papers

Abstract

Listen

Distributed Non-Convex Model Predictive Control for Non-Cooperative Collision Avoidance of Networked Differential Drive Mobile Robots

Highlights

Collision avoidance is the central problem to many applications, such as autonomous navigation through human crows [1], crowd simulation for computer graphics and VR [2], multiple AGVs in an automated warehouse [3]
This strategy is an extension of Priority-Based Non-Cooperative Distributed Model Predictive Control (NCDMPC) (PB-NC-Distributed MPC (DMPC)) [31], it can deal with the problem of the loss of the prediction consistency property and reduce the computation time as well
Using PB-Non-Cooperative DMPC (NC-DMPC) in Networked Control Systems (NCS) with timevariant coupling topology, the time-delayed predictions used by robot v for its higher priority neighbors V( ) differ from the predictions computed by robots V( ), which results in a loss of the prediction consistency property

Summary

INTRODUCTION

Collision avoidance is the central problem to many applications, such as autonomous navigation through human crows [1], crowd simulation for computer graphics and VR [2], multiple AGVs in an automated warehouse [3]. We present a novel multi-DDMRs collision avoidance strategy that significantly outperforms existing methods This strategy is an extension of Priority-Based NCDMPC (PB-NC-DMPC) [31], it can deal with the problem of the loss of the prediction consistency property and reduce the computation time as well. A “safe” priority assignment should be proposed: 1) all robots can find a trajectory to simultaneously avoid the obstacle and the robots, 2) all robots do not need to add any sensors, 3) computation time should be maintained Based on these requirements, the concept of urgency is introduced. Using PB-NC-DMPC in NCS with timevariant coupling topology, the time-delayed predictions used by robot v for its higher priority neighbors V( ) differ from the predictions computed by robots V( ), which results in a loss of the prediction consistency property.

13: Sorting Pu from small to large to generate the corresponding priority

OBJECTIVE FUNCTION

NUMERICAL RESULTS

CONCLUSIONS