利用大规模视觉导航机器人进行快递分拣，涉及调度系统的设计、视觉导航机器人设计、大规模机器人模拟软件设计、路径规划算法研究，涉及技术栈java、C++、c、spring、swing、netty、openCV、ardunio、调度、路径规划、嵌入式、PID控制。——Using large-scale visual navigation robot for express sorting involves the design of scheduling system, visual navigation robot, large-scale robot simulation software, path planning algorithm research, and technology stack Java, C + +, C, spring, swing, netty, opencv, ardunio, scheduling, path planning, embedded, PID control
知网论文地址 如果不方便下载，在issues留下邮箱，有空我会发给你
我国物流行业的分拣环节基本还处于人工分拣阶段，存在分拣效率低下、人工成本高、差错率高等问题。自动化物流分拣正朝着智能化方向发展，AGV（自动引导车）具有智能化程度高和灵活性高的特点，应用AGV来分拣快递包裹可以极大地提高分拣效率、降低人工成本和减小差错率。 本文研究的对象是一种多AGV物流分拣系统，在对多AGV物流分拣系统进行设计的基础上，对视觉导航及定位、多AGV路径规划两个关键技术进行了研究。论文主要研究工作包括三个部分：
本论文的研究对应工程主要分为三块，express（调度系统）、smartcar-vision（分拣机器人）、simulator（大规模机器人模拟系统）。
调度系统是整个多AGV物流分拣系统的大脑，负责信息统筹、调度和控制快递分拣机器人。该工程基于java、swing、spring boot、neety等技术，实现了论文中所述三种路径规划算法：改进A*算法、dijkstra算法、贪婪算法，以及多机器人之间防冲突算法，结合simulator工程可以模拟大规模机器人的调度。
大规模机器人调度模拟视频，点击观看
该工程是分拣机器人的大脑，实现了视觉导航、机器控制、调度，该工程基于C++、openCV、PID控制等技术，运行在Jeson Tk1嵌入式板（其实就是一个ubuntu系统），底层电机控制使用了Arduino。该工程依赖于硬件，有了这份代码也只有助于研究视觉导航的实现，很难实际运行起来。
一台分拣机器人造价不便宜，所以没有使用大规模真实机器人进行分拣实验。为了验证论文提出的算法的有效性，开发了本模拟软件，本工程模拟了机器人运行的物理特性，以及送料系统，使随机生成的快递包裹模拟放置在机器人托盘上，并完全适配于调度系统，调度系统没有特殊定制就可以像调度真实机器人一样调度模拟机器人。
smartcar-vision工程依赖于物理硬件，所以很难运行起来，里面代码可以用来学习机器视觉。这里我们主要讲“大规模机器人调度模拟”如何运行起来。
express和simulator都是基于spring开发的swing工程，之间采用netty通信，找到main函数就能运行起来了，先运行express，再运行simulaotr。两个工程运行都需要提供一分地图元数据Excel文件，仓库也为大家准备了一份Graph.xls，大家运行的时候需要在ComGraph文件中修改文件路径。
paper link If it’s not convenient to download, leave an email in issues, and I’ll send it to you when I have time
The sorting process of China’s logistics industry is still at the stage of manual sorting, which has the problems of low sorting efficiency, high labor cost and high error rate. Automated logistics sorting is developing in the direction of intellectualization. AGV (automatic guide car) has the characteristics of high intelligence and high flexibility. Using a large number of AGV to pick up the express package can greatly improve the sorting efficiency, reduce the labor cost and reduce the error rate. The object of this paper is a multi AGV logistics sorting system. On the basis of the design of the multi AGV logistics sorting system, two key technologies of visual navigation, positioning and multi AGV path planning are studied. The main research work of this paper consists of three parts:
The research of this paper is divided into three parts，express、smartcar-vision、simulator。
The scheduling system is the brain of the whole multi AGV logistics sorting system, which is responsible for information co-ordination, scheduling and control of the express sorting robot. Based on Java, swing, spring boot, neety and other technologies, this project implements three path planning algorithms: improved a * algorithm, Dijkstra algorithm, greedy algorithm, and conflict prevention algorithm among multiple robots. Combined with the simulator project, it can simulate the scheduling of large-scale robots。Large scale robot scheduling simulation
The project is the brain of sorting robot, which realizes visual navigation, machine control and scheduling. Based on C + +, opencv, PID control and other technologies, the project runs on the embedded board of Jason TK1 (in fact, it is a Ubuntu system). Arduino is used for the bottom motor control. The project depends on hardware, with this code only helps to study the implementation of visual navigation, it is difficult to actually run.
Large scale robot scheduling simulation
A sorting robot is not cheap, so there is no large-scale robot sorting experiment. In order to verify the effectiveness of the algorithm proposed in this paper, the simulation software is developed. This project simulates the physical characteristics of the robot operation, and fully adapts to the scheduling system, which has no special customization.
smartcar-vision Engineering relies on physical hardware, so it’s hard to run. The code inside can be used to learn machine vision. Here we mainly talk about how “large scale robot scheduling simulation” works.
Express and simulator are based on spring development of the swing project. Netty communication is used between them. Find the main function and run it. First run express, then run simulator. Both projects need to provide a map metadata excel file, and the warehouse has also prepared a graph.xls for you. When you run, you need to modify the file path in the comgraph file.