[Documentation] [TitleIndex] [WordIndex

Overview

The package aims to give an optimal path for an agent to get its destination in an obstacle environment. The agent starts from a random position except the barrier area in the map, and then generates a feasible path to its destination while avoiding all the obstacles. Furthermore, it can support multiple agents to plan in the map.

Quick Start

Installation

cd catkin_ws/src
git clone http://github.com/micros-uav/micros_hopfield
cd ..
catkin_make

Running

-Setup rviz

rosrun rviz rviz

set the config parameter in rviz : File-open config-user.rviz -Run server node

source devel/setup.bash
rosrun micros_hopfield plan_server

Note:Run these commands under ~/catkin_ws directory for the terrain map loading.

-Run Client node Open a new consol

source devel/setup.bash
rosrun micros_hopfield plan_client i

where i is the client ID, i = 1,2,3...

Note:run different clients in independent terminal.

Result demo

alt text

* The green texture presents a terrain with different altitude

* The blue cylinder is an obstacle

* The red and blue lines are the pathes generated for two different agents

Note:The package is inspired by and adapted from [1]. Related details about neural network based path planning may also be found in [2], [3] and [4].

Reference

[1] Chonghong Fan, Youzhang Lu, Hong Liu, Shangteng Huang. Path planning for mobile robot based on neural networks. Computer Engineering and Application, 2004, 8: 86-89. (in Chinese)

[2] Roy Glasius, Andrzej Komoda, Stan C.A.M. Gielen. Neural network dynamics for path planning and obstacle avoidance. Neural Networks, 1995, 8(1):125-133.

[3] Simon X. Yang, Max Meng. An efficient neural network approach to dynamic robot motion planning. Neural Networks, 2000, 13(2):143-148.

[4] Simon X. Yang, Max Meng. Neural network approaches to dynamic collision-free trajectory generation. IEEE SMC Part B, 2001, 31(3): 302-318.


2024-07-20 13:23