This is project inspired from the Mathworks Matlab - Excellence in Innovation Repository : https://github.com/mathworks/MathWorks-Excellence-in-Innovation.git The details of the project is illustrated on the project 209 readme with A* Path planning and obstacle avoidance in a warehouse example.
Contributors: Shubhankar Kulkarni, Sanskruti Jadhav
Autonomous navigation on off-road environment such as agricultural lands is significant in the development of autonomous vehicles. Off-road terrain brings alot of complex problems such as uneven-ground with uncertain paths for an autonomous vehicle to traverse. In order to overcome these limitations a robust path planning algorithms and dynamic obstacle avoidance model must be deployed on the vehicles. Navigating complex terrain at speed and with minimal human supervision is a major challenge for developing such systems.
The project is an experiment of finding the shortest path to a given set of start and end points namely waypoints using RRT* Path planning algorithm in a rough terrain and then deploying these waypoints on the turtlebot robot. Mainly the tools used were MATLAB for generating the waypoints using a map of the given environment in pgm file, further more feeding this shortest path to a robot in Gazebo Ros simulation using the simulink model.
Following are the steps taken for running this project -
-
Below is a rough terrain environment suitable for the application - source: husky clearpath , husky playpen environment :
-
Launching the robot in this environment using the launch file which contains the world file location in the catkin workspace
-
Generate a pgm file through slam mapping of this environment, the pgm file of the chosen environment is already available by default in the husky clearpath installation.
- Once the pgm file is available, use a matlab mlx script RRT planner to generate a path using RRT* algorithm. One major challenge is to transform the coordinates of gazebo world with the exact coordinates from pgm file format.
- A roslocation.cpp script was used to establish a correlation with map coordinates on the pgm file and the gazebo simulated environment
- After creating the waypoints, the last point is goal of the trajectory which needs to be transformed back into gazebo coordinates. The mathworks path following with obstacle avoidance slx file takes this goal point as input to then generate the corresponding velocity upon connection with gazebo simulation. a rosinit command must be initialized in the command line of matlab to establish ros master connection with the turtlebot.
- The robot then traverses the given path and goal set in Gazebo environment after the simulink model runs successfully.
Autonomous Navigation using RRT* path planning algorithm was achieved using turtlebot3 burger model in Gazebo ROS simulated environment. The scope of this project was limited to small rough terrain environment and can be further extended with additional optimizations such as testing dynamic obstacle avoidance cases and other worst case scenarios. The outcome of this analysis was an introductory study of autonomous navigation of robots in a rough terrain concepts and its workings using MATLAB and ROS GAZEBO environments.
There are following improvement points which needs to considered for further optimizing this model:
- Using an actual off road terrain robot such as husky in an agricultural land which is already available in clearpath robotics git will prove to be more significant contribution for further study and analysis.
- Currently this project experiments using pure pursuit controller,optimizing the control of the robot to overcome the troughs of the environment by accounting uneven ground can be another approach that needs to be further investigated.
We are thankful to the support of mentors and advisors of this Project:
Dr. Roberto G. Valenti (Senior Research Scientist,MathWorks Excellence in Innovation Program Lead)
Dr. Bing Li (Professor,Clemson University International Center for Automotive Research (CU-ICAR))
