Ransac Corridor Control

ROS package ransac_corridor_control repository. This package aims to solve the problem of intra-row autonomous navigation with an unisensory approach. The unique sensor device considered is a laser LIDAR. The package can also be used in any navigation problem well-characterized by a corridor with some sort of stochastic walls. The problem of leaving one corridor and entering in the parallel one is not covered here.

• ransac_corridor_control package in action!

Installation

Just clone the repository into a ROS workspace and compile it with catkin_make.

• There are no dependencies others than the ones commonly required by ROS

System Requirements

Some requirements are:

• GCC version 5.4.1 or later
• CMake version 2.8.3 or later
• Git
• python 2.7

Algorithms

The main algorithms used by the ransac_corridor_control are:

• The method Random Sample Consensus (RANSAC);
• An Extended Kalman Filter implementation;
• A PID control.

Usage

The Ros package ransac_corridor_control can easily be initialized by the command:

roslaunch ransac_corridor_control full.launch

The command above will launch three nodes, which are called:

• bisector;
• filter;
• control;
• to_twist.

bisector node

This node subscribes to the topics:

• \scan of message type sensor_msgs/LaserScan.

It is from the topic \scan that the data from the corridor environment provided by a laser scan sensor are made available for the method RANSAC, which basically estimates support lines of the corridor's wall and a bisector line.

This node publishes to the topics:

• \bisector_coeffs of messsage type ransac_corridor_control/LineCoeffs3Stamped (custom message);
• \bisector_pcl of message type sensor_msgs/PointCloud2;
• \points_ransac of message type sensor_msgs/PointCloud2.

It is from the topic \bisector_coeffs that the coefficients of the bisector line are made available to the ROS environment. The others two topic are just to data visualization. The topic bisector_plc contains generated points of the support and bisector lines and the topic points_ransac contains the points used by the RANSAC in the estimation process.

filter node

This node subscribes to the topics:

• \bisector_coeffs of messsage type ransac_corridor_control/LineCoeffs3Stamped (custom message)
• \car_command of message type ransac_corridor_control/CarCommandStamped (custom message).

With the coefficients of the last computed bisector line, of the last command sent to the car and with a simplified kinematic vehicle model, this node uses the EKF to estimate a bisector line less noise than the one computed by the node bisector.

This node publishes to the topics:

• \filtered_bisector_coeffs of messsage type ransac_corridor_control/LineCoeffs3Stamped (custom message)
• \filtered_bisector_pcl of message type sensor_msgs/PointCloud2

It is from the topic /filtered_bisector_coeffs that the filtered coefficients of the bisector line are made available to the ros enviroment. The topic \filtered_bisector_coeffs is just for data visualization.

control node

This node subscribes to the topics:

• \filtered_bisector_coeffs of messsage type ransac_corridor_control/LineCoeffs3Stamped (custom message)

With the bisector line coefficients from the topic \filtered_bisector_coeffs the control node computes an steering angle that orients the vehicle in the direction of reducing its distance from the filtered bisector line (lateral error).

This node publishes to the topics:

• \car_command of message type ransac_corridor_control/CarCommandStamped (custom message).

The control command to the low-level platform control implementations are sent by the topic \car_command. The message publised by the topic \car_command contains three fields which are:

• speedLeft (left wheel speed)
• speedRight (right wheel speed)
• steerAngle (steering angle)

to_twist node

This node publishes to the topics:

• \car_command of message type ransac_corridor_control/CarCommandStamped (custom message).

This node publishes to the topics:

• \cmd_vel of message type geometry_msgs/Twist

This node just convert from the message ransac_corridor_control/CarCommandStamped to the message geometry_msgs/Twist.

Parameters

All setting parameters are available to customization in the file yaml/config.yaml. Some parameters are:

• l: float parameter of the distance between car axle;
• laser/verbose: bool parameter for debugging purposes;
• laser/window_width: float parameter of the width of the window of interest;
• laser/window_length: float parameter of the length of the window of interest;
• laser/threshold: float parameter of the formation of the consensus group threshold;
• laser/iterations: int parameter of the number of iterations realized by the method RANSAC
• control/kpt: float parameter of the control proportional gain;
• control/kit: float parameter of the control integral gain;
• control/kvt: float parameter of the control velocity gain (like derivative);
• control/krt: float parameter of the control rescaling gain;
• control/lin_vel: float parameter of the target linear velocity;
• control/wait_time: float parameter of the waiting time to start to send the target velocity;
• kalman/verbose: bool parameter for debugging purposes;
• kalman/q00: float parameter of the element associated with the angle between the bisector line and the vehicle of the process covariance matrix Q;
• kalman/q11: float parameter of the element associated with the linear coefficient of the bisector line of the process covariance matrix Q;
• kalman/q22: float parameter of the element associated with the vehicle linear acceleration of the process covariance matrix Q;
• kalman/q33: float parameter of the element associated with the vehicle angular acceleration of the process covariance matrix Q;
• kalman/r00: float parameter of the elements associated with the angle between the bisector line and the vehicle of the measurement covariance matrix R;
• kalman/r11: float parameter of the elements associated with the linear coefficient of the bisector line and the vehicle of the measurement covariance matrix R;
• kalman/r22: float parameter of the elements associated with the vehicle linear velocity of the measurement covariance matrix R;
• kalman/r33: float parameter of the elements associated with the vehicle angular velocity of the measurement covariance matrix R.

Licensing

This software is being made available for research purposes only. See the LICENSE file in this directory for conditions of use.

Bugs & Feature Requests

Please report bugs and request features using the Issue Tracker.