path following with obstacle avoidance in simulink -凯发k8网页登录

introduction

this example uses a model that implements a path following controller with obstacle avoidance. the controller receives the robot pose and laser scan data from the simulated robot and sends velocity commands to drive the robot on a given path. you can adjust parameters while the model is running and observe the effect on the simulated robot.

start a robot simulator

start a simple matlab-based simulator:

this matlab-based simulator is a ros-based simulator for a differential-drive robot. the simulator receives and sends messages on the following topics:

replacing the matlab-based simulator with gazebo:

you can also use gazebo simulator with a simulated turtlebot®. see (ros toolbox) for instructions on setting up the gazebo environment. see (ros toolbox) for instructions on setting up network connection with gazebo. after starting the virtual machine, launch gazebo office world using the desktop shortcut. the simulated turtlebot in the gazebo simulator, receives velocity commands, as messages of type geometry_msgs/twist, on the /cmd_vel topic. you also need a localization algorithm to get robot position in the gazebo. see localize turtlebot using monte carlo localization algorithm for instructions on finding robot location in gazebo environment.

open existing model

this implements the path following with obstacle avoidance algorithm. the model is divided into four subsystems. the following sections explain each subsystem.

open_system('pathfollowingwithobstacleavoidanceexample');

process inputs

the 'inputs' subsystem processes all the inputs to the algorithm.

there are two subscribers to receive data from the simulator. the first subscriber receives messages sent on the /scan topic. the laser scan message is then processed to extract scan ranges and angles. the second subscriber receives messages sent on the /ground_truth_pose topic. the (x,y) location and yaw orientation of the robot is then extracted from the pose message.

the path is specified as a set of waypoints. this example uses a 3x2 constant input. you can specify any number of waypoints as an nx2 array. to change the size of the path at run-time, you can either use a variable sized signal or use a fixed size signal with nan padding. this example uses a fixed size input with nan padding for the waypoints that are unknown.

open_system('pathfollowingwithobstacleavoidanceexample/inputs','tab');

compute velocity and heading for path following

the 'compute velocity and heading for path following' subsystem computes the linear and angular velocity commands and the target moving direction using the pure pursuit block. the pure pursuit block is located in the mobile robot algorithms sub-library within the robotics system toolbox tab in the library browser. alternatively, you can type robotalgslib on the command-line to open the mobile robot algorithms sub-library.

you also need to stop the robot once it reaches a goal point. in this example, the goal is the last waypoint on the path. this subsystem also compares the current robot pose and the goal point to determine if the robot is close to the goal.

open_system('pathfollowingwithobstacleavoidanceexample/compute velocity and heading for path following','tab');

adjust velocities to avoid obstacles

the 'adjust velocities to avoid obstacles' subsystem computes adjustments to the linear and angular velocities computed by the path follower.

the vector field histogram block uses the laser range readings to check if the target direction computed using the pure pursuit block is obstacle-free or not based on the laser scan data. if there are obstacles along the target direction, the vector field histogram block computes a steering direction that is closest to the target direction and is obstacle-free. the vector field histogram block is also located in the mobile robot algorithms sub-library.

the steering direction is nan value when there are no obstacle-free directions in the sensor field of view. in this case, a recovery motion is required, where the robot turns on-the-spot until an obstacle-free direction is available.

based on the steering direction, this subsystem computes adjustments in linear and angular velocities.

open_system('pathfollowingwithobstacleavoidanceexample/adjust velocities to avoid obstacles','tab');

send velocity commands

the 'outputs' subsystem publishes the linear and angular velocities to drive the simulated robot. it adds the velocities computed using the pure pursuit path following algorithm with the adjustments computed using the vector field histogram obstacle avoidance algorithm. the final velocities are set on the geometry_msgs/twist message and published on the topic /mobile_base/commands/velocity.

this is an enabled subsystem which is triggered when new laser message is received. this means a velocity command is published only when a new sensor information is available. this prevents the robot from hitting the obstacles in case of delay in receiving sensor information.

open_system('pathfollowingwithobstacleavoidanceexample/outputs','tab');

note: to use gazebo simulator, select the /cmd_vel topic in the publish block.

configure and run the model

configure and run your model and observe the motion of the robot in the simulator.

see also