Plan Path for a Differential Drive Robot in Simulink

This example demonstrates how to execute an obstacle-free path between two locations on a given map in Simulink®. The path is generated using a probabilistic road map (PRM) planning algorithm (mobileRobotPRM). Control commands for navigating this path are generated using the Pure Pursuit controller block. A differential drive kinematic motion model simulates the robot motion based on those commands.

Load the Map and Simulink Model

Load the occupancy map, which defines the map limits and obstacles within the map. exampleMaps.mat contain multiple maps including simpleMap, which this example uses.

load exampleMaps.mat

Specify a start and end locaiton within the map.

startLoc = [5 5];
goalLoc = [20 20];

Model Overview

Open the Simulink model.

open_system('pathPlanningSimulinkModel.slx')

The model is composed of three primary parts:

  • Planning

  • Control

  • Plant Model

Planning

The Planner MATLAB® function block uses the mobileRobotPRM path planner and takes a start location, goal location, and map as inputs. The blocks outputs an array of wapoints that the robot follows. The planned waypoints are used downstream by the Pure Pursuit controller block.

Control

Pure Pursuit

The Pure Pursuit controller block generates the linear velocity and angular velocity commands based on the waypoints and the current pose of the robot.

Check if Goal is Reached

The Check Distance to Goal subsystem calculates the current distance to the goal and if it is within a threshold, the simulation stops.

Plant Model

The Differential Drive Kinematic Model block creates a vehicle model to simulate simplified vehicle kinematics. The block takes linear and angular velocities as command inputs from the Pure Pursuit controller block, and outputs the current position and velocity states.

Run the Model

simulation = sim('pathPlanningSimulinkModel.slx');

Visualize The Motion of Robot

After simulating the model, visualize the robot driving the obstacle-free path in the map.

map = binaryOccupancyMap(simpleMap);
robotPose = simulation.Pose;
thetaIdx = 3;

% Translation
xyz = robotPose;
xyz(:, thetaIdx) = 0;

% Rotation in XYZ euler angles
theta = robotPose(:,thetaIdx);
thetaEuler = zeros(size(robotPose, 1), 3 * size(theta, 2));
thetaEuler(:, end) = theta;

for k = 1:10:size(xyz, 1) % Plot the Robot's poses at every 10th step
    show(map)
    hold on;
    
    % Plot Start Location
    plotTransforms([startLoc, 0], eul2quat([0, 0, 0]))
    text(startLoc(1), startLoc(2), 2, 'Start');
    
    % Plot Goal Location
    plotTransforms([goalLoc, 0], eul2quat([0, 0, 0]))
    text(goalLoc(1), goalLoc(2), 2, 'Goal');
    
    % Plot Robot's XY locations
    plot(robotPose(:, 1), robotPose(:, 2), '-b')
    
    % Plot Robot's pose as it traverses the path
    quat = eul2quat(thetaEuler(k, :), 'xyz');
    plotTransforms(xyz(k,:), quat, 'MeshFilePath',...
        'groundvehicle.stl');
    light;
    drawnow;
    hold off;
end

© Copyright 2019 The MathWorks, Inc.