Indoor Mobile Robot Mapping and Navigation

中文说明

An indoor mobile robot simulation project integrating MATLAB, Simulink, ROS 1 Noetic, Gazebo 11, and TurtleBot3 Waffle. It implements an end-to-end workflow from sensor interaction and 2-D lidar SLAM to reactive obstacle avoidance and custom A* multi-goal patrol. The repository includes the control and navigation code, a custom Gazebo world, reproducible map and goal data, and simulation results.

Scope note: the navigation algorithms and controllers are implemented in MATLAB; the ROS Navigation Stack, AMCL, and gmapping are not used.

Highlights

• Turtlesim motion control with both a MATLAB script and a Simulink model.
• A custom 8 m x 6 m Gazebo indoor world with a TurtleBot3 Waffle launcher.
• Keyboard teleoperation with live LaserScan and camera visualization in MATLAB.
• Online 2-D lidar SLAM using MATLAB lidarSLAM.
• Interactive waypoint selection and sequential patrol with reactive obstacle avoidance.
• A custom 8-connected A* planner with Euclidean heuristic, obstacle inflation, waypoint tracking, and multi-goal patrol.
• Best-effort zero-velocity cleanup guards for interrupted controller runs.

System architecture

Gazebo 11 / TurtleBot3 Waffle (Ubuntu, ROS 1 Noetic)
        |  /scan, /odom, camera topics
        v
MATLAB ROS Toolbox (Windows host)
        |-- teleoperation and sensor display
        |-- lidarSLAM -> occupancyMap
        |-- interactive goal selection
        `-- reactive patrol / custom A* patrol -> /cmd_vel

Repository layout

matlab/
  task1_turtlesim/             MATLAB turtlesim controller
  task2_teleoperation/         Keyboard control and sensor dashboard
  task3_slam_navigation/       SLAM, waypoint patrol, and A* pipeline
  ros_network_config.m         Environment-based ROS network configuration
simulink/task1_turtlesim/      Final two-lap Simulink model
ros1_ws/src/my_robot_sim/      Catkin package with launch and world files
assets/images/                 Result figures
assets/demos/                  Demonstration recordings
tests/                         MATLAB tests for the standalone A* planner

Tested project configuration

• MATLAB R2024b
• Simulink
• ROS Toolbox
• Robotics System Toolbox
• Navigation Toolbox
• Ubuntu 20.04.6 LTS with ROS 1 Noetic and Gazebo 11
• TurtleBot3 Waffle packages
• VMware NAT networking between the Windows host and Ubuntu guest

The ROS environment can be reproduced on Ubuntu 20.04 / ROS 1 Noetic when the dependencies below are installed.

Reproduction

1. Build the ROS package in Ubuntu

Copy ros1_ws/src/my_robot_sim into a Catkin workspace, then run:

cd ~/catkin_ws
catkin_make
source devel/setup.bash
export TURTLEBOT3_MODEL=waffle
roslaunch my_robot_sim start_world.launch

Required ROS packages include gazebo_ros, xacro, and turtlebot3_description. The launch file loads worlds/my_world.world, publishes the robot description, and spawns a TurtleBot3 Waffle at the world origin.

2. Configure MATLAB-to-ROS networking

In MATLAB, from the repository root:

addpath(genpath(fullfile(pwd, "matlab")));
setenv("ROS_MASTER_URI", "http://<ubuntu-vm-ip>:11311");
setenv("ROS_IP", "<windows-vmnet-ip>");

The two IP addresses must belong to the same VMware NAT subnet; network addresses are configured through environment variables rather than fixed in the source code.

3. Run the tasks

Task 1 — turtlesim script:

1. Start roscore and turtlesim_node in Ubuntu.
2. Run matlab/task1_turtlesim/turtlebot.m in MATLAB.
3. For Simulink, run simulink/task1_turtlesim/prepare.m, then open turtle_circle_2laps.slx.

Task 2 — teleoperation and live sensors:

1. Launch the custom world as shown above.
2. Run control_turtlebot from matlab/task2_teleoperation.
3. Use W/S/A/D to move, X to stop, and Q to quit.

Task 3 — mapping and navigation:

rosconnect.m
step1_check_lidar.m
step2_online_slam.m          # move the robot while scans are collected
step3_build_save_map.m
task2_step1_show_map_pose.m
task2_step2_select_goals.m
auto_patrol.m                # reactive patrol
task3_step1_astar_planning.m # offline path preview
task3_step2_astar_nav.m      # single-goal navigation
task3_step3_astar_patrol.m   # multi-goal patrol

The included gazebo_map_20251103_1956.mat contains a MATLAB occupancyMap. The included task2_goal_points.mat contains five sample world-coordinate goals; selecting goals again overwrites this file.

Reference algorithm parameters

Component	Configuration
Lidar SLAM	20 cells/m, 3.5 m maximum range
Loop closure	threshold 200, search radius 3.5 m
Reactive patrol	front sector +/-30 degrees, 0.35 m safety distance
A*	8-connected grid, Euclidean heuristic, no diagonal corner cutting
Planning safety	occupancy threshold 0.65, 0.25 m map inflation
Path following	every fourth grid point used as a waypoint

Results

Stage	Result
Simulink two-lap controller
Teleoperation and sensors
SLAM occupancy map
Reactive patrol
A* patrol

The screenshots and videos were captured from completed simulation runs. The included five-goal MAT file is a reproducible sample and is not asserted to be the exact waypoint set used in every recorded run.

Verification

Automated tests cover ROS environment configuration and standalone A* planning, including straight-line planning, routing around a wall, blocked endpoints, and diagonal corner blocking:

results = runtests("tests");
assertSuccess(results);

All ROS package XML files are well formed. Runtime reproduction additionally requires a running ROS/Gazebo environment and correct host/guest networking.

The ROS package was also verified in a clean temporary Catkin workspace on Ubuntu 20.04.6 / ROS Noetic / Gazebo 11.13.0. catkin_make completed successfully, roslaunch started a fresh ROS master, Gazebo server/client, and spawn_tb3; the TurtleBot3 laser, differential-drive, joint-state, and odometry plugins initialized before the bounded validation run was shut down.

Known limitations

• The reactive controller can enter local minima in concave or tightly constrained environments.
• The A* follower uses simple waypoint tracking plus a short-range front-sector check; it is not a full local planner.
• The saved occupancy map and goals assume the same Gazebo world frame and initial robot placement.
• AVI demonstration files may need to be downloaded because GitHub does not preview every codec.

Author

Jiaxin Wu — @5Elaine