Skip to content
This repository was archived by the owner on Aug 20, 2026. It is now read-only.
wukongxzeroPublic archive

About

This is a home service robot that uses previous concepts slam ,localizationa and mapping for the udacity robotics software engineer course

Resources

Stars

0 stars

Watchers

1 watching

Forks

Latest commit

 

History

3 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Home Service Bot

A full autonomy stack for an indoor mobile robot: AMCL localisation on a pre-built map, Nav Stack path planning and obstacle avoidance, and a pick-and-place pipeline. Built on ROS 1 Noetic with Gazebo simulation.

What it does

The robot navigates autonomously through a building, picks up a virtual marker from a designated zone, and delivers it to a drop-off zone — simulating a warehouse pick-and-place operation.

Full pipeline: localise → navigate to pick zone → pick up marker → navigate to drop zone → place marker.

Stack

Component Package What it does
Localisation amcl Particle filter on pre-built map from LiDAR + odometry
Map server map_server Serves pre-built occupancy grid
Path planning move_base (global: A*, local: DWA) Plans and executes collision-free paths
SLAM (map building) rtabmap_ros RGB-D SLAM to build the map before autonomy
Navigation ROS Nav Stack Global + local costmaps, recovery behaviours
Marker logic add_markers (custom C++) Shows/hides RViz markers at pick/drop zones
Pick logic pick_objects (custom C++) Sends navigation goals and triggers pick/place

How AMCL works

Adaptive Monte Carlo Localisation — particle filter on a known 2D map:

  1. Scatter N particles (pose hypotheses) on the map
  2. Propagate each particle with odometry + noise (motion model)
  3. Weight each particle by how well its predicted LiDAR scan matches the actual scan (sensor model: likelihood_field)
  4. Resample — particles cluster around the most likely pose
  5. Adapt N — more particles when uncertain, fewer when localised

Needs: /scan (LiDAR), /odom (wheel odometry), a pre-built map. TF tree: map → odom → base_link.

Costmap parameters

# common
obstacle_range: 1.0
robot_radius: 0.1
inflation_radius: 0.5

# local costmap (rolling window)
width: 2.5, height: 2.5, resolution: 0.05
rolling_window: true

# global costmap (full map)
width: 20.0, height: 20.0, resolution: 0.05
static_map: true

Run the simulation

catkin_make
source devel/setup.bash

# Full home service task (opens all launch files):
src/scripts/home_service.sh

# Individual test scripts:
src/scripts/test_slam.sh        # test SLAM
src/scripts/test_navigation.sh  # manual teleoperation
src/scripts/pick_objects.sh     # test autonomous navigation goals
src/scripts/add_marker.sh       # test marker show/hide

Repository layout

src/
  my_robot/          URDF, Gazebo world, robot description
  slam_gmapping/     SLAM configuration
  navigation/        move_base, costmap configs
  pick_objects/      C++ node: sends navigation goals
  add_markers/       C++ node: RViz marker lifecycle
  map/               Pre-built occupancy grid
  scripts/           Shell scripts to launch scenarios
  rvizConfig/        RViz configurations

Tech stack

ROS 1 Noetic · C++ · Gazebo · AMCL · rtabmap_ros · move_base · RViz

About

This is a home service robot that uses previous concepts slam ,localizationa and mapping for the udacity robotics software engineer course

Resources

Stars

0 stars

Watchers

1 watching

Forks

Releases

Packages

Contributors

Languages