exploratory-SLAM

Exploratory-SLAM is a minimal, reproducible ROS Noetic setup to simulate a Husky robot exploring an unknown world with online SLAM and an OpenCV-based frontier ranker. It launches Gazebo + Husky, maps with gmapping, plans with move_base, and continuously publishes the next best frontier goal.

demo_exploratory.1.mp4

Setup

• Prerequisites: Docker and Docker Compose plugin.
• Setup Husky Navigation: ./scripts/vendor_husky_nav.sh
• Allow X11 (Linux): xhost +local: (revoke with xhost -local:)
• Build image: docker compose -f docker/docker-compose.yml build
• Start container: docker compose -f docker/docker-compose.yml up -d
• Shell into container: docker exec -it husky_gui bash
• First run builds automatically via docker/entrypoint.sh into volumes: /ws/build, /ws/devel, /ws/install.

Run

1) World + Husky (Gazebo)

• roslaunch frontier_ranker_ros sim_boxes.launch
• This enables UST10 2D LiDAR on Husky and loads worlds/exploratory.world.

2) SLAM + Navigation (gmapping + move_base)

• roslaunch husky_navigation gmapping_demo.launch
• Sanity checks:
  • rostopic hz /front/scan (>5 Hz), rostopic hz /map (>1 Hz)
  • rosservice list | grep make_plan (expect /move_base/make_plan)
  • If namespaced (e.g., /husky/move_base/make_plan), set:
    • rosparam set /frontier_ranker/make_plan_service /husky/move_base/make_plan

3) Frontier ranker

• roslaunch frontier_ranker_ros exploration_ranker.launch
• Publishes next goal on /move_base_simple/goal.

4) RViz (optional)

• roslaunch husky_viz view_robot.launch

Reproducibility

To ensure consistent results, this section lists all key configuration parameters that influence the exploration behavior and navigation characteristics.

Frontier Ranker Configuration

File: src/frontier_ranker_ros/config/params.yaml

Parameter	Description
alpha	Weight for frontier cluster size (larger = prefer bigger frontiers)
beta	Weight for path length (larger = prefer shorter paths)
gamma	Penalty for revisiting previously explored regions
min_cluster_size	Minimum frontier cluster size considered valid
frontier_connectivity	Connectivity threshold between frontier pixels
search_radius_m	Search radius in meters
dilate_kernel	Dilation kernel size used in frontier extraction
plan_tolerance	Allowed deviation in goal plan
path_sample_step	Path sampling step for evaluation
update_rate	Ranker update frequency (Hz)
max_plans_per_cycle	Maximum plans evaluated per iteration

These parameters define how aggressively or conservatively the robot explores.
For faster but less exhaustive runs, reduce gamma and increase alpha.
For more cautious exploration, increase beta and plan_tolerance.

---

Navigation Parameters (Vendor Configuration)

Located in:
src/vendor/husky/husky_navigation/config/

costmap_common.yaml

footprint_padding: 0.08

costmap_local.yaml

inflation:
  inflation_radius: 1.6
  cost_scaling_factor: 2.5

planner.yaml

DWAPlannerROS:
  occdist_scale: 2.0          # Prioritize clearance around obstacles
  path_distance_bias: 4.0     # Encourage staying close to the global path
  goal_distance_bias: 8.0     # Weight goal proximity in trajectory evaluation

  max_vel_x: 1.0
  max_vel_trans: 0.9
  max_vel_rot: 1.2

Tips

• Code in src/ is bind-mounted read-only; rebuild inside the container with catkin build.
• To wipe builds (clean slate): docker compose -f docker/docker-compose.yml down -v.