Exercises#

This page contains four take-home exercises that reinforce the concepts from Lecture 13. Each exercise asks you to write code from scratch based on a specification – no starter code is provided.

All files should be created inside your ~/enpm605_ws/src/ workspace in the appropriate packages (nav_demo or your own extension package).

Exercise 1 – Build, Save, and Reload a Map

Goal

Build an occupancy grid of the Husarion world using slam_toolbox, save it to disk, and reload it for AMCL-based navigation.

Specification

  1. Launch Gazebo and slam_toolbox in mapping mode:

    # T1
ros2 launch rosbot_gazebo husarion_world.launch.py
# T2
ros2 launch nav_demo map_nav.launch.py mode:=mapping

  2. Drive the robot with teleop to cover the entire environment. Aim for at least one loop closure by returning to your starting area after exploring all rooms.

  3. Save the map under ~/enpm605_ws/maps/my_map using nav2_map_server:

    ros2 run nav2_map_server map_saver_cli -f ~/enpm605_ws/maps/my_map

  4. Verify the two files my_map.pgm and my_map.yaml exist and open the .pgm in an image viewer.

  5. Re-launch in navigation mode pointing to your saved map and confirm the robot localizes after a single 2D Pose Estimate click in RViz2.

Verification

  • White, black, and grey regions of the .pgm correspond to free, occupied, and unknown space.

  • In RViz2, the AMCL particle cloud converges to a tight cluster around the robot within a few seconds of motion.

  • A Nav2 Goal click successfully drives the robot to the requested pose.

Exercise 2 – Inflation Radius and Path Quality

Goal

Investigate how the inflation layer’s radius affects the planned path and the robot’s clearance from obstacles.

Specification

  1. Locate the Nav2 parameter file used by map_nav.launch.py (typically nav_demo/config/nav2_params.yaml).

  2. Set the inflation_layer.inflation_radius of the global costmap to each of the following values, one at a time:

    • 0.10

    • 0.30

    • 0.55

    • 0.80

  3. For each value, launch navigation, send the same goal from RViz2, and:

    • Capture a screenshot of the planned path overlaid on the global costmap.

    • Note whether the path is feasible (the robot reaches the goal) or fails (no plan, recovery loop, or collision).

  4. Write a short reflection (one paragraph) answering: at what radius does the planner start refusing to plan through narrow doorways, and why?

Verification

  • Four screenshots labelled with their inflation radius.

  • One paragraph of analysis tying the result to the robot’s circumscribed radius.

Exercise 3 – Sequential Goals with BasicNavigator

Goal

Write a Python node that drives the robot through a sequence of three goals using the nav2_simple_commander API and reports feedback while moving.

Specification

Create nav_demo/nav_demo/sequential_goals.py.

  1. The node must:

    • Instantiate a BasicNavigator.

    • Set the initial pose from the latest map -> base_link TF lookup (use a 2-second timeout).

    • Call waitUntilNav2Active() before sending goals.

  2. Define three goal poses (read from ROS 2 parameters):

    • goal_1 (default [2.0, 0.0, 0.0])

    • goal_2 (default [5.0, 1.0, 1.57])

    • goal_3 (default [-5.42, 11.22, 3.14])

    Each parameter is a length-3 double_array of \((x, y, \text{yaw})\).

  3. Send the goals one at a time with goToPose. Between goals:

    • Print feedback every second: distance remaining and time elapsed.

    • When the goal completes, log the final TaskResult (SUCCEEDED / CANCELED / FAILED).

  4. After all three goals, exit cleanly with rclpy.shutdown().

Verification

# T1
ros2 launch rosbot_gazebo husarion_world.launch.py rviz:=False
# T2
ros2 launch nav_demo map_nav.launch.py mode:=navigation
# T3
ros2 run nav_demo sequential_goals

The robot visits all three goals in order, and the terminal shows distance-remaining feedback for each leg.

Exercise 4 – Cancel-on-Distance Behavior

Goal

Extend the sequential-goals node from Exercise 3 with a safety behavior: if the robot has not made progress (distance remaining has not decreased) for five seconds, cancel the current goal and move on to the next one.

Specification

  1. Inside the goal-monitoring loop, track distance_remaining from each feedback message.

  2. Maintain a last_progress_time timestamp; update it whenever distance_remaining decreases by at least 0.05 m.

  3. If now - last_progress_time > 5.0 seconds while the goal is still active:

    • Call self._navigator.cancelTask().

    • Log "No progress -- cancelling goal." at WARN level.

    • Continue with the next goal in the sequence.

  4. The node must always finish, even if some goals were cancelled, and print a final summary like:

    goal_1: SUCCEEDED
goal_2: CANCELED
goal_3: SUCCEEDED

Verification

Place a wall in Gazebo (or pick a goal inside an obstacle) so that one of the goals becomes unreachable. The node should cancel that goal after five seconds of no progress and continue with the remaining goals without crashing.

This Page