/Dual-Filter-Safety-Critical-Control

Primary LanguagePython

E-FastSOD-Zone-Safety-Critical-Control

Obstacle Avoidance and Path Following for Mobile Robots using Salient Object Detection and buffer D-ZCBFs

This program demonstrates a novel algorithm that combines Salient Object Detection (SOD) and buffer robust CBFs obstacle avoidance to enable a vehicle equipped with a depth camera to avoid obstacles and reach a specified destination or follow a corresponding trajectory. The simulation environment is based on Ubuntu 20.04, ROS Noetic, and Gazebo 11.

Simulation Environment

The package is primarily used in a Gazebo simulation environment that includes the TurtleBot3 and RealSense camera ROS packages. The simulated robot is a TurtleBot3 Waffle Pi equipped with a RealSense D435i depth camera. The Rviz and Gazebo files have been configured to include depth camera data.

Scenarios

The package is mainly used in the Gazebo simulation environment and is divided into three types of scenarios:

1. Static Environment

Launch the Gazebo simulation environment:

roslaunch type2_D_ZCBFs turtlebot3_static_cylinder_obs_world.launch world_env:=1
  • world_env:=1: One obstacle environment
  • world_env:=2: Five obstacle environment
  • world_env:=3: Ten obstacle environment
  • world_env:=4: Twenty obstacle environment

Launch the salient object detection algorithm, point cloud filtering, obstacle circle construction and other algorithms for obtaining obstacle information:

roslaunch type2_D_ZCBFs turtlebot3_dangerous_obstacle_detection.launch

Finally, launch the vehicle obstacle avoidance algorithm:

python3 src/sod_avoidence/type2_D_ZCBFs/static_obstacles_avoidence/scripts/static_obs_typeII-d-zcbfs_avoidence.py

2. Dynamic Environment I (Simple circular obstacles, both static and dynamic in the simulation environment)

Launch the Gazebo simulation environment:

roslaunch type2_D_ZCBFs turtlebot3_mix_cylinder_obs_world.launch

Launch the salient object detection algorithm, point cloud filtering, obstacle circle construction and other algorithms for obtaining obstacle information:

roslaunch type2_D_ZCBFs turtlebot3_dangerous_obstacle_detection.launch

Launch the obstacle movement code:

python3 src/sod_avoidence/type2_D_ZCBFs/move_obstacles_avoidence/scripts/control_obs_move.py

Launch the vehicle obstacle avoidance algorithm:

python3 src/sod_avoidence/type2_D_ZCBFs/move_obstacles_avoidence/scripts/move_obs_typeII-d-zcbfs_avoidence.py

3. Dynamic Environment II (Simulated warehouse environment with 10 moving Turtlebot3 robots)

Launch the Gazebo simulation environment:

roslaunch type2_D_ZCBFs turtlebot3_warehouse_world.launch

Launch the launch file for the salient object detection algorithm, point cloud filtering, obstacle circle construction and other algorithms for obtaining obstacle information:

roslaunch type2_D_ZCBFs turtlebot3_dangerous_obstacle_detection.launch

Launch the vehicle movement code:

python3 src/sod_avoidence/type2_D_ZCBFs/move_obstacles_avoidence/scripts/control_robot_move.py

Launch the vehicle obstacle avoidance algorithm:

python3 src/sod_avoidence/type2_D_ZCBFs/move_obstacles_avoidence/scripts/move_obs_typeII-d-zcbfs_avoidence.py

4. Path Following (Sine function trajectory and elliptical trajectory)

4.1 Sine Function Trajectory Environment (including generating trajectories and displaying trajectories in Rviz)

Launch the Gazebo environment:

roslaunch type2_D_ZCBFs turtlebot3_sin_path_following.launch

Launch the launch file for the salient object detection algorithm, point cloud filtering, obstacle circle construction and other algorithms for obtaining obstacle information:

roslaunch type2_D_ZCBFs turtlebot3_dangerous_obstacle_detection.launch

Launch the vehicle trajectory tracking and obstacle avoidance algorithm:

python3 src/sod_avoidence/type2_D_ZCBFs/path_following_with_obstacles/scripts/sin_path_following_typeII-d-zcbfs_avoidence.py

4.2 Elliptical Trajectory Environment

Launch the Gazebo environment:

roslaunch type2_D_ZCBFs turtlebot3_elliptical_path_following.launch

Launch the launch file for the salient object detection algorithm, point cloud filtering, obstacle circle construction and other algorithms for obtaining obstacle information:

roslaunch type2_D_ZCBFs turtlebot3_dangerous_obstacle_detection.launch

Launch the vehicle trajectory tracking and obstacle avoidance algorithm:

python3 src/sod_avoidence/type2_D_ZCBFs/path_following_with_obstacles/scripts/elliptical_path_following_typeII-d-zcbfs_avoidence.py

The trajectory can be changed in python3 src/sod_avoidence/type2_D_ZCBFs/path_following_with_obstacles/scripts/path_generator.py. All comparison data, charts, and videos for each scenario are available in the daten_analyse folder.