Team Members

Name	Section	BN
Ahmed Nasser Ahmed	1	8
Ahmed Hisham Eid	1	9
Abdelrahman Ahmed Mohamed Farid	1	33
Youssef Walid Hassan	2	34

Launch Instructions

1. Install ROS melodic
2. Initialize catkin workspace
3. Place code inside src/ directory
4. Run catkin_make
5. Run roslaunch vehicle_sim_launcher corner.launch gpu:=true to launch gazebo simulation
6. Run rosrun slam_code main.py in another terminal to run the code

File Structure

Inside vehicle_sim/worlds we can find the gazebo worlds used Inside vehicle_sim/launcher/vehicle_sim_launcher/launch/corner.launch we can find the corner.launch file used Inside slam_code/main.py we can find the main python script for the project.

Code structure

main.py:

• The main loop consists of:
  • Subscribers
  • Publishers
  • The ROS loop which executes once every 1 second and is responsible for calling normalize_and_publish_map()
• Global variables
  • particle: particle class which contains the car position
• scan_matching_callback
  • Callback for scan matching
  • Responsible for updating the robot position globally
• transformed_lidar_reading_callback
  • Callback for the lidar readings
  • Responsible for:
    • Parsing the lidar PC2 into an np array
    • Transforming lidar points correctly into world and map point of views
    • Checking for out of bound particles
    • Filtering out car particles
    • Casting rays using cast_ray_and_update_map
• cast_ray_and_update_map
  • Casts rays using different methods:
    • naive: the usual unit vector casting method
    • bresenham: the bresenham algorithm which is more accurate
    • skimage_line_2d: only used for 2d ray casting, uses skimage.draw.line method
  • Updates map accordingly by log odds likelihood method
• normalize_and_publish_map
  • Responsible for mapping the grid array into probabilities and publishing this on /map_pc2 topic
  • Uses vectorized ways to quickly map into PC2 format bresenham3d.py:
• Contains 3d bresenham algorithm
• Copied from geeksforgeeks