rosInterctionRecognition

intersection recognition(ros ver.)

Requirement

• MeCab (Python)

Run

$ roslaunch intersection_recognition simple_hypothesis.launch

ROS Param

• simple_hypothesis.launch

  • extended_toe_finding
  • scan_hz : Frequency of 2D-LiDAR
  • distance_thresh : Threshold of LiDAR-beam
  • read_scenario.py
  • scenario_path : Path to .txt-file that a scenario is saved to

• hypothesis_simple.launch

  • toe_finding
  • scan_hz : Frequency of 2D-LiDAR
  • scan_off_set : How many adjacent point clouds to ignore
  • epsilon1 : Threshold
  • epsilon2 : Threshold
  • epsilon3 : Threshold

• hypothesis.launch

  • intersection_recognition_tensorflow.p
  • model_full_path: Path to load Neural Network model(.h5)

Node

• check_scenario : Compare scenario and hypothesis of intersection.
  • /rotate_rad : Specifies how much the robot should rotate with /rotate_rad (if /rotate_rad == 0, then go straight).
• cmd_vel_controller2 : The robot will rotate at the angle specified by /rotate_rad to publish cmd_vel.
  • /turn_finish_flg : Publish this flg when the robot rotates at the angle specified by /rotate_rad.
• /emergency_stop_flg : When this flg is true, the robot stop the move and the calculation.

Movie

On simulator (simple_hypothesis.launch)

https://www.youtube.com/watch?v=1WaGJrMmR0E

toe-finding Ver.

https://youtu.be/fdbXFAEWZBk

Machine Learning Ver.

https://youtu.be/1tNvuxRLBRw