Skip to content

Latest commit

 

History

History
92 lines (69 loc) · 3.2 KB

README.md

File metadata and controls

92 lines (69 loc) · 3.2 KB
Raw

ecn_turtlebot

[ROS package] M1 group proyect

In collaboration with Mahmoud Ali

The aim of this project is to perform localization using an Extended Kalman Filter (EKF) and beacons as visual markers. Detailed information can be found in the PDF report.

Software

  • ROS kinect
  • C++
  • LaTeX

Hardware

  • Turtlebot 2
  • Omni 60 (omnidirectional camera capruting 360° panoramic images at 60Hz)

turtlebot

Nodes description

Image processing nodes (using OpenCV)

image_rectifier rectifies images published by camera (Omni 60)

  • Subscribes to
    • camera_info: intrensic parameters and distortion coefficients required for image rectification
    • image_raw: image itself
  • Publishes to
    • image_rect: rectified image

raw_image rectified

image_tiles combines 5 different images into single panoramic image (no stitching)

  • Subscribes to
    • camera0/image_rect: rectified image from camera0
    • camera1/image_rect: rectified image from camera1
    • camera2/image_rect: rectified image from camera2
    • camera3/image_rect: rectified image from camera3
    • camera4/image_rect: rectified image from camera4
  • Publishes to
    • image_tiles: resulting panoramic image

tiles

blob detector_v1 outputs distance between a beacon (blob) and the omni camera. HSV thesholding and morhological operations are used as pre-processing methods.

  • Subscribes to
    • image: often remapped to image_tiles in our case
  • Publishes to
    • beacon_distance: distance in cm
  • Parameters
    • invert: boolean variable used to determine if image should be inverted or not (useful for solving discontinuity in HSV transformation)
    • HMin: minimum threshold value for Hue channel
    • HMax: maximum threshold value for Hue channel
    • SMin: minimum threshold value for Saturation channel
    • SMax: maximum threshold value for Saturation channel
    • SMin: minimum threshold value for Value channel
    • SMax: maximum threshold value for Value channel

green_blob green_mask

blob dist interface manages the distances associated with the beacons in the scenario

  • Subscribes to
    • blue\beacon_distance: often remapped to image_tiles in our case
    • red\beacon_distance: often remapped to image_tiles in our case
    • green\beacon_distance: often remapped to image_tiles in our case
  • Publishes to
    • beacon_distances: distances in cm [blue, red, green]

Localization nodes

odom interface extract relevand data from the nav_msgd/Odometry coming from the turtlebot

  • Subscribes to
    • odom: contains information about the 3D pose of the turtlebot
  • Publishes to
    • odom_2D: publishes 2D pose data (x, y, theta) and their covariance

ekf_node implements the extended kalman filter (not fully tested)

  • Subscribes to
    • odom2D: contains 2D pose data (x, y, theta) and corresponding covariance
    • beacon_distances: contains distances in cm [blue, red, green]
  • Publishes to
    • robot_pose_ekf: 2D pose estimation