Shared Robovision project for AEP, TU Berlin 2015.

This project contains 2 main components:

• Image-Processor
• Replayer

The image-processor is the part of the project processing the images coming from the two in-field cameras. It process each frame in order to find and locate all the player markers contained in the image. It is possible to log a specific part of a match for a later replay.

The replayer can replay any previously logged recording. It has settings for the replay speed.

It uses a simple data format, making it easy to manually create/edit logs for training purposes.

This project requires 2 USB cameras (preferably external ones) connected to the host computer.

The first camera should solely oversee the left side of the football pitch and the second one the right side.

The project has been built using Eclipse.

It uses the following libraries:

• Aruco 1.2.5+ (http://sourceforge.net/projects/aruco/)
• OpenCV 2.4.10

Install both libraries in the system (tested on Ubuntu 14.04+).

The following 2 tutorials are a good guidance for installing ArUco and calibrate the camera(s).

• http://maztories.blogspot.de/2013/07/installing-aruco-augmented-reality.html
• http://maztories.blogspot.com.es/2013/07/camera-calibration-with-opencv.html (needed for getting the intrinsics files)

** Linker Flags: **

• -std=c++11 -lm

** Include paths (-I):**

• {path_to_opencv_installation}/include
• {path_to_aruco_installation}/utils
• {path_to_aruco_installation}/src

**Libraries (-l): **

• opencv_core
• pthread
• aruco
• opencv_imgproc
• opencv_highgui
• opencv_calib3d

Library Search Path (-L):

{path_to_opencv_installation}/lib

Copy the 'res' folder in the same folder of the executable.

Image-processor:

#!bash

./shared-robovision intrinsics_left.yml intrinsics_right.yml teams_info.yml [log_file.txt]

Replayer:

#!bash

./replayer log_file.txt teams_info.yml [time_between_frames_in_ms] [step]

Config.h contains the following variables for tweaking the application:

• **MARKER_SIZE **: size of the marker (in meters)

• **PLAYER_RADIUS **: size of the player in the map

• **ROBOT_HEIGHT **: distance of the marker from the ground (in meters)

• **FIELD_HEIGHT **: height of the field image (in pixels)

• **FIELD_WIDTH **: width of the field image (in pixels)

• **CAMERA_HEIGHT **: distance of the camera from the ground (in meters)

• REAL_FIELD_HEIGHT: height of the field (in meters)

• **REAL_FIELD_WIDTH **: width of the field (in meters)

• **DEBUG **: run in debug mode

• **DEBUG_CAMERA **: use only local cameras

Here is an example of a team.yml:

team_1=TU-Berlin
short_1=TUB
colour_1=255,255,0
players_1=792,4,192.168.0.1;849,1,192.168.0.2;1009,5,192.168.0.3
#
team_2=DragonKnights
short_2=DK
colour_2=0,0,255
players_2=466,6,192.168.0.4;650,1,192.168.0.5;354,10,192.168.0.6
##

team_#: full name of the team #

short_#: short name of the team # (shown in the field)

colour_#: colour of the team #

players_#: tuples of 'id,number,IP_Address' separated by a ';'.

#: team separator (optional)

##: end of file (optional)

The team file is required for correctly displaying the information of the players on the field and being able to communicate with the players.

Note: The program currently supports only 2 teams.