OSkAR: OpenCV Based Speaking Robot

|| Main || Project Details || Videos || Project Proposal v1.0 ||

Navigate using the links above.
For my personal webpage, go HERE

Current News


Things are going well with OSkAR! Spending a lot of time on classifiers now.


Work is beginning on the robotic platform to carry the beaglebone system. The platform itself is created from a LEGO Mindstorms NXT kit, and the platform can be controlled via bluetooth from an Android phone app called "NXT Remote Control". A mobile powerpack has been ordered to act as a power supply for the BBB. I am going to attempt to power the USB hub from the same supply, but I am not sure if that will be good enough to power all devices on the hub. I plan on having the camera and a speaker plugged in.

Description: W:\Project Website\images\robot_1.jpg


Compiled first OpenCV Program on the Beaglebone. Indicates whether the object of interest is on the left or right. Output is viewed via SSH Terminal.



I have not yet had success with cross-compiling code. I compiled all my code directly on the BBB. (The application was small, so it took less than a minute to compile.)

Compilation via Terminal: g++ oskar.cpp -L/usr/lib -lopencv_highgui -lopencv_core -lopencv_imgproc -lopencv_objdetect -o oskar)

 When attempting to run the OpenCV application using Angstrom's default desktop environment, a borderless white box would simply appear and do nothing. After some Google searching, I found a workaround solution posted HERE that involves installing a new desktop environment (Xfce).

 I also had to make sure that my code specified the correct resolution of my camera feed, and that it resized frames to match. I may post some code snippets to better explain this.

The resolution that was displayed on the BBB itself during testing was 320x240. It was kept small for performance reasons.

Description: W:\Project Website\images\beaglecv.png

---------- 1/10/2014----------
Used a CableWorks HDMI to Micro-VGA Converter to connect BBB up to an old VGA monitor.
Did not work out of the box, had to switch the BBB resolution via Putty Terminal and input the proper monitor settings. (1280x720 with refresh rate of 60)
Followed these steps:

 1. In BBB's uENV.txt, put:

2. ssh into BBB, login as root
3. Type:
export XAUTHORITY=' ls/var/run/gdm/auth-for-root-*/database'
Hit enter.
4. Type:
export DISPLAY=:0.0

Hit enter.
5. Type:
xrandr --output HDMI-0 --mode 1280x720 --rate 60
it enter.


Set up cross compiler for Beaglebone Black on Linux Netbook using
tutorial located here.


Created First Classifier for Door Frame Corners

Preliminary work has been done on creating classifiers for feature recognition in a corridor.

Description: W:\Project Website\images\door_class_1.pngDescription: W:\Project Website\images\door_class_2.png


Created Project Website