Guard Dog Robot Project

Its been three months since the last post in this blog..
Mainly due to finally delivering the GuarddoG thesis along with a presentation in an event organized by the aueb graphics group

I have started to use ARCH ( instead of Ubuntu Server ) in the GuarddoG onboard computer and thus some of the recent work has been finding pacman equivilents to apt-get packages , writing a new service starter that preloads v4l2 compatibility shared objects and some other environment/setup code ..

It is kind of weird to have “officially finished” with something that was such a big part of my everyday life for the last 4-5 years.. and I certainly want ( and probably will ) continue to actively develop it..! The truth though is that major work needs to be done , especially in the hardware/exterior part of the robot.

As always new commits can be found on github .. and due to the repository getting so big , I am thinking of “forking” a new one , probably RoboVisionNG or something like that to start off a fresh git history tree to reduce the size of the repo and remove obsolete history objects..
The RoboVision repository will still remain though..

It seems too good to believe ;P , but after all this time the guarddog “documentation” , my BSc thesis seems to be complete

Some issues remain with the table of contents and other typographic details but the fact of the matter is that the theory , ideas and details have now been passed from thoughts on my mind to paper and I am very happy .

Subjects analyzed for computer vision and robotics are camera calibration ( pinhole model , calibration , rectification ) , basic image processing, disparity mapping ( depth mapping by binocular stereoscopy ) , integral images , haar cascades , corner and feature detection , lukas kanade pyramid optical flow , ransac , A* Pathfinding and many others as well as lots of notes on implementing all these in a compact stack with serious performance considerations.

I hope that people around the web will find my work useful and it will help them understand and organize all those different concepts easier on their mind , since this is the goal of this work..

I have enlisted GuarddoG(RoboVision) in the Ohloh software directory..!
It generated some very interesting statistics..

RoboVision
Mostly written in C
Very few source code comments
Only a single active developer
Estimated project cost: $1,312,081

Codebase for Logic Only
49,134 lines (est.)
12 person-years
Estimated Cost
$ 656,665

https://www.ohloh.net/p/RoboVision
https://www.ohloh.net/p/RoboVision/analyses/latest

Similar Projects

https://www.ohloh.net/p/mrpt
https://www.ohloh.net/p/ROSorg
https://www.ohloh.net/p/yarp
https://www.ohloh.net/p/6399