Building a semi-autonomous drone with Python
blog.yhat.com
blog.yhat.com
Maybe I'm overreacting, but I always get a bitter taste in my mouth when I start a tutorial whose title claims to teach something but then gaps a huge problem using a product.
This is a form of growth hacking that screws over people who come to learn.
So many companies do this, it's very frustrating.
They could turn this into such a positive by introducing the technical way first... They'd then have it all laid out to describe their product and how much code it saves.
Now, http://dronegames.co/ is still touting the 2013 Games.
Parrot has released many new drones since the AR.Drone, but from what I can tell it's not that the hackers are now working on the Bebop or Rolling Spider; I'm not sure where they went.
I'm a committer on the ar-drone node library, but lately I've mostly been working on turboshrimp, a clojure library for AR.Drone control: https://github.com/wiseman/turboshrimp (API docs at http://wiseman.github.io/turboshrimp/com.lemondronor.turbosh...). I've also got an Android app (written in Clojure)[1] and a desktop app that uses the CamShift tracking algorithm with OpenCV, which lets you select a target of interest and then will continue to track that target[3].
The AR.Drone is still a really nice, relatively hackable robotics platform. Running Linux and having USB lets you do things like connect an RTL-SDR dongle[4] and display an overlay of nearby aircraft on the drone video stream[5].
[1] https://github.com/wiseman/shrimpdroid
[2] https://github.com/wiseman/turboshrimp-tracker/
[3] https://www.youtube.com/watch?v=In6KFGakGGw
[4] http://lemondronor.com/blog/indexphp/2013/4/cheap-ads-b-on-a...
[5] http://lemondronor.com/blog/indexphp/2013/5/augmented-realit...
https://github.com/AutonomyLab/bebop_autonomy https://github.com/AutonomyLab/ardrone_autonomy
these are quite well-used in the research community.