Let's Make Robots!


Great open-source platform for robotics, AI, computer vision researchers and DIY enthusiasts.

We are using Dagu's tracked chassis. The body of the vehicle is 3D printed. The BeagleBoard-xM is used as on-board computer to control wide range of sensors (such as ultrasonic range finders, compass, cameras, etc.).

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Pardon my ignorance but what advantages do you get from the analog camera and USB frame grabber that you wouldn't get from a USB web cam?

I love the 3d printed body and will definitely try to emulate your polished look with a design of my own on my 48-wheeled Rover 5.



There are several reasons why we decide for analog cameras:

  • They are smaller than typical web cams and it allows us to place two cameras on the moving part. Required frame grabbers are located in the middle of the robots body where there is enough space for them.
  • They have rather wide angle which is important in manual control mode. Also, it is possible to exchange the lens if even more wide angle will be required.
  • Analog camera plus grabber are usually cheaper then decent web camera.

> my 48-wheeled Rover 5.

Could you please say a few words about your experience with these wheels? How it works on the normal carpet, thick carpet or tiles or wet tiles, etc.?

Thanks for the informative response regarding the analog cameras, sounds like you have done your homework when it comes to spec'ing out your robot. I was thinking about picking up some cheap USB webcams for my robot SPANR but now I think I'll need to investigate further (lens viewing angle being an important factor in robot vision I expect).

I've yet to fully experiment with different surfaces but I have had it running on the rug in the video (tough, rough surface), thick carpet, an uneven polished concrete floor and astroturf (fake grass, plastic bristles). The robot strafes (runs sideways) on all surfaces with some wobble/change in orientation which is sometimes random and other times it can be anticipated (centre of gravity affects it, as does unevenness of the surface). As has been suggested in the comments on my robot page I may need to implement feedback from gyro/accelerometer/compass etc which I hope to look into. One major problem with the wheels is that I cannot turn on the spot (tank driving where I drive the left wheels forward and right wheels backwards together) as the O-rings which provide grip will pop off their wheels due to the slippage required on the wheels and the direction of travel/friction is perpendicular to the small wheels (this doesn't happen on all the wheels but it's often/regular enough that I'll need to stop and fix the wheels after attempting a turn).

Thank you very much for the information about wheel performance! I am currently thinking about different alternatives for trucks (different types of wheels) and your experience is a great help in decision-making process.

Ah, sorry, yeah, the xm looks like a nice board. I have a 512meg RPi, so looking forward to seeing how this progresses. Thanks for sharing. She's gorgeous.


How necessary is the Beaglebone?  Any chance of getting this to work with the Raspberry Pi.

> Very cool

Thanks! :-)

> How necessary is the Beaglebone?

We are using BeagleBoard-xM, not Bone.

> Any chance of getting this to work with the Raspberry Pi.

Until recently, RPi was irrelevant for us because we want to have real-time video compression on-board. Before publishing the API for hardware encoder it was not possible and that was the reason why we were using BB-xM (DSP-based h.264 encoder). However, now we are considering the RPi based version as well. Technically it should not be a big problem. It is just a question of time and money.

This has to be one of the classiest projects posted on LMR. When I saw the video from drivers seat view in the bot I thought immediately apart from the great robotics platform this is you could with a little more creative thinking create a game system with this.

Since we can control the vehicle from all over the world over the Internet, we were thinking about organizing races. One of our previous models were actually based on RC car: http://veter-project.blogspot.de/2011/11/how-to-build-beagleboard-based-wifi.html . However, due to the lack of place to build racing circuit and some other organization problems we drop the idea and decide to concentrate on robotics platform for researchers. But theoretically, it should be very easy to mount our electronic on some kind of RC car and drive around with it over the Internet.

Very nice body design! Looks like a great 'all in one' robot. The video processing works really well too. Are you capturing the feed with a camera and then aplying an edge detect algorithm and then using OpenCV to detect objects?