I'm going to build an autonomous vehicle with four wheels and two movable cameras. I'm a neuroscientist and my aim is to use the control principles we study in the brain to control the vehicle. I'm new to Arduino so I need some help choosing which parts to buy. I've already got an Arduino Uno and a Mega 2560.
Can you recommend a good chassi? Preferably one with four independent wheels/motors and a breadboard. Medium size, strong enough to carry some extra gear.
Can you recommend a camera that's good with arduino? (Do I need a shield here?)
What is the best solution for power? Is there a good rechargeable(?) battery that can power an Arduino, six motors and two cameras?
There are tons of robot chassis around and you need to understand your size/weight/speed requirements etc and budget and then choose one that seems to fit.
As far as cameras are concerned, these are not related to the Arduino in any way and you can use whatever you want (subject to your chassis being able to support them). If you were expecting the Arduino to do something to or with the video stream from the cameras, you're probably barking up the wrong tree and need to look for a different platform. But if you just want to use the Arduino as an I/O device to control the 'bot and perhaps point the camera in the right direction, and have some other device deal with video processing and the complex logic, that would be fine. Arduinos are just simple controllers, though - don't ask them to do any sort of video processing.
You could also build a chassis by yourself with some wood or aluminum or something as a base, and then you buy the wheels and motors or perhaps continuous servos for the wheels, and assemble them together. This will probably not looking amazingly professional as some preassembled chassis but eh well, at least you make it by yourself. As for the camera, you could use webcams and then connect them with your computer and then use a computer vision library, such as opencv, for the video processing. But in this case of course your compose needs to be traveling with the robot, so you could perhaps use a laptop or something like that as the computer.
Nice, thanks everyone. @PeterH, I think you're right about the cameras, maybe I should aim to stream the video to a proper computer with a graphics card and send the results of image processing back to the Arduino. But how do the teams in for example the Autonomous Vehicle Competition solve this? Do they have higher capacity microcontrollers of some kind?
I had a quick look around and I think I'll buy the Pirate-4WD Mobile Platform unless someone has a better suggestion..
The dfrobot base looks like a good choice, as it appears to be large and powerful enough
to carry a wifi camera or two, but you need to check the motor torque specs and estimate
the overall weight. I use NiMH AA-cells for most of my robot projects, they come in
2200-2500 mAH energy ratings nowadays, which will run small robots for an hour and more.
For a larger robot, I have a 6V SLA battery for the motors and the NiMH AA-cells for
electronics.
pololu.com has some bases, but they may be too small for this project.
There are smaller RF cameras, that may or may not be suitable, and depending on whether
you want to do onboard processing versus remote processing on a central PC, etc. https://www.dynaspy.com/
It takes a lot of cpu power to do vision processing, and small processors like Arduino will
not handle it. One possibility is the surveyor system for that, you can get the robot too,
Also, Edelman has a bunch of nice brain-based robots, in case you've not seen them,
but they could use some updating as vision science progresses of course, http://www.nsi.edu/~nomad/
For reference, one of the important things with a mobile robot is, the base does not
need to be very fast. Most RC cars run far too fast to be controlled by the usual robot
s.w., so most robot bases are relatively slow moving.
In line with RR's comment, I also use XBee modules for control and status comms between
all my robots and my PC, even though the robots are "autonomous". The XBee Pro modules
will provide up to 100m line of sight range. They are much too slow for transferring video,
of course.
chrisharris:
But how do the teams in for example the Autonomous Vehicle Competition solve this? Do they have higher capacity microcontrollers of some kind?
I don't know, but I would assume that they use computers rather than microcontrollers, and probably high spec computers at that. I would think that the starting point for your search should be something like a RPi, but they're still not at all powerful in computing terms and if your solution is vision based you might well need a more conventional computer - perhaps a laptop if you're short of space.
If Autonomous Vehicle Competition refers to sparkfun, those are likely to be basic
Robomagellan robots, which use microcontrollers and GPS, and there is a large amount
of info on Robomagellan on the web.
If AVC refers to the Darpa Challenge, well those are automobile-sized and many have
multiple 3-Ghz netbooks computers on board. Every single one of those relies mainly on
SICK laser scanners for navigation. http://www.hizook.com/projects/sick-lms-100-laser-rangefinder-lidar
Hi all, thanks for all your advice. Here's the robot I ended up getting, driving around on its own
It's a Rover 5 I got from SparkFun ($60). I'm very happy with it.
Power turned out to be straightforward. Still haven't decided on cameras.
I have many more questions but I'll use new threads for those.
