I'm working on a bachelor's project, together with 5 other students, where we're going to continue development of and improve a laser guided AGV.
The AGV in picture and movies is the one produced by a group last year and we're keeping most of it. The major change we're planning is to remove the loading area on top and instead have it fetch a trailer of some sort. Minor changes include adding turn sensors on drive wheels to get better movement feedback - one of the problems with the old one is unreliable position data from the NAV200 but it will hopefully improve if we feed it somewhat accurate velocity data. We might also stick some ultrasonic sensors for collision avoidance on it. Otherwise the main problem is to rewrite the control software to make it a lot faster and more reliable. Hopefully we'll also be able to find out why the bluetooth connection to the central computer drops every now and then.
010218 update Been a little busy at university so no time for updates. We're completely rewriting the software, both AGV firmware and the controlling / route planning PC software. So far the bluetooth and motors are working, so we have a pretty fast computer controlled RC tank :) We're installing the homemade tachometers so we can create a control loop for the propulsion.
And we've also changed the plans of how you will work with the AGV. Now you'll mark disallowed areas on the map and the AGV will find the shortest route to the goal automatically, instead of you having to place navigation nodes and draw edges/lanes.
Today we're (re)starting the work of integrating the NAV200 with the rest of the system.
Good luck with your project. It looks like a nice platform to start with.
The SRF05 is a favorite ultrasonic sensor of many on this site, but there are others out there of course. The Ping sensor is another widely available one.
For turn sensors, are you looking at optical encoders? These could help you not only detect how far you’ve turned the tracks, but also provide feedback so you can be sure you are driving straight.
What mechanism will you use to find the trailer and ensure the robot can hook up successfully?
I think we’ll probably use some simple mechanism to measure wheel turns, like an ir diode with a stripe pattern on the wheel. Gotta look into if the ATmega can do the counting, or if it will cause too many interrupts… I’m not too happy with the choice of putting tracks on the AGV as I’ve got a feeling they might slip when turning while moving, thus causing speed measure errors.
Hopefully the position data from NAV200 will be enough for docking, as long as the trailer/loading table stays put. I think the previous year’s team got 10mm precision with their code but we’ll of course try to improve that!
hay hay i was planning on getting some of those tracks , would you recomend them for outside use(only going over grass ) and how much noise do they make? in the video it seems like alot
I don’t think they’d be suitable for grass. Maybe concrete or asphalt if you have to go outdoors, but they’re pretty smooth and there are thin “slits” in the rubber (approx 0.5mm) so I think they would slip on the grass and tons of dirt would get stuck in the tracks.
— edit: our motors make a lot of noise so I can’t really hear if the tracks do as well
