Thanks to robotshop for the roboard and the sabertooth motor controller.
Now I need sensors… lots of sensors.
First thing I want is a gps, compass, and tilt sensors.
Also going to need a range finding device. Would like a simplifid laser rangefinding. But more or $100 not $2000.
Yeah the wooden tracks are really neat. Really proud of them. Took a very long time in frount of a drill press. But there is no glue, or hardware holding anything together. It is oldschool dowel rod construction. They sounds amazing going down the road.
I2C devices can be attached to the bus and addressed diffrently? I have done a series of pics on a serial bus and just gave each one a diffrent address. Would the sonar sensors work the same way? They look good but will have to do the research.
SRF02 vs SRF08: if building an array I would be inclined to go with the cheaper model. But besides for the 12cm close end range lose what else do you lose for money.
Another thought was maybe a SRF235 on a turret?
And yes the laser rangefinder hack is something I wanted to try for a long time, but never have money to try. But maybe soon.
Yeah the tracks where a project in themselfs. Lots of time in frount of a drill press. I am thinking about a redesign this summer however to make them more sturdy. Might not have a problem really. They work surprizingly well.
The insperation for them came from: woodgears.ca/tracked_vehicle/index.html
Steampunky looks interesting. Will have to do some research to find out what I want the final design to “look like” Thanks for the artisitic input I am very bad when it comes to that sort of thing.
I allready have a older usb gps. But if it does not work the USGlobalSat is my second choice.
It is controlled by the internet now. It has a bullet2 wifi router set up as a bridge. It is from: ubnt.com/bullet but it is temporary until I can get antenna for ubnt.com/sr9. I have the radios but not the connectors or the antenna. But with 900mhz NLOS wifi radio is just pefect for a robot where you will be moving around ect. I highly recommend ubnt products. High quality stuff.
Well I don’t think weight vs torque will be a big issue. The motors and gearbox is from a very large powerwheels jeep designed to hold two toddlers. Weight vs power comsumption is my big worry. The closer to stall those motors get the closer they become big waste of power.
Like the idea of balsa. Getting it into a barrel shape should be fun.
I am going to be upgrading the battery soon. To a 12v 20Ah motorcycle battery. Once I get my external power connector finished. Might work on that tonight.
Been almost a year since my last update. But the project had alot of advances in the summer and got a completly working chassie but got stalled late summer by an attemt at sterio vision system. That yeilded the result that the roboard is not a sutable plateform for advanced image processing.
Then my wife had baby number 3 and then winter hit and roboard went into a coma for a still unknown reason it would not boot.
Luckly it was under warrenty and the Robotshop people quickly and professonally RMA and I got the Roboard back as good as new.
Now the first body was WAY too big (for developent indoors) so I am redesigning it to be a portalble version of the first one.
The platform is going to be extreamly small. Not to spoil the surprize but to give you a sence of scale the track feet pads are going to be made from popsicle stick segments. And body will be balsa wood. without battery the whole thing will be less then a pound. Later this tiny robot will be implanted into larger platform(s) where the propulsion system to match. Reusing the Roboard’s i/o and support structure. For any number of robots.
On the vision side we tried several vision packages. But the biggest problem with USB cameras is the USB compression. AND that most drivers that we could find do not allow you to pull a single frame on demand they wanted to stream a video stream of data across the bus(which for 2 camera even at extreamly low RES and Framerate the roboad had perforemance issues.) And the drivers that did support the single frame on demand would just BSOD the roboad on initilizing. Think the whole processor instruction set diffrences are to blame there. And I ran out of tim to reverse engineer existing drivers and add what I need.
Conclusion: The plan has always been that the Roboard is the hardware and sensor foundation of a much larger network of computers build on the platform. So crunching all the computations onto it was quickly going to become obsolete and unnessesary. I am sure with enough software grease it would work.
Proposed work arounds is get either serial cameras like the CMU series or get a TCPIP cam where the images could be stored and accessed on demand to several diffrent proccessing units at once. Almost having a “camera” server. This will work for very large platforms but less feasable as the platform shrinks.
Congratulations on the new arrival to your family! Based on your propultion (Parallax continuous rotation servos) the platform cannot be too large. You’ll likely need DC gear motors if you want to make is larger. Can you clarify what happened with creating a vision system with the RoBoard? Did you use a pre-made software package?
How about an array of ultrasonic rangefinders (SRF08) for close to mid range object avoidance, laser for mid to long rage avoidance or mapping and GPS for waypoint navigation?
Roboard supports I2C so try to stick with an array of I2C sensors for simplicity. I didn’t mention IR but they are also sometimes used in tandem with ultrasonics as they complement each other’s weaknesses. Just be weary that some people report “issues” with IR sensors in direct sunlight and with some lighting.
Those Hokuyo Lidars are ideal if you have the budget for them (UTM-30LX is drool worthy) and are perfect for Roboard. Otherwise perhaps hack a golfing laser rangefinder and have it sweep 180 degrees in front of the tank (always wanted to hack one of these things).
