Done & Gone

This project is long since defunct; I maintain a website at http://dinomite.net.

Busy-time

We get more and more busy as we near our goal...we're getting close. Things of note this week, OverallPathfinding is working (read: it compiles) and I'm going to be testing that today or tomorrow (Saturday/Sunday) and if we can get it and StepperCalculator running, we just might be moving the car on Sunday; Chris says that the National Instruments (NI) stuff that runs the Senior Design team's systems (throttle, brake, steering, speed) is ready to go.

Late last night (11pm-2am) was trying to do some more testing (simulator) of VehicleHandler, but it just found some problems. Tay and I worked on that this morning (it was before Noon) today as well, but problems aren't solved yet. Finding is good, though! On Wednesday, I spent a little bit of time watching videos from last year's start and at the meeting we went over things that we should watch out for. I'm most concerned about shadow confusion (which Tay is putting effort into mitigating) and the car being indecisive, either not going anywhere or turning in circles.

Monday

We had that meeting with the GCCIS guy and he was much happier seeing our new list of computer needs; he said it would be much more do-able, including the firewire card (for video) and CF cards/adapters that we're going to use instead of hard drives so that we don't have to worry about vibration. That means, however, that any writing will have to be done in RAM. We're going to make good use of RAM-disks and put things like /var and /tmp in RAM. I'm going to have to figure out how that's going to happen.

Tay and I also drove around to get some videos so that he can do more realisitic testing on his vision code. We drove around the loop, through the parking lots and on some walkways making sure to capture many different situations like going through gates and approaching insurmountable obstacles.

Hours today: 10

Wiring Day!

Sunday was wiring day and we got a lot done, though ran into a number of problems. We were unable to do the testing we wanted because it took a lot longer to install of the wires and connectors than we expected. Chris and Mark got the dashboard built, and it looks a lot better. The module boxes that they built look a lot better as well, and will definently be more reliable than what we once had. I worked on some of the connectors, and they are a lot harder to deal with than they should be. Quite annoying. We got them all done, except for one of the boxes, which we ran out of pins that we're going to have to order. Problems, but that's why we plan in a lot of leeway.

I wrote up a perl program to write low-level test code (commands for the ethernuts) so that we can test the workings of the hardware and associated communication stuff. As long as that turns out well (Darren worked out some kinks on Sunday), we'll be set to plug in the computer and drive, assuming StepperCalculator, VehicleHandler and Harware Interface work out, which they should. More code reviews concerning those this week.

Tay and I also went over what computers we're going to ask for at the meeting tomorrow with the donations coordinator for GCCIS. The initial estimate of $12k worth of computers (Dual 3.6GHz Xeon with 2GB of ram was our vision machine) was much more than we need. From testing, we've found that our algorithms aren't all that intense and run well-enough on Tay's 2.5GHz Athlon 64. The computers we're looking to get are two 2.6GHz machines, each with 512MB of ram. Meager by comparison (Dell's lowest priced machines) but should be OK for all of our testing. Two of those work out to around $1k, still much less than most of the other GC teams. Once again, we undercut the competition!

Hours: 8

Thursday and Friday

We did code review on Friday, looking through Tay's Hardware Interface code. Doing the code reviews is good for the obvious reason of finding errors and the somewhat connected, making suggestions about their algorithms and sparking new ideas. It's also great because we all gain a much deeper understanding of everyone else's code.

I rewrote some of the SPacket class, and (over) zealously removed some of the methods. Tay fixed up a a couple of those errors (Java and C++ have very different string classes) and it's better than the written-for-C-strings-but-used-on-real-strings code that it once was. I also commented the entire class heavily, so the next time changes need to be made, they'll be a lot easier to implement.

I also fixed a few problems that I had been having with the dual 350 that's probably going to end up running everything but the vision processing.

Hours Thursday and Friday: 19

Wednesday

Gotta get back onto the GPS code, as we're going to be using the USB GPS for now. It'll post to the world model (as it should), so when we replace it with the Telogis (or something else) no other changes will be needed. I also updated the constants file and a couple other small things.

Hours today: 11

Tuesday

Finished up the LIDAR code and worked on getting Rhubarb's OverallPathfinding code on the simulator. It did get integretad properly (communicating with the needed modules), but didn't actually do anything. There' more work to be done on that....

I got the OS back onto our dual Pentium 2, which we're going to run everything (but vision) on when we start doing in-car testing this week; it had some bad ram that had done bad things. It should be all set, and if we can we'll get the Pentium 3s back in there.

Hours today: 10

Monday

Finished up the laser code, got it pulling from the database and everything; just need to make it return data directly for subscriptions.

Hours: 9.5