Friday

9:30AM-5PM
Worked in the MDRC lab on:
-Simulator and fake ethernut integration with VehicleHandler (our hardwar interface)
-Got the dualie computer that we have registered and setup with our code
-Continued working on the LIDAR code; requires some refactoring to integrate with our WorldModel thing
6PM-7:30
-Got the simulator working with VehicleHandler; we are ready to drive the car....sort of. There are always other things that we depend on, but this is one piece that's tested and ready to go!
-Laser stuff is coming together; got the world model things that I need defined in the wiki, so everyone else knows what's up.

Thursday

9AM-2:30PM
-Worked on the fake ethernut/simulator with Tay in the MDRC lab
-Updated bios, tried P3 processors, installed Debian on new machine; we'll probably use it for Vehicle Handler.
5-8PM
-Turning LIDAR code into GCART-ified goodness

Tuesday

10AM-12:30
-Team meeting
Noon-2:30
-Looked at Damn Small Linux (50 megs with X) and Bonzai Linux (180 megs with KDE); probably going to use Bonzai and just remove KDE and such. Bonzai is down to 140M with GCC.
2:30-6
-Fake ethernut: gonna have it send out data (GPS, laser?) as well as receive the velocity/turnrate
-testing: it works!

Monday

8:30AM-10AM
-Started work on fake-ethernut code
10AM-Noon
-did some work on the GPS stuff
Noon-5PM
-back to the fake ethernut with Tay's help
-FakeNut.h working, in CVS; by working I mean the code is proper. Doesn't yet compile.
-Started working on a small Linux install; can we use 256MB CF cards?
5-7PM
-Debian install ~230 megs with GCC
10PM
-144 megs without GCC and some other stuff

GPS data pulling

I've been working on the GPS data pulling and passing it on to Rhubarb's code (overall pathfinding). I'm grabbing data from Chris' USB GPS using gpsd, and then passing it on to Rhubarb using his update() method from the OverallPathFinding module. I did most of it in Perl, because I wanted to get the regular expressions for parsing the incoming gpsd stuff down quickly. I got that moved over to C, with much ado. Character arrays are really a very bad invention.

Hours Monday: 9
Hours Tuesday: 10

This week

We decided on the GPS that we're going to get, the Garmin 16A which can be had for less than $300. In the mean time, I've been working with Chris' TN-200 (Rayming TripNav 200) which, after much trouble, I got to work with gpsd, though I can't get a fix with it in my room. I've started writing the code to provide this GPS info to Rhubarb, as I'm pretty sure gpsd will work with the Garmin 16A.

Monday: 10
Tuesday: 9
Wednesday: 8
Thursday: 8
Friday: 2

Video of simulator

I took the simple obstacle avoidance code that I had and made a video of the simulator doing it's thing.

I also got an e-mail back from Riegl about their laser scanner; it's $27,500, which is a lot more than the SICK unit. Sensors are really expensive.

Hours worked: 8

From further research, it seems to me that the GPS16/16A from Garmin is our most reasonable choice for a GPS unit. The next step up in accuracy is also a significant jump in price (thousands of dollars). So, for $170 we can get good enough accuracy and speed-of-reporting from the GPS16.

I still haven't heard back from Riegl as to the price of their laser scanner, but I hope it's less than the SICK.

Hours yesterday: 7

GPS units

I started looking at other GPS units yesterday, since we're not going to be using Applanix's $80,000 POS LV, even with their 5% discount. I looked at Garmin's products and found the 16/16A which provide better than 3 meter accuracy with WAAS, 2 second recovery from dropout an updates every second for the GPS 16 and 5 times per second for the 16A. I also talked with a guy from CMC Electronics who gave me some insight into what I should be looking at for our vehicle.

Hours worked yesterday: 5

Lasers and GPS

I scoured the internet today looking for laser radars that fit our needs. SICK, the most well known company in the challenge, has a number of systems that would work, the most appropriate of which is the LMS221. The LMS221 has a 45M/80M max range depending on visiblity and reflectivity, 180 degree field of view and connects via RS 232/422 at a rate of one 180 scan every 13ms with an angular resolution of 1 degree. The SICK LMS221 can be had for less than $8,000. The other option, made by Riegl is their LMS-Q120 which provides a .04 degree angular resolution scan in 800ms or a 1 degree scan every 32ms. The Riegl unit provides an 80 degree field of view and communicates via Ethernet. I haven't yet gotten a price from Riegl.

Hours worked: 9

Making a Shelter

We built a shelter in the barn today. It's the width of the barn (30 feet), 13 feet tall and 26 feet long, made of translucent plastic. It's supported by steel cable that we strung and tacked in place with screw-washers and zip ties. Greg's dad came (from Maryland!) and setup a 20,000 watt heater that keeps the enclosure pretty warm. We also gathered all of the tools from the Mechanical Engineering lab (where we had been working previously) and moved them to the barn. All of this took quite some time and a lot of hard work.

Hours worked: 13