So, how was the class AVC course this year?

[wholder]

I wasn't able to attend, but I did notice a few things from photos and videos that seemed new.  First, the tall, chain length fence looked to have been replaced by a much smaller fence that would be better for reducing GPS multi path.  The course seemed more challenging, but narrower to my eye.  How did these changes affect your results?  Any other insights that might help with planning for next year?

Wayne

[Ted Meyers]

Wayne,

Hope you can make it next year!

I didn't use GPS, but I heard from the other competitors that GPS was bad; possibly due to the building and/or the mountains.  Also, I suspect that an 8 foot lane is really narrow compared to the GPS error margins for most consumer GPS units.

Yes, the course was very challenging!  We spent hours calibrating and testing to it on Friday and Saturday morning.  In the end, we opted to take the dirt covered shortcut, which worked out quite well.  We avoided the barrels and it was a lot shorter.  The only downside was some loss in wheel encoder ticks (likely due to slippage), but it was consistent enough that we could compensate.  Going through the hoop and over the ramp was almost required, due to the narrow lanes.  The lower fences were great for spectators, but the hay bales made it hard to see the bots at times, and loose hay caused some traction problems.

Ted

[Thomas Roell]

For us it was a mixed bag. I really like the new course a lot. The hey balls make it tricky and the dirt path is fun as well.

We were running GPS. A simple programming mistake with the waypoint software added a off by 2m on Friday afternoon, so we were not able to do a good setup. On Saturday one of our rovers (Leon's) simply refused to work and I spend all of my time trying to figure out what got it killed. Got a lot of data, but haven't gotten any time (or motivation) yet to look at it.

GPS seems to be a tad better. Given our troubles I could not go throu all of the tests I wanted, so no good data. On some runs we were able to get 12 GPS satellites with WAAS kicking in, on some others only 8. I tried a few runs (which of course all ended up in the hey balls) with disabling WAAS, and various elevation filters for the satellites. Didn't do any experiments with GPS+GLONASS. On this site there is a shift during the day on where the GPS fix pulls you. We had a filter for that, but that did not work out that well due to the said hardware problems (actually electric and hardware).

What was really frustrating is that we did some testing the days before (the software was the same as last year's with some more filters on), on similar courses, on various parking lots with various interference. On all of those runs there were only a few bad ones, and about 95% of the runs had less than a +/- 30cm error. Zero hardware failrues.

Scheme that we were using was GPS + Z-GYRO + WheelTick.

- Thomas

On Sunday, September 18, 2016 at 6:58:01 PM UTC-6, wholder wrote:

[Minuteman]

The new course is fun. It's a good mash-up of newer and traditional obstacles. The hoop and jump are a bit of a joke, now, as it's easier to hit them than to get around them. The hay bales make it difficult to see what is going on at some points, but they were really good at not breaking suspension parts. The non-gps bots ruled the day again, and were still given a bonus for doing so. (Pete Dokter was not happy about that one as he hadn't authorized it). The balance bots, one potato and three potato (?) killed it.

I took 3 bots, which in this case were too much. I upgraded the code in Minuteman, which took some debugging. The emaxx was a last-minute effort and needed a lot of extra attention. It was only sort-of running on the day of the race. I never got it set up on the course, and had to use minuteman's waypoints, but it managed to navigate most of the course, even making it through the hoop. Now, I guess it's ready for next year!

Roadrunner (Rich) did well, but managed to miss one of the final waypoints on his last run, so had to settle for 2nd.

What I liked best this year was the flexibility on starting. They had everyone from the heat line-up, but start individually. So, you could wait for the course to clear off a bit and avoid crashes.

We are already planning next year!

On Sunday, September 18, 2016 at 6:58:01 PM UTC-6, wholder wrote:

[Thomas Roell]

On Monday, September 19, 2016 at 6:20:23 AM UTC-6, Minuteman wrote:

We are already planning next year!

Yup. My son said he would do the programming himself next year instead of relying on my shitty stuff ;-)

- Thomas 

[Minuteman]

Taking on an advisory roll eh? At least you're still on the team I guess.

[Thomas Roell]

I am not so sure, as to whether I'll be just relegated to being the taxi driver who drops him there ;-)

- Thomas

[Minuteman]

Ironically, I heard that Lyft will be all self-driving vehicles within 5 years. :)

[Jon Watte]

Note that Volvo and Lyft and Uber and a few other claim that.

But Google, who has gotten the furthest, is not.

Self-driving on a freeway or separated highway? No problem.

Following the signs of a road worker who's manually enforcing one-way around some debris removal, in a school zone, heading straight into the sunset?

Yeah, no, that's actually QUITE A BIT harder!

Sincerely,

jw

Sincerely,

Jon Watte


--
"I find that the harder I work, the more luck I seem to have." -- Thomas Jefferson

--
You received this message because you are subscribed to the Google Groups "diyrovers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to diyrovers+unsubscribe@googlegroups.com.
To post to this group, send email to diyr...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/diyrovers/b190ead4-d552-4bf0-a681-3f13581f7d3e%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.