Got the word from Peter over at at SparkFun earlier today. He reviewed our video and gave the thumbs up. So we’re officially in the autonomous vehicle contest next month.
Complete update. Decided its just easier to host the video(s) here. On that note, here is the video submitted to Sparkfun.
Our race day vehicle is finally ready for some action. Here are some pics:
After two weeks of battling the speed control, the Traxxas Slash is ready. It has 4 stationary IR range finders, one scanning IR range finder, a compass, a GPS, and a wheel encoder.
We’ve using X-Bee modules to send information and commands between the vehicle and a laptop. After we added all the sensor information and other feedback to Proto-2, I started to see communication drop outs when attempting bi-directional comm. Since it is a stop-gap mechanism for test and debug, I wasn’t too worried about it but wow has it been annoying.
The modules are the X-Bee 900 Pro‘s. We had the baud rate ramped up to 115.2 but decided after some calculations to drop back to 38400. No help. So I dug through the manual again. Likely, we should be using API mode but for the moment we kept it simple and just left them in the default broadcast configuration.
The downside of this mode is that the modules are not really communicating point to point even though only two are in the system. So they duplicate broadcast messages to help assure the messages get to everyone in the message. The default duplication is set to 3. I lowered it to 1. If we drop a message here or there, it shouldn’t really effect the performance of the system significantly.
Also, the vehicle is really and end-point, not a router and will never be for this configuration so I changed it to be an end-point not a router. However, that required a firmware upgrade which was seamless using the X-CTU software other than it defaulting the module back to 9600 baud. Honestly both could be set as endpoints I think but I didn’t test that configuration since I was in a rush to get the comm working.
I also tried lowering the packet ratio to zero prior to the other changes to immediately transmit characters rather than bundling messages. It proved too not help but might be a good option at some point. I also lowered the XOFF buffer maximum to 128 characters down from the default of 319 since none of our message strings exceed that length.
If we were going to rely on the radios for good communication, the API interface is the way to go not the transparent default. The changes cleared up the congestion but I made mass changes so cannot specify precisely which one did the trick. If you run into similar issues, I can check the exact configuration and setting changes.
In the end, one more hurdle cleared.
Geeknight is coming down from Victoria, BC, Canada for the contest. That’s a heck of a commute. Per his post at SparkFun, he’s looking for good lodging options in the Boulder area. If you have suggestions, let him know.
SparkFun added a spectator registration site to help predict how many people are coming to view the contest. If you are planning to attend, sign on up and put on a Team Hellhound T-shirt. Wait, we don’t have T-Shirts.
Anyway, let the good folks know if you are planning to attend. How many beers you plan on consuming, etc. How hard can it be? Wait, did they ask how many beers?
Anyway, come on out and support all the crazies.
Update: Fixed the borked registration link to point to the proper page.
Today, we added a couple of new Sponsors. Normally, sponsors provide some material or fiscal support. Not us. We flag them sponsors when some company, individual or organization goes out of their way to help us. Every bit of support is useful and these guys step up to the plate regularly.
Battery Buddies gave us some free product and significant technical advice. If you use batteries with Dean’s connectors, these simple additions make your life much simpler. Hopefully one day SparkFun will carry both Deans connectors and Battery Buddies caps.
Next up, the Colorado Rotor Heads has provided a great deal of RC knowledge and expertise. If you already fly RC Helicopters or would like too, this club rocks. The members rank from sheer novices, like Floyd and I, to top-notch pilots who make helicopters do things that seem impossible. Check them out if you are in the area.
Also, a new Cool Company has been added as well. Our competitor and soon to be friend over at Bluer White mentioned the general uselessness of double sided tape for mounting. If double sided tape mounting is a need, take a look at HeliProz’s offering for ZEAL Vibration Absorption tape. Our friend at Battery Buddies turned me onto it. While it seems pricey, the stuff is amazing: great adhesion and provides vibration isolation. I love the stuff. In fact, it sticks so well that it broke the housing loose from a EM-406 GPS receiver when I pulled it off the Traxxas today. Strange they attach a housing with crappy solder joint.
Managed to quickly get the new encoder installed on the Traxxas vehicle over lunch today. I only had my cell phone camera with me and the results indicate how well I can manage to take shots with it. I am completely unable to hold it still enough for close-up shots.
We managed to get the encoder mounted. Floyd constructed a few cables while I was fabricating the bracket. The GPS was also replaced on the Traxxas. Someday I’ll remember everything I need to check on a data sheet before ordering. No worries. Who cannot use a few extra bits?
Encoder is the fuzzy object left of the right wheel.
Floyd 'The Man' crimping pins.
Lots of software development of late. Floyd got a nice speed control loop working in the micro controller which alleviates fence post bouncing the in the future. He also updated the board to output all the sensor information. The end result is a lot of updates on my end that are rather dull but quite necessary.
The global waypoint follower is also behaving correctly now along with a minimalistic world model to profile speeds when obstacles are the in the path. No avoidance behavior yet but at least the robot is attempting to avoid destroying itself by slowing down. Someone once told me speed kills. Very apt for runaway robots.
Hopefully videos will arrive soon.
Bang! Fencepost wins.
Sometimes things go awry even when you suspect all your bases are covered. Nothing like buggy code, open loop speed set points and a narrow region with big buildings to occlude GPS. Wack! Too bad I didn’t have video it was pretty funny to watch.
Ah well, a little broken plastic and a snapped steering linkage. Linkage is fixed and little CA will hold the broken bits for a while anyway. Good thing we have multiple platforms. At this rate, most of them will be down for repairs. The good news is I found the bug but I’m going to have to find a larger place to test to avoid future destruction.
Repair Tools
