Sparkfun Electronics held their 6th annual Autonomous Vehicle Competition last weekend, and this year was bigger than ever before. The action was at Boulder Reservoir in Colorado, but anyone could follow along (with a few technical difficulties) on the YouTube LiveStream. (Part 1), and (Part 2).
The story of the day was Team SHARC’s Troubled Child, which won the ground vehicle doping class. Rather than mess around with miniature cars, Team SHARC built their ‘bot out of a freaking Jeep, a 1986 Jeep Grand Wagoneer to be exact. Troubled Child had no problem getting around the course. One could say it carried the entire team. Literally – the rest of Team SHARC’s robots are riding along on top of Troubled Child in the picture up there.
There was also plenty of action in the aerial competition. Sir Crash-a-Lot was the first drone to find a watery doom at Boulder Reservoir. The last we saw of it on the stream, the team was looking for some divers.
Aircraft can not be hand launched at the AVC. Not a problem for rotary-winged vehicles, but this rule has led to some interesting solutions for fixed wing aircraft. The disguised “Team Falcon” showed up with an incredible compressed air launcher, which used a gallon water jug to fire their delta-winged plane to a clean run. Team Karma550 wasn’t quite as lucky, with their helicopter crashing hard, and throwing up quite a bit of smoke.
We’re still waiting for more detailed results, but if you want the full scores, they are available on Sparkfun’s AVC scoreboard page.
Registration is open for Sparkfun’s 2014 Autonomous Vehicle Competition (AVC)! Every year the fine folks at Sparkfun invite people to bring their robots, rovers, and drones to Colorado to see who is the king of the hill – or reservoir as the case may be. We see plenty of robots here at Hackaday, but precious few of them are autonomous. To us that means capable of completing complex tasks without human intervention. Sparkfun has spent the last five years working toward changing that. Each year the robots get more complex and complete increasingly difficult tasks.
The competition is essentially a race through the Boulder reservoir. Time is key, though there are multiple ways to gain bonus points. For aerial vehicles there are two classes: fixed and rotary wing. Planes fall under the fixed wing category. Helicopters, gyrocopters, tricopters, quadcopters, and beyond fall into rotary wing. We’re holding out hope that e-volo shows up with their Octadecacopter. Ground vehicles have a few more class options. Micro/PBR class is for robots with a build cost less than $350 total, or small enough to fit into box that’s 10″x6″x4″. The doping class is unlimited. Sparkfun even mentions costs over $1kUSD+, and weights over 25LBS. Non-Traditional Locomotion class is for walkers, WildCats and the like. Peloton is Sparkfun’s class for robots that don’t fit into the other classes.
Sparkfun is also making a few changes to the course this year. A white chalk line will be drawn through the course, so robots don’t have to rely on GPS alone for navigation. We’re hoping to see at least a few vision systems using that chalk line. Aerial robots will have to contend with three “Red Balloons of Death”. Robots can navigate around the balloons without penalty. The balloons can be bumped or even popped for bonus points, but the robot must do this with its own body. Projectile weapons are not allowed. To say we’re excited about the AVC would be an understatement. As much as we enjoy watching the big players at competitions like the DARPA Robotics Challenge, we love seeing individuals and small teams of hobbyists compete every year at the AVC. Click on past the break for Sparkfun’s AVC 2013 wrap up video.
Continue reading “Sparkfun’s AVC 2014: Robots, Copters, and Red Balloons of Death, Oh My!”
[Tim] is getting his drone ready for SparkFun’s 2013 Autonomous Vehicle Competition on June 8th. He has a pretty good start, but was having some problems accurately measuring travel distance. The technique he chose for the task was to glue magnets onto the axles of the vehicle and monitor them with a hall effect sensor. Those sensors are finicky and a few problems during testing prompted him to look at a redundant system. Right now he’s experimenting with adding an optical mouse sensor to the autonomous vehicle.
Recently we saw the same concept used, but it was meant for tracking movement of a full-sized automobile. If it can work in that application it should be perfect here since the vehicle is much closer to the ground and will be used in ideal conditions (flat pavement with clear weather). [Tim] cracked open an old HP mouse he had lying around. Inside he found an Avago ADNS-5020 sensor. After grabbing the datasheet he discovered that it’s simply an I2C device. Above you can see the Arduino Leonardo he used for the first tests.
[Tim] coded functions to monitor the chip, including some interesting ones like measuring how in-focus the surface below the sensor is. This brings up a question, is there limit on how fast the vehicle can travel before the sensor fails to report back accurately?
Team 0x27 was the winner of this year’s AVC, the Autonomous Vehicle Contest put on by SparkFun Electronics. You’ll find video of the two runs from this entry (the third run did not finish). We love it that there’s an on-board camera recording both video and sound of the race from the vehicle’s point of view. They haven’t updated their team page yet but we’re sure they will once their done celebrating.
On the first run the team opted not to use obstacle avoidance, and here you can see it annihilating one of the barrels from the course (this is the second one it took out with hulk-like rage). These collisions didn’t keep it from finishing the circuit. On the second run it didn’t slam into anything. Because of the hoop-deduction (a bonus for threading the needle during the run) the official time came in at 2.08 seconds. Still, the unadjusted time of 32.08 seconds is a course record and beat the fastest finisher from the airborne group of compeititors. Nice.
Seriously, this video just cracks us up!
Continue reading “AVC from the vehicle’s perspective”
[Paul Breed] participated in this year’s Autonomous Vehicles Contest put on by SparkFun Electronics. As with most projects, the deadline really snuck up on him and he ended up cramming a bunch of code development into the waning days before the competition. His experiences are shared in this recent blog post.
One big part of the hardware is a laser range finder used for wall following. This is explained well in the video after the break, but you can see the side-pointing blue box in the image above. [Paul] also spent a lot of time preparing for the checkpoint portion of the course where the vehicle would need to pass through a red hoop. He worked long and hard on an image processing setup to find and navigate those hoops before learning that they would be positioned at known locations and it would be much easier to use a path following technique to complete the challenge.
He had a few follies along the way. At one point during debugging the car — which was connected to his laptop via Ethernet — it got away from him. This ripped the NIC right out of the back of his computer. And on the day of the event he had some low battery issues that zapped his laser calibrations. But [Paul] rolled with the punches and ended having what sounds like a really exciting experience participating in the contest.
Continue reading “Cramming for Sparkfun’s Autonomous Vehicle Competition”