I met Tom and Chris at the AVC in 2014 and enjoyed meeting them. I think the issue is the rover code is designed to work with GPS, and the SparkFun building isn't a great place for GPS. Mountains and the building make it hard. I'll be back next year and hope to complete all my runs for all my rovers. My rover was the one that pushed the hoop around. It was the highlight of the even if you ask me. :)
I won heavyweight, and got 3rd in welterweight and had the highest score. ;) I can tell you all the bots that won only used dead reckoning with NO gps. Also, all 3 winners are using their own custom code, not 3dr/diydrones code. The most impressive bot I saw was USU Racecar. They had a scanning lidar, and it really seemed to work. It was a class project they built in only 10 weeks. There was another team using a neural network to drive their bot. They managed to complete once during testing, but not during competition. If you have any more questions feel free to ask.
Presumably they're manely using machine vision with some using dead reckoning & GPS. The real value is knowing who the best were & what algorithm they used. They only show who the best were.
Comments
I met Tom and Chris at the AVC in 2014 and enjoyed meeting them. I think the issue is the rover code is designed to work with GPS, and the SparkFun building isn't a great place for GPS. Mountains and the building make it hard. I'll be back next year and hope to complete all my runs for all my rovers. My rover was the one that pushed the hoop around. It was the highlight of the even if you ask me. :)
Congrats Jesse!
We should encourage Tom to visit AVC again in 2017 :)
Though the rover code truly would need some love to compete with tough competititors.
I pretty like the fact that they made it harder :)
No problem. Just trying to clarify the brand identity. Many people assume that 3DR and/or DIYDrones owns Ardupilot, and that is not correct.
I apologize! I seem to be unable to edit my post to correct to ArduPilot/APM:Rover
There is no such thing as 3dr/diydrones code.
I won heavyweight, and got 3rd in welterweight and had the highest score. ;) I can tell you all the bots that won only used dead reckoning with NO gps. Also, all 3 winners are using their own custom code, not 3dr/diydrones code. The most impressive bot I saw was USU Racecar. They had a scanning lidar, and it really seemed to work. It was a class project they built in only 10 weeks. There was another team using a neural network to drive their bot. They managed to complete once during testing, but not during competition. If you have any more questions feel free to ask.
Presumably they're manely using machine vision with some using dead reckoning & GPS. The real value is knowing who the best were & what algorithm they used. They only show who the best were.
Cool. I see lots of Pixhawks on the vehicles. Any of them running PX4 stack, or were they all Ardupilot?