Decapitron!
Decapitron is an autonomous fighting robot. It won second place at Robogames this year. Decapitron's brain is a Gadget Gangster Propeller Platfom USB, with a servo shield and some custom wiring.
For sensors this robot uses two colorPALs, a Ping sensor and an IR sensor on top to detect if the robot is flipped over. It can continue to operate even if it is flipped. It also has a radio controlled safety relay. Decapitorn's armor is aluminum on top and bottom, with a steel sheet to protect the front. For weapons it has two teeth cut out of chrome-moly steel that are powered by two 400 ounce/ inch servos. The advantage of using the prop is that I can take reading continuously from all sensors while controlling the motors and the teeth. Since each sensor is on a cog, if something goes wrong with a sensor the main loop is not interrupted. It just stalls out the cog that polls that input in a way that the main loop recognizes. In robot combat sensor failure is not a theoretical issue.
Decapitron is mostly a defensive robot that uses its teeth to keep its enemy in front and push them around. You are not allowed to pin for more than five seconds in robot combat, but usually it could quickly reacquire the target with sonar. On the first day of combat I had a stupid problem where I did not tune the light sensors to be sensitive enough and it missed the red stripe of the arena twice. Fortunately one time when the sensors failed my opponent had already been counted out and it was double elimination. I had calibrated them the ColorPALs the day before, but not carefully enough.
On the second day I won until I faced Carrapato from Brazil in the finals. In this match a hit caused the little AAA battery holder for receiver to open and spill the batteries, deactivating (decapitating?) the robot. If I had just screwed the battery case shut instead of snapping it-- or even taped it-- I think I could have won. I am not at all damaged and ready for next year. I think there is an autonomous division in a robot competition in Ohio this year. I can mail the robot to someone if they want to enter it.
I heard that a picture of Decapitron may be in Servo magazine. Here are some pictures of the outside, the inside (check out the guts!) and some movies of it in action.
![DSC00501.jpg](https://forums.parallax.com/uploads/attachments/50326/90448.jpg)
![guts.jpg](https://forums.parallax.com/uploads/attachments/50326/90449.jpg)
![DSC00511.jpg](https://forums.parallax.com/uploads/attachments/50326/90450.jpg)
![decapitron1.jpg](https://forums.parallax.com/uploads/attachments/50326/90451.jpg)
http://youtu.be/7ccDh-L6daA
http://youtu.be/8stBXv4GP3c
For sensors this robot uses two colorPALs, a Ping sensor and an IR sensor on top to detect if the robot is flipped over. It can continue to operate even if it is flipped. It also has a radio controlled safety relay. Decapitorn's armor is aluminum on top and bottom, with a steel sheet to protect the front. For weapons it has two teeth cut out of chrome-moly steel that are powered by two 400 ounce/ inch servos. The advantage of using the prop is that I can take reading continuously from all sensors while controlling the motors and the teeth. Since each sensor is on a cog, if something goes wrong with a sensor the main loop is not interrupted. It just stalls out the cog that polls that input in a way that the main loop recognizes. In robot combat sensor failure is not a theoretical issue.
Decapitron is mostly a defensive robot that uses its teeth to keep its enemy in front and push them around. You are not allowed to pin for more than five seconds in robot combat, but usually it could quickly reacquire the target with sonar. On the first day of combat I had a stupid problem where I did not tune the light sensors to be sensitive enough and it missed the red stripe of the arena twice. Fortunately one time when the sensors failed my opponent had already been counted out and it was double elimination. I had calibrated them the ColorPALs the day before, but not carefully enough.
On the second day I won until I faced Carrapato from Brazil in the finals. In this match a hit caused the little AAA battery holder for receiver to open and spill the batteries, deactivating (decapitating?) the robot. If I had just screwed the battery case shut instead of snapping it-- or even taped it-- I think I could have won. I am not at all damaged and ready for next year. I think there is an autonomous division in a robot competition in Ohio this year. I can mail the robot to someone if they want to enter it.
I heard that a picture of Decapitron may be in Servo magazine. Here are some pictures of the outside, the inside (check out the guts!) and some movies of it in action.
![DSC00501.jpg](https://forums.parallax.com/uploads/attachments/50326/90448.jpg)
![guts.jpg](https://forums.parallax.com/uploads/attachments/50326/90449.jpg)
![DSC00511.jpg](https://forums.parallax.com/uploads/attachments/50326/90450.jpg)
![decapitron1.jpg](https://forums.parallax.com/uploads/attachments/50326/90451.jpg)
http://youtu.be/7ccDh-L6daA
http://youtu.be/8stBXv4GP3c
Comments
The most common thing you fail for in autonomous combat is your robot gets stuck. You can't stay still for 10 seconds, or be stuck in the same place. It's a little challenging to keep a robot running for two minutes without getting stuck. It's more challenging if there is another unpredictable robot in the same space. Still more challenging if the robot is trying to bump into you and harm you. The Peanut Tin of Terror did pretty well last year because it didn't get stuck. But it couldn't take a hit. The electronics weren't even fastened down, they were just stuck in a can. When Toro+ hit it with it's drum the robot flipped and the electronics spilled out.
http://www.youtube.com/watch?v=TxE3z9ezQjM
Autonomous combat is interesting to me because there have only been people trying it for a couple years now. This time there were about nine entries and half of them barely worked at all, including me on the first day. There is so much room for improvement in the robots. It's still at the point where people are trying just to make the robot work reliably. When most people clear that hurdle I think the strategies will really develop.
Well, perhaps it lost, but the name is a winner!
Amanda