ÜberBots Hybrid Mode (Video)

[Uberbots]

So we finally got around to taking a couple videos of autonomous doing its thing... id say we are satisfied for now. It makes a full lap (sometimes 5 lines depending on the ball positions), and knocks off two balls.
consistently.

Video Here

If you scroll around in the playlist bar, you can find more videos/pictures of our robot.

See you at the CT Regional!

[Tom Bottiglieri]

How is it driven? All autonomous?

[Uberbots]

Originally Posted by Tom Bottiglieri

How is it driven? All autonomous?

There is a 'theoretical' all autonomous mode where it uses a custom made camera to locate the balls.

however, this is not implemented as of yet, and so probably will never be (in the spirit of tools/laptops down).
We tell it where the balls are with the IR remote, and basically push those positions onto a stack. As the robot makes laps, it pops these set positions off of the stack, and corrects to them using a closed loop driving system.

Trial A Demonstrates its ability to hit balls in other positions.

[Dave Scheck]

First off, nice work. It's definitely not the easiest thing in the world to drive that far with that kind of accuracy, so I commend you for your hard work.

I don't want to stir up a controversy, but I would like to know what people think about this topic. Billy explained the functionality of their robot as:

Originally Posted by Uberbots

We tell it where the balls are with the IR remote, and basically push those positions onto a stack. As the robot makes laps, it pops these set positions off of the stack, and corrects to them using a closed loop driving system.

My question is simply is this legal? My answer is I don't know because I see both sides.

On one hand, I would say this is legal because the action taken when one of three buttons are pressed, the ball location corresponding to that button would be pushed onto a stack. That has the same repeatable outcome every time...a button is pressed, a position goes on the stack. How the robot interprets this is irrelevant.

On the other, I would say this is illegal because based on what I've read in Update 3, an action is something that the robot physically does (I don't agree with this, but it is what it is). For the sake of argument, there are 9 possible actions that are taken (i.e. left then left, left then middle, left then right, etc...). This is clearly more than the 4 allowed actions.

Originally Posted by Team Update 3

What can be contained in the message(s) sent to the Robot? The message should contain a single command that directs the Robot to do something, or provides a single “snapshot” of information about the state or condition of the field that the Robot can use to determine what action to take.

If your buttons told the robot where both balls were at the same time, I would agree that this represents "a single snapshot of information about the state or condition of the field". Queuing up snapshots, I fell, is not a single snapshot because the same button has changed its meaning to represent a different state.

I also see issues with this being a message sequence because the robot's actions are dependent on the order that you press the buttons in. The update clearly says that message sequencing/encoding is not allowed

Originally Posted by Team Update 3

The same would also be true if pressing Button #6 and then Button #7 initiated one action, but pressing Button #6 and then Button #8 initiated a different action. This would be a form of high-level encoding, and is contrary to the intended use of the Signaling Device.

Please don't take this as an attack towards the work that you've done. I definitely appreciate the work that you've put in to make this system actually work. In light of some of the recent rulings at competitions, I think that you should put a Q&A in so that you don't run into problems and have to back out all of your hard work at the competition.

[Manoel]

Originally Posted by Dave Scheck

My question is simply is this legal? My answer is I don't know because I see both sides.

Dave,

I had the same doubt a few weeks ago (before we realized our auto mode was great for the first ball but didn't leave time to go for the second ) and asked this question on the Q&A forums.

I knew the intent of the rules allowed it, because high-level suppervisory commands are exactly what the GDC wanted us to do. The letter of the rules, as we agree, seem to disallow this.

Even the GDC answer is not mentioning the letter of the rules but, based on the link I posted above, they should be fine if questioned at an event.

By the way, 1124, we really hope to play with you guys next week in Connecticut - will you please let us knock off that first trackball?

[Dave Scheck]

Manoel

The way I read the Q&A response that you're referring to is that it's legal to do what you're described after the first ball was knocked down. I completely agree with that. It's a single action "knock down the ball from the next overpass" and could be rerun just as a "turn left" command could be.

If that's what the Uberbots are doing, then I think that answers my question from above.

However, if they are sending the two commands before that first ball is knocked down, I think my question still stands.

Dave

[Tom Bottiglieri]

Originally Posted by Dave Scheck

Manoel

The way I read the Q&A response that you're referring to is that it's legal to do what you're described after the first ball was knocked down. I completely agree with that. It's a single action "knock down the ball from the next overpass" and could be rerun just as a "turn left" command could be.

If that's what the Uberbots are doing, then I think that answers my question from above.

However, if they are sending the two commands before that first ball is knocked down, I think my question still stands.

Dave

Easy fix:
Press the second button after the first ball is knocked down.
We are debating whether we should use this approach or if we should simply have the RoboCoach drive the robot. I've seen success with both.

[Aren_Hill]

Our first approach was to simply push a button according to which position the ball was in. That was inconsistent and had like a 30-40% chance of hitting the ball off. For the eliminations we just switched to the robocoach driving the robot forward backward right left and its much nicer

[Uberbots]

Originally Posted by Aren_Hill

Our first approach was to simply push a button according to which position the ball was in. That was inconsistent and had like a 30-40% chance of hitting the ball off. For the eliminations we just switched to the robocoach driving the robot forward backward right left and its much nicer

how do you drive it while still remaining within the rules?

[Dave Scheck]

Originally Posted by Tom Bottiglieri

Easy fix:
Press the second button after the first ball is knocked down.
We are debating whether we should use this approach or if we should simply have the RoboCoach drive the robot. I've seen success with both.

Agreed, providing that
1. The stack approach is deemed illegal (which really is the heart of my question)
2. The IR sensor is positioned such that it can be sent a signal from the robocoach station while the robot is downfield.
3. The IR system works under the hostile IR conditions that can be at the event.

Originally Posted by Aren_Hill

we just switched to the robocoach driving the robot forward backward right left and its much nicer

What kind of range/reliability did you have with that approach? I'm assuming that you had commands like "Drive forward for N loops" and repeatedly had to press the button.

[Uberbots]

I dont take any of this as an attack to my work, because i am completely aware that it is borderline legal/illegal. The reason we have it like this, though, is that we can justify that it does the same action every time, which is store a specific number to a variable, push that variable onto the stack, and then wait a half a second so that the robocoach doesn't inadvertently set invoke another press due to the nature of the IR signal.

We do realize the potential problems, and have backup code that does exactly the same thing, its just a bit riskier. the theory of this mode is the coach has 4 buttons, a1 a3, b1 b3. a and b are the different sides of the fields, and the numbers are the positions of the balls. the robot defaults to a2 b2, until the robocoach sends it a command that alters its default.

[smurfgirl]

Originally Posted by Manoel

By the way, 1124, we really hope to play with you guys next week in Connecticut - will you please let us knock off that first trackball?

Of course! Sharing is great.

I can't wait to see all the neat stuff other teams have to offer at the CT Regional.

[Uberbots]

we did a test today to show what happens when there are robots in our way.
its like they werent even there to begin with... it plows through them. the avoidance algorithms would work if we could position the sonar where it didnt see our own robot.

[seanl]

how are you detecting where your colored ball is? because the opposite team will get the points if you knock down their ball.
are you using the ir sensor.

[Qbranch]

Why is this robot 'almost a year old' according to the description?

Looking fancy shmancy guys... can't wait to hear some more details.

-q