What COTS parts would you like to see?

[marshall]

Originally Posted by Mr V

AndyMark sells this bumper kit for the AM14U3 http://www.andymark.com/AM14U3-p/am-3152.htm Much of that was included in the version shipped as the KOP chassis, it included the right and left angle brackets, the thumb screws and the t nuts. There was also a separate instruction sheet on how to use those parts to create bumpers. It did not include the corner angles but personally I prefer to glue and screw the two pieces together and have found that method to be more than robust enough.

Not what I want. It's not simple enough (And yes, I know how simple it is). Andy knows what I want. He had a 3D printed version of it at champs 2 years ago. Someone at AM should tell him to make it already.

Originally Posted by hrench

a CAN plug-in vision aiming module.

Ain't gonna happen anytime soon for a lot of reasons. Unless of course all you want is the PIXY CAM with a CAN interface... you can make that happen without a whole lot of effort... use the HERO board to do it maybe?

All that being said, 900 is actively talking about how to "cheesecake" vision onto other robots in the future... no idea what that means at the moment but we're talking about it.

The biggest issue with vision isn't the vision or the detection. It's getting more teams to add encoders to their drivetrains and mechanisms. You need feedback to make use of the vision data.

I'd love to see more COTS options for encoder integration.

I would also love to see the CIM-Encoder that AM introduced this season improved. It's hollow and the little center spacer plate isn't square. Make it filled in, more robust, and square the center plate.

And I REALLY hope brushless is coming based on what CTRE was showing off. I'd love to see it happen.

[Rangel(kf7fdb)]

Originally Posted by marshall

Ain't gonna happen anytime soon for a lot of reasons. Unless of course all you want is the PIXY CAM with a CAN interface... you can make that happen without a whole lot of effort... use the HERO board to do it maybe?

All that being said, 900 is actively talking about how to "cheesecake" vision onto other robots in the future... no idea what that means at the moment but we're talking about it.

The biggest issue with vision isn't the vision or the detection. It's getting more teams to add encoders to their drivetrains and mechanisms. You need feedback to make use of the vision data.

I'd love to see more COTS options for encoder integration.

I would also love to see the CIM-Encoder that AM introduced this season improved. It's hollow and the little center spacer plate isn't square. Make it filled in, more robust, and square the center plate.

And I REALLY hope brushless is coming based on what CTRE was showing off. I'd love to see it happen.

Why couldn't a gyro be an alternative to having encoders on the robot? If the gyro was part of the board, it could likely still be kept all together. I'm not sure if we even used the encoders for our vision tracking this year and we hit every autonomous shot at champs. I'd imagine a game with a bigger goal would be even better.

Edit: Granted, we did use the encoders for going forward a set distance in teleop. This probably would effect our accuracy in autonomous. I would imagine not by too much though. In teleop though, auto tracking would be unaffected by not having encoders.

[Andrew Schreiber]

Originally Posted by Rangel(kf7fdb)

Why couldn't a gyro be an alternative to having encoders on the robot? If the gyro was part of the board, it could likely still be kept all together. I'm not sure if we even used the encoders for our vision tracking this year and we hit every autonomous shot at champs. I'd imagine a game with a bigger goal would be even better.

Edit: Granted, we did use the encoders for going forward a set distance in teleop. This probably would effect our accuracy in autonomous. I would imagine not by too much though. In teleop though, auto tracking would be unaffected by not having encoders.

Gyros tend to need to be level. You can cheat to an extent with some of the IMU boards doing a self leveling routine on startup and I guess that's probably a viable approach. Just don't forget it.

But why Encoders - At some point you need to get to a point on the field and that's pretty hard without encoders in the drive.

[GDG 2337]

FR6-ZZ HEX PS2 Radial Bearings at Boca Bearing, showing 4700 in stock

http://www.bocabearings.com/bearing-...12-FR6ZZHEXPS2

[asid61]

Originally Posted by GDG 2337

FR6-ZZ HEX PS2 Radial Bearings at Boca Bearing, showing 4700 in stock

http://www.bocabearings.com/bearing-...12-FR6ZZHEXPS2

And look at that- only $14 each! What a steal. /sarcasm
Seriously though, nice find. Good to know there's a source out there for 3/8" hex bearings in tough situations. However, I would still want a 7/8" OD 3/8" hex bearing available from Vex or AM.

[JesseK]

Originally Posted by Peter Johnson

We've used standard cut-to-length non-addressable 12V LED strips (Amazon sells lots of these in various colors) and this tiny board from pololu to control them: https://www.pololu.com/product/2802

Basically the polulu board is a simple digital switch (think of it like a spike without the ability to reverse the voltage polarity), controlled by a PWM signal. The linked one above is 3A but they also have higher current ones. The wiring diagram shows how to wire it up. Load is LED strip, source is either 5V or 12V (depending on LED strip voltage) from VRM. The PWM comes from a PWM output on the roborio. You do need 5V in from somewhere too; if you're using 5V LEDs or a PWM from a MXP board you can get 5V from that connection, or you can get it from the VRM or RoboRIO DIO. I think we used the Talon class in software to control the PWM output--set(0) turned the LEDs off, set(1) turned them on. Easy enough to do das blinkenlights from there with software timing.

Thanks for sharing! We just picked up a few of these to experiment with since our 2016 bot would have had at least a few uses for any kind of LED driver feedback, but we couldn't find any spikes in-house at the time.

Originally Posted by Greg Needel

So we built a prototype of a product about a year ago to do simple LED control called the "blinkin Board" ...

Cool! I will keep an eye out just in case something comes out.

[hrench]

I want a CAN plug-in vision aiming module

Originally Posted by marshall

Ain't gonna happen anytime soon for a lot of reasons. Unless of course all you want is the PIXY CAM with a CAN interface... you can make that happen without a whole lot of effort... use the HERO board to do it maybe?

All that being said, 900 is actively talking about how to "cheesecake" vision onto other robots in the future... no idea what that means at the moment but we're talking about it.

Yes, I guess I'm talking about 'cheesecake'. Making code and 'teaching' it in the time we have after a running robot has always been the impediment.

I want a module that has the programming built in to actually run the wheels and drive forward/back for the best distance. Or for a turret, to aim azimuth and elevation.

And it would ideally have a 'learning' mode where you'd manually make shots and it would recognize where the aim was to duplicate it.

Originally Posted by marshall

The biggest issue with vision isn't the vision or the detection. It's getting more teams to add encoders to their drivetrains and mechanisms. You need feedback to make use of the vision data.

Our robot has had a camera and encoders on it every year, necessary for any autonomous.

Originally Posted by marshall

I'd love to see more COTS options for encoder integration.

I would also love to see the CIM-Encoder that AM introduced this season improved. It's hollow and the little center spacer plate isn't square. Make it filled in, more robust, and square the center plate.

And I REALLY hope brushless is coming based on what CTRE was showing off. I'd love to see it happen.

Yes this would be great. Also a 'cheesecake' aiming system could have a camera/multiple cameras pointed at the carpet and read speed and distance from that. Maybe I'm getting a little too complicated.

On the flip side, if every team has automated aiming and shooting, it may make amazing robots but less fun of a game.