What sensors do your teams use

[snekiam]

We used the CTRE magnetic encoders on our shooter to measure velocity, andymark's string potentiometer to measure position on our arms, these grayhill encoders on our drivetrain to measure velocity for (theoretical) automatic shifting, and a navx on our drivetrain for straight driving. I'm sure there are one or two sensors I am forgetting, but all of these are in closed loop P, PD, or PID control schemes. If you have any questions about how any of this works, let me know.

[ILAMtitan]

I gotta give a shout out to the humble potentiometer. We picked up a handful last year, and they proved super usefull this year too. They wire directly to an analog in Port, and can essentially give you abaolute position of a mechanism. Getting some with more than 300deg of rotation help protect it in case of over rotation.

Here's two versions I recommend keeping around:
Continuous rotation: http://www.digikey.com/product-detai...-103-ND/274005
10turn: http://www.digikey.com/product-detai...03L-ND/2058386

[Steven Smith]

One other question for you guys/gals as you discuss your sensor needs:

Are there any broad areas of sensing technology, how to do sensor integration (mechanical, electrical, programming), etc that you feel your teams could use help with?

AKA, in addition to WHAT to use, do you feel there are adequate resources on HOW/WHY to use sensors, that could be addressed by linking existing content or creating new content specific to FRC?

[DaveL]

I too would like to help my team improve their use of encoders.

I'm most interested in encoders that could be used for flywheel shooters that don't require any machining to install. In 2012, I remember teams used light emitting sensors to count light and dark marks on a spinning disk.

Are these still viable?
If so, which ones work well (high reliability, easy installation, non-contacting) and are reasonably priced?

Dave
Build Mentor 2976

[ILAMtitan]

Originally Posted by DaveL

I too would like to help my team improve their use of encoders.

I'm most interested in encoders that could be used for flywheel shooters that don't require any machining to install. In 2012, I remember teams used light emitting sensors to count light and dark marks on a spinning disk.

Are these still viable?
If so, which ones work well (high reliability, easy installation, non-contacting) and are reasonably priced?

Dave
Build Mentor 2976

Dave,
What you're referring to is a simple IR beam break sensor (example: https://www.adafruit.com/product/2168). One of the modules is just an IR LED, and the other is a receiver that tells you if it sees the LED or not. I don't see any problem with the electronics not being reliable, but you would have to manufacture you're own spinning disk to serve as an interrupter. Those adafruit ones seem pretty straightforward to mount, but I'm sure there are some better ones out there that come in a single module.

While I think mechanically attaching a proper high speed encoder would be the right solution for a flywheel shooter, these are still pretty useful. You can use them as limit switches and even "linear encoders" with the right interrupter.

[GeeTwo]

If you are spinning it fast enough (quite likely), you can put the light/dark pattern directly onto the disk, and use a reflective optical sensor. If you do this, use a relatively inexpensive sensor and be ready to replace it, because it'll be located right next to that spinning wheel, and FRC robots have been known to collide with one another or the field elements. On occasion.

[thatprogrammer]

Originally Posted by GeeTwo

If you are spinning it fast enough (quite likely), you can put the light/dark pattern directly onto the disk, and use a reflective optical sensor. If you do this, use a relatively inexpensive sensor and be ready to replace it, because it'll be located right next to that spinning wheel, and FRC robots have been known to collide with one another or the field elements. On occasion.

You can also use a $10 hall effect sensor to do the same thing with a magnet. We found this pretty effective this year.

[ASD20]

Originally Posted by DaveL

I'm most interested in encoders that could be used for flywheel shooters that don't require any machining to install. In 2012, I remember teams used light emitting sensors to count light and dark marks on a spinning disk.

If you want easy to install, I would recommend encoders that are designed to fit directly on a motor or gearbox. Vex sells a VersaPlanetary encoder, which is easy to assemble and fits right into a VersaPlanetary gearbox just like any other stage. This is what I would recommend for a shooter encoder. Andymark also now sells there PG gearmotors with encoders built in. The only assembly required is plugging in a wire. They also sell a CIMcoder now, which is designed to fit right on the shaft of a CIM. I have not used this before, so I can not speak to how well it works or how easy it is to install, but I imagine it is pretty straightforward. Andymark also sells an encoder that fits directly onto most of their gearboxes. It is kind of a pain to install and my team has broken way too many of them, but it is probably the easiest way to get an encoder on your drive train.

All of these require no machining and interface nicely with other COTS components.

[ollien]

Originally Posted by ASD20

They also sell a CIMcoder now, which is designed to fit right on the shaft of a CIM. I have not used this before, so I can not speak to how well it works or how easy it is to install, but I imagine it is pretty straightforward.

I cannot even think of recommending the CIMCoder. We had four of our drivetrain and by the time we reached competition, only two were reporting any information. I've also heard some horror stories of the shell cracking on high vibration applications like a shooter.

YMMV

[marshall]

Originally Posted by ollien

I cannot even think of recommending the CIMCoder. We had four of our drivetrain and by the time we reached competition, only two were reporting any information. I've also heard some horror stories of the shell cracking on high vibration applications like a shooter.

YMMV

Agreed. They need to be solidly filled and not a shell. Additionally, it would be nice if the little black square was actually a square and not a rectangle. Cranking down on it with the wrong orientation will crack the housing. It has a lot of potential but it needs some revisions.

[electroken]

Originally Posted by DaveL

I too would like to help my team improve their use of encoders.

I'm most interested in encoders that could be used for flywheel shooters that don't require any machining to install. In 2012, I remember teams used light emitting sensors to count light and dark marks on a spinning disk.

Are these still viable?
If so, which ones work well (high reliability, easy installation, non-contacting) and are reasonably priced?

Dave
Build Mentor 2976

For wheeled shooter speed measurement we've had good luck with this gear tooth sensor from Allegro Microsystems:

http://www.allegromicro.com/en/Produ...Cs/ATS667.aspx

As long as you can get the face of the sensor close enough to the teeth of a ferrous metal gear, you'll see one pulse per tooth that passes. They're around $7 each.

[ttnn58]

gyro&accelometer - for PID and stuff i think...
microswitch - to know if the ball is secured inside the robot
magnetic door sensor - to know if the scaling arm lifts the robots high enough for the scale to count

[David Lame]

We used the built in accelerometer of the roborio as a tilt sensor.

This year, in autonomous, we had a problem. Our robot was top heavy. If we tried to get over obstacles in autonomous, if we went too fast, we tipped. If we went too slow, we got stuck. The tilt sensor was our salvation. You can see it in action in this youtube video.



https://www.youtube.com/watch?v=roVc581joiA

As 247 goes over the moat, the announcer indicates that we look stuck. What you see on the film is that we drift back to the back side of the moat, and then suddenly charge forward.

Well, the drift isn't a drift. We programmed it to wait two seconds and check the tilt. If we weren't on flat ground, we would go slowly back, and then gun it full speed to finish the job.

It worked quite nicely. We also used it as an anti-tip sensor. If the tilt got too high, we cut wheel speed or moved in reverse to prevent tipping. That one didn't always work, as seen in the same video at about 1:10.

[gerthworm]

Originally Posted by Steven Smith

AKA, in addition to WHAT to use, do you feel there are adequate resources on HOW/WHY to use sensors, that could be addressed by linking existing content or creating new content specific to FRC?

I've seen a number of resources online of various qualities, but it would be nice to see a single published whitepaper that describes how to design a robot sensing system from top to bottom.

Challenges with such a document:

• Keeping it up to date with changing technologies and parts availability.
• Satisfying a large range of audiences.

I'd expect some mentors or students don't want to think about sensing system design much, and just need a quick answer as to "what should I use for situation XYZ". Others will have this design experience, and just aren't familiar with what's commonly used for FRC...

Perhaps a community-maintained wiki for this (and other topics) is an option? I haven't seen one yet, although I haven't looked very hard either...

EDIT: After a brief amount of searching, I came across https://firstwiki.github.io/wiki/tech , but it looks like there is plenty of work to be done still.

[marshall]

We use students as limit switches... the robot runs into them and they let us know about it. Decent feedback but a little slow and if they get hit too hard then they stop working sometimes. The good thing is we seem to have a never ending supply of them for replacements.