Autonomous Sensing

[Pavan Dave]

What kind of sensors does your team use to make their autonomous mode work? Do you use gear tooth sensors, potentiometers, limit switches, sonar, or something? How do you use these sensors? Do you run across (or think you might run across) any problems if certain variables (starting location, etc.) don't work out for you?

Answers can be from the past as well as present.

Thanks.

[joshyboy9987]

We don't have anything working so far, but we've been working on some code to use the accelerometers and gyro that come in the kit. They can give you both linear acceleration (in two directions) and rotational acceleration. By doing some Calculus, you can use them to get distance pretty simply.

Also, for obstacle evasion we were thinking about a ultrasonic rangefinder. Pretty much any rangefinder that doesn't use light, because almost all the boundaries are glass!

[ubergeek5075]

Well, last year we used Optical encoders on the drivetrain, and were extremely successful. We knew within 0.15" where we were. Also, because of this, we created our own force-feedback system to keep ourselves in place when other robots were playing defense on us.

On last year's arm manipulator, we used the usual potentiometers for positioning data.

We've been working with gyros and accelerometers since 2006, but haven't found use for much of the code I wrote that year, since we replaced it last year with encoders, the camera, etc. However, this year with the lack of the need for a camera, we may be using these again.

Also, last year we started experiencing with ultrasonic range finders, specifically the EZ1 from maxbotix. We didn't end up using them last year, but we can tell within an inch, between 6" and 20+ feet. However, they may be our only eyes this year if we can get the logic working. Granted we found 2 problems: that the FOV of ultrasonic sensors was too large to "scan" area, and that they only scan at 20Hz.

[Stuart]

Ping ))) x 3
possibly 4 shaft encoders(hmm why 4 and not 2 Pavan)
1 x gyro


and to those of you counting yes that is 8 sensors . . wonder how thats going to work 8)

[popo308]

Use encoders on your drive wheels to sense how far you have gone.
depending on how you build your robot you want sensors on your arm or whatever your manipulator is obviously...

[Guy Davidson]

Our plan for this year includes quadrature optical encoders, a gyro, absolute magnetic encoders, rangefinders (either IR or Ultrasonic, but of the vareity that returns an analog value and not that you have to send a high to ping and you have to use an interrupt to receive), and possible an accelerometer. This is going to be a blast I have some code written for all of them if you want to see.

As for starting location not working on anything like that, I haven't run into any such problems yet. Are there any specific scenarios you have in mind?

[Donut]

Originally Posted by joshyboy9987

We don't have anything working so far, but we've been working on some code to use the accelerometers and gyro that come in the kit. They can give you both linear acceleration (in two directions) and rotational acceleration. By doing some Calculus, you can use them to get distance pretty simply.

Prior to last year's competition I messed around with a Vex robot and the 2006 kit accelerometer and gyro. The gyro works fine for angles, and I have yet to have a bad experience using gyros for direction sensing in 3 years using them. However, I don't believe accelerometers (or at least the kit ones) are that great for linear distance measuring.

With the Vex robot, the accelerometer is absurdly sensitive to noise (with the kit accelerometer, touching the robot with my finger was enough to make it angry), and you get innaccurate for distances extremely fast. It may be better with FRC, and if you could get a more reliable accelerometer or a better ADC with more resolution it might be possible, but I would not recommend using the kit accelerometer for distances.

Now for the main thread:

Our team has used pots and limit switches forever, for limiting motion and setting exact positions on arms and whatnot (the most fun of these being an elevator in 2005, where we used a set of switches and a plate that would slide over them to sense 5 different heights). The gyro has started getting used recently on my team and in my opinion is the most beneficial sensor you can use; we use it for turning in autonomous and for driving straight in both autonomous and manual mode (your drivers will thank you!). We installed an accelerometer for tip prevention last year and had already tested and proven code, but never got to use it since we never got to calibrate for that specific sensor. Banner sensors are pretty effective for line detection and can be used as a sensor beam like a garage door with a reflector (I think these used to come in the kit pre-2004).

This year will be a gyro, pots, and limit switches as usual. I think this year's programmers want to try encoders for distance and then a line sensor for detecting the finish line (nice way to tell when you're right under the overpass). I've suggested uses for sonar but with a smaller programming team than normal I think they only want to try one new sensor and stick with encoders.

[anonymoose]

Our Team used ultra sonic and our camera for our automous mode. i don't know much more info then that cause i am a rookie.

[Darunada]

We were playing with the accelerometer today,

Holding the accelerometer in the air and moving it gave us a nice smooth reading. Dragging the accelerometer on the table, it was very choppy and probably not reliable if the robot does any amount of... touching...

[mwtidd]

were planning on using the cmu cam,

I was also interested in using a rangefinder,
if anyone has gotten one of these to work,
and is willing to share that portion of their code
it would be greatly appreciated.

[mluckham]

Holding the accelerometer in the air and moving it gave us a nice smooth reading. Dragging the accelerometer on the table, it was very choppy and probably not reliable if the robot does any amount of... touching...

Gentle Bumping

[Phalanx]

We have in the past few years used:

CMUCam
Gear Tooth Sensors or Encoders
Yaw/Rate Gyro
Maxbotic EZ-1 Sonic Range Finder

So far our tentative plans are to use the following:
*disclaimer This is subject to change at anytime without notice*

KOP Gear Tooth Sensors to measure the distance traveled.
We don't see the need for quadrature encoders this season.

300 degree/sec Yaw Rate/Gyro instead of the KOP one to help keep us straight and to turn left.

IR Sensors to select which position the ball is to knock down, or to simply race.

Undoubtedly we'll use limit switches, potentiometers, or encoders for our arm/manipulator but we haven't finalized that design yet.