Sensors
 

Can you guys tell me some useful sensors to have?

Re: Sensors
 

Potentiometers and encoders can tell you distance and angle, depending on how they are used and assumptions that are made.
Limit switches tell you if you have (or better yet, are just about to) hit something.
Ultrasonic sensors do the same thing as limit switches, except that they also tell you how far it is before you hit it. (There are other range-finding sensors but those can cost a pretty penny.)
Light sensors can be useful for line following.
Cameras help you sight on a target.

Care to elaborate on a potential application of said sensors, or is this more of a "We are about to lose money from this year, what should we stock up on?" type of question?

Re: Sensors
 

Well I was wondering specific sensors that are better than others, and I am somewhat worried about others spending OUR robotics budget. Our 'providers of money' told our team that we had an unlimited budget, so everyone else in the building are purchasing outrageous amounts of stuff on things we don't need and we don't want and nobody will use. Someone bought 500 dollars worth of bee bots, and they were only used once, to drive around with markers attached to draw on our floors, also someone bought 6 new macs even through we have 2 computer rooms filled with 20 perfectly fine PCs and we have a few imacs, most of the macs have not even been turned on yet after 2 weeks. And we haven't even gotten anything since we held a competition a few months back. So I guess I am kind of worried about our budget being blown.

Re: Sensors
 

The best sensor is one you have experience with.

Get a handful of encoders from USDigital and play around with them as a team. Learn how they work, and make sure you can make them work. That way, when the season comes, you'll be able to pick out the correct encoder for the correct job.

Do the same for potentiometers. Single turn and multiple turn, and as a team figure out what applications you would use each for.

For limit switches, get a few different sizes. Some applications, you might want something large (like sensing the ball in this year's game - the huge ones from the KoP were great for that). Other applications, you might want something smaller (like the smaller KoP ones). For other applications, you might want something even smaller (think minibot from 2011). Learning the best way to mount these and their performance characteristics can help the team during the build season.

Other than that, I would simply look at previous year's games, and see what sort of sensors were popular on robots. Get some of them and start playing!

Re: Sensors
 

Light sensors, depending on the type, can also be used to count game pieces, regulate shooter feed mechanisms, detect mechanism position, and act as encoders. Passive light sensors should usually be avoided unless you intend to provide it with a light source since ambient light can vary greatly from venue to venue.

Re: Sensors
 

Gyroscope - This will make your autonomous routines 100000x more accurate. Your robot will be able to carry out quick and accurate turns despite drivetrain inconsistencies and battery voltage. It may also be used to keep a certain heading whille driving in a line.

Any sensor for line following (dependant on field of course) - Using a line-following technique is one of the simplest ways to achieve a consistent autonomous routine.
-My team has used these sensors for arm limits as well.

Encoder - Besides autonomous consistency, encoders open up proportional control (arm locking & arm presets) and digital limits.

Potentiometer - Similar to encoders, yet absolute. Encoders will read "0" on startup, while potentiometers will always read values from their current position. Potentiometers have physical limits as well. (not for drivetrain)

Dipswitch - Dipswitches on a robot may be used to set autonomous routines, delays, and other robot settings. Your team may have these capabilities built into its dashboard.

Kinect - This was the most useful sensor my team has ever used, but implementing it on the robot was not easy. With the kinect you are given a grid of depth values - tracking game pieces, scoring zones, etc. are all possible with this sensor.

Re: Sensors
 

Gyroscope:
Always, always, always use a gyroscope on your robot for autonomous, you would not believe how much better it is when you have to drive in a straight line. Be warned, when you place your robot, it must be off and once you turn it on, you cannot move it or the gyro will be calibrated wrong.

Accelerometer:
Useful for things this year like balancing the bridge, it determines angle so you can tell if the bridge is level or not.

Shaft Encoders:
Lots of different types, there are optical, magnetic etc. (more, but I can't think of any). Put these on your drive train, shooters that use wheels, anything that spins freely like wheels. they don't have physical limits and therefor cannot be broken by spinning too far. use these to judge speed of drivetrain, Set up PID loops in your code to tell shooter wheels to run at a set speed (you can control speed based on RPMs rather than voltage, it's much more accurate) These work well for high speed, but also work with slower speeds too.

Potentiometers and string-pots
Have physical limits! Cannot be used for wheels or shooters as they cannot fully rotate forever. Good for arms and turrets, could also be used with crab or swerve drive (to some degree) You can measure rotational distance and angles.

Light sensors:
Use them to track lines like in 2011

Sonar/Ultrasonic:
Determine distance from objects/other robots/field walls

Lasers and Sensor: same uses as above

Camera: Vision tracking, line following, goal finding, game piece finding etc.

Kinect: Same as camera + sonar/ultrasonic and laser.

Limit switches: determine limits rotationally (turrets or arms) or linearly (slides, elevators, pneumatic cylinders etc.) You can also use them to detect game pieces.

All I can think of right now, sorry if I missed any big ones, but thats what I know.

Edit: Get a variety of the same type of sensor, some work better in other locations than others and get a variety of sensors as well.

Re: Sensors
 

The one I haven't seen mentioned is the Absolute Encoder.
To my understanding they are similar to potentiometers in that they don't judge angle from the starting position, but unlike potentiometers, they don't have physical limits.

They are generally expensive, but we use at least one every year.

Re: Sensors
 

The banner laser trip sensor works really good for counting game pieces, but don't mount them on plastic as the plastic warps and unaligns them.

Re: Sensors
 

Kevin,
When faced with a gift, I would go for some long term needs like tools or a few extra batteries. A good crimper, DVM, auto wire strippers, and extra DSC, and PD are always useful.

Re: Sensors
 

An accelerometer does NOT return angle. It returns the sum of all acceleration applied to it in the axis it measures. If you are NOT moving in any way (or at a perfectly constant speed... very hard to do in FRC), the only acceleration you will see is that of gravity, from which you can determine the angle of gravity relative to the axis of the accelerometer with one or two accelerometers.

If you are trying to balance a bridge using the accelerometer, you will also see acceleration from the driving, which will make the sensor see gravity as being in a different place.

While I have seen 6-axis IMU algorithms which can accurately keep track of their position in 5 dimensions indefinitely (it's hard to keep track of yaw indefinitely without any other sensors), that is far too complicated for an FRC robot (as cool as it is)

A gyro is most likely the right sensor for auto balancing.

As for useful sensors, consider the interface. For absolute sensors, an analog interface is the easiest to interface with the current control system. For incremental sensors, a quadrature or single-phase digital output is the easiest. Anything else (especially USB) is significantly more difficult. My favorite sensors of all time would probably the 1 or 10 turn analog potentiometer, and the Banner reflected light sensor.

Final thought - We on 33 have a policy against using an incremental sensor for measuring absolute position. This means that any mechanism that has a defined, fixed limit (such as an arm or elevator) MUST use an absolute sensor. We do this in case the robot ever reboots during a match, so it dosen't have to re-initialize itself.

Re: Sensors
 

An accelerometer can most certainly be used for auto balancing. If the robot is stationary but there is acceleration on a horizontal axis, the robot is not on a level surface (or it is being pushed, but let's assume the former). To correct for this you drive away from the acceleration (up). This is more likely to work if you scale the motion based on the difference between the vertical acceleration perceived by the sensor at that moment and the normal acceleration due to gravity, by using the vertical axis to calculate this, the affects of the robots motion are reduced. For even more accuracy, you could also have two sensors on opposite sides of the robot, one on either side of the pivot. Averaging the two values would minimize the effects of a shifting bridge. If the robot COG was centered and the sensors were set at equal distances from the COG, the average would be 9.8 m/s^2. the position of the sensors in relation to the COG is important, if that condition is not met an average of 9.8m/s^2 only means that the sensors are equally far away from the bridges pivot.

Re: Sensors
 

Thank you everyone for the info!

Re: Sensors
 

I should have said let's you calculate angle based on gravity. Which would be a viable auto balancer, just tell it to set the angle to 0. I had said angle because of how I have personally used the accelerometer in phones and iPods. Sorry for any confusion.

Re: Sensors
 

$40 and an Arduino will get you a 3 axis accelerometer and Gyro that can give you pitch, yaw and roll. As was mentioned the yaw is going to drift over time. 6 axis IMU's have improved allot. Our team is going to work with them this fall. Maybe we can get field centric drive working. If we could only firure out how to get a magnetometer working on the bot, neat things could be done. All those magnetic fields put the damper on them. An IMU is what is needed for good bridge balancing.