Sensors needed for swerve?

[elbuo]

What sensors are needed to make a perfect swerve drive?

[AmoryG]

I'm not sure about a "perfect" swerve, but I would think 1 gyro and a servo on each swerve module would make a swerve drive more intuitive. Having directional control was a nice little feature on our last robot. Our team also put on hall-effect sensors so that when we turn the robot on, the wheels would automatically realign themselves.

[daltore]

Last year, we cut gear teeth into the top plate of our swerve modules and put a smaller gear on a 10-turn pot attached to the robot. I wouldn't suggest it, because it means you can't turn the module continuously around, but we also had the motors mounted on the modules.

I would actually try those absolute encoders that came in the kit. Glue the magnet onto the end of a shaft that turns with the module, then place the chip above it. If you're using CAN and the Jaguars, you can do position control directly (but you have to make a voltage divider circuit for the Jaguar, because the board runs off of 5v, but the Jaguar ADC is 3v), or you can do your own PID in the code. One idea I've played around with is using a Soft Pot and the corresponding wiper as an absolute encoder that can sit right above the module.

Never heard of the gyro idea, sounds intriguing. How well does it work for you?

[JewishDan18]

This year, team 20 has a half-swerve drive (each pair on the sides is linked). We have a gyro and a 10 turn pot on each side, and it drives exceptionally well. The gyro maintains the robot's heading, and the pots allow the wheels to spin about 5 turns total (took some effort to get the wheels to spin that far around). Plus, they auto-center at the start of the match, or whenever they're idle.

[elbuo]

I used 1 encoder.
Wheels center each other when ever the joystick is idle. And the joystick and motors are "linked".
If I am at 45 degrees in my pad, the motors align themselves at 45 degrees and stop spinning until I change the direction.
I didn't know there where absolute, I though they didn't know the center. I will look for those...) Thanks!

PS: I have 1 encoder per wheel right now. But I will ask the mechanical team to link them just like team 20

[Foster]

Look at this page that describes the Team 1640 Swerve Drive. We use a Cherry Magnetic Sensor at the top of the pivot to tell the direction. It works well, the system knows which way the wheel is facing at all times.

But, our programming team came up with a "least turn" algorithm. Since there are drive motors are for each wheel and they don't really have the concept of "forwards/backwards", the programmers know which way to spin the wheels that gets them aligned in the fastest way. So unlike the "side linked crab" swerve drives that you see, we can go from going one direction to another by only moving the wheels < 90 degrees.

It also allows us to create and drive in snake mode to make it easier to follow the ball.

[Gdeaver]

To expand on what Foster said, this year team 1640 went with 4 wheel independent steering. Each wheel can independently rotate 360 degrees. We used the window motors for the steering. A 3/4" 15 tooth sprocket was mounted to the window motor and a 1/8 x 1/2" neodymium magnet is centered and atached on top of the sprocket. This assembly drives the wheel module 1: 1. A AN8 360 Cherry Electric absolute position sensor is mounted 2.5MM away from the magnet. The Cherry sensor was chosen over many other similiar sensor for many reasons.
1) its non-contact.
2) It is a robust sensor being IP67 and automotive qualified. Reverse battery protected. High static discharge rated. Sealed and designed for the nasty automotive environment.
3) Cost we paid about 35$
4) The sensor is tolerant of the magnet position misalignment and distance much more than other like sensors.
5)The Cherry sensor puts out a voltage from .5 to 4.5 over the 360 degree rotation. This avoids the non-linearity at the rails (0 and 5 volts) that other 0 - 5 volt magnetic sensors show.
We looked at some other like sensors that were SPI but did not want to have the programmers work out the communications. Their plate was full enough. Analog is much more simple. The voltage output was very stable and filtering was not needed. We are very satisfied with the Cherry product. Again, if your into swerve, I would check out the link above that Foster posted. There is allot of information- pictures and white papers on what we did and how we approached swerve. Swerve- crab-pivot drive is a major under taking.