Mousebot?
I know I've been chatting a lot more than doing, but I was running this idea through my head.
If I used an old mouse instead of the ball caster, I can use it to keep track of where my robot is heading.
If I'm using the mouse to track my robot, then I don't need the wheel encoders.
Why not just slap a pair of cheap Radio Shack hobby motors to the back of a mouse, build a simple H-Bridge circuit and connect THAT to my QuickStart board. I could even hack the mouse buttons into front bumper switches. I'm sure 1 cog could read the USB input from the mouse, then another can run the motors, and then give it some autonomous capability.
That's something I can probably start by this weekend or so.
Cheers,
John Gay
If I used an old mouse instead of the ball caster, I can use it to keep track of where my robot is heading.
If I'm using the mouse to track my robot, then I don't need the wheel encoders.
Why not just slap a pair of cheap Radio Shack hobby motors to the back of a mouse, build a simple H-Bridge circuit and connect THAT to my QuickStart board. I could even hack the mouse buttons into front bumper switches. I'm sure 1 cog could read the USB input from the mouse, then another can run the motors, and then give it some autonomous capability.
That's something I can probably start by this weekend or so.
Cheers,
John Gay
Comments
Parallax sells an optical mouse sensor. You could try that for your odometry needs. I've heard the mouse approach isn't as accurate as normal encoders though. It would still be fun to try.
http://www.instructables.com/id/Mousebot-Revisited/
Another one;
http://letsmakerobots.com/node/26463
My main goal is start cheap and simple. 2 hobby motors, a handfull of transistors and diodes for a H-bridge, and slap it onto a mouse for tracking. Then practice making drive reliably. First doing straight lines. Boxes. The inevitable figure 8 (-=
Then giving it some simulated free will. Chasing the dogs and cat around for fun.
Wheel encoders tell how far a wheel has turned, but can't know if the wheel has slipped. The mouse input, either ball or optical can tell how much the 'bot has traveled regardless of the wheels moving. Either way is open to errors. I'll be adding whiskers to the buttons for wall detection. Resolving the differences between how far the 'bot THINKS it's traveled and when it hits a wall is an exercise to solve in software.
Computer mice don't need to be as accurate as a normal wheel encoder becuase us humans keep adding the subtle corrections automatically (I think I just quoted one of erco's posts). I'd be very surprised if odometry using a mouse sensor is as accurate as using enocoders on wheels.
Not that I think you shouldn't try mouse odometry. I even hope you prove me wrong.
Just got this for $17.99
Not much, but it's a start. First get it running, then put my prop in it and start coding.
Plan B is under development.
I thought the idea was to use the encoders of a mouse for odometry?
Are you going to transplant guts from a PS/2 mouse into the clear mouse?
As I previously mentioned, Parallax sells a mouse sensor. It might be easier to install the sensor from Parallax than trying to transplant mouse guts.
SPECS: Voltage range: 1.5-3V
Nominal voltage: 1.5V
Normal load: .091 oz-in 6.5 g-cm
RPM at no load: 7600
RPM at nominal: load 5200
Current mA (no load): 270mA
So, nominal voltage is 1.5V and max is 3V. So I should be find with a single OR dual battery pack for the motors. But, the motor driver is a little more troublesome?
As I understand the circuit for a basic PNP-type H-Bridge, the motor supply voltage at the collector should be higher than the switching voltage on the base.
Will using MOSFET's fix this? I studied this stuff over 30 years ago and only remember the basics now.
Otherwise, I'll just look for higher voltage motors that fit the worm gear and motor mount.
Cheers,
John Gay