I’ve been reading around on the forum, and ive seen people talk about how odometry may become inaccurate over time during driver period due to maybe moving too fast so sensors can’t keep up, or getting shoved around by other bots
I think odometry will be important for my bot during driver and I’m looking at ways to keep it consistent
Is it a good idea to try combining 2 wheel + imu and 3 wheel to have 2 sources of heading, and then average the 2? Or is that overkill?
(On a separate note, how do the 3 wire port expanders work)
That’s way overkill, you’d need 5 tracking wheels which takes up a lot of space. One thing teams do to keep odometry consistent is using reset points on the field with known coordinates to eliminate any error that has accumulated.
Why? You’d get better results by just practicing instead of trying to make odometry consistent for the entire match.
I didn’t mean 3 track+ 2 track + imu, I meant just using 3 tracking wheels, but also using imu for heading (probably should have clarified)
I need odometry for something to assist me during the driver period with aiming, doing it manually would slow me down a fair bit and would make my strategy kind of useless
The issue with Odom during driver control, is that as soon as any tracking wheel leaves the ground, your accuracy is gone. In auton , this can be prevented by pid control to keep your robot from popping wheelies and you don’t have to worry about intense robot interaction. As soon as another robot rams you at full speed, your turret will start missing its shots. There really isn’t much you can do about it.
You can also just have your tracking wheels on a rotation point such as a hinge and have them rubber banded to the ground. This completely solves the issue of them leaving the ground, however odometry drift is unpredictable at times. It generally occurs when your tracking wheels are slightly misaligned. However there are a bunch of different variables that go into it.
It may be possible to improve the performance of odometry in driver control by developing a traction control or can be like a slew rate pretty much, at the cost of having a robot that is slightly slower when it comes to controller → robot responsiveness. Not to mention I am unsure if odometry can last a whole round without losing its accuracy regardless. Probably worth testing.
That’s more reasonable, but in my experience IMU is more accurate than 2 tracking wheels for heading, so we used 2 IMUs and averaged them for more accuracy.
You’d be surprised how good you can get with practice.