I am building an autonomous electric vehicle that simply needs to travel a variable distance (8-12 meters) very fast (aiming for 3 seconds) with preferably +- 3cm error to the left/right.
I figure that I am going to need strong DC motors for this since speed is 10x more important than error. So I was wondering if I want maximum speed, should use 4 moderate DC motors (1 per wheel) like this:
or if I should use 1 really high-torque DC motor and have a 1-wheel-drive set-up ? like this
Also, I plan on using an Arduino to control the set-up and most likely an electronic speed controller.
Thank you for reading. =)
you need 4 esc's with the first option ! Single motor would be easiest and then you will correct the direction with the steering servo. 4 motors would require 4 speed controllers probably an encoder for each wheel (so they can spin at the same speed) and a tank drive control scheme for steering. rc vehicles are made to get to speed fast! Hopefully you have a friend or hobbyshop close by with rc stuff.
Put two lines on the ground 10 meters apart and just see what type of chassis works best for the surface type that you are running on. Make sure you can do it with manual control first and you'll get a better idea of the challenges. My bet is on a 4wd rally chassis with a nice soft rubber tire compound, depending on the surface.
Thank you for the information about 4 ESC's. I had not known about the 1-motor-to-1-ESC limitation before. In that case, I agree that a 1 motor set-up will be easier to start off with.
My dad used to build lots of RC planes, so he is helping me out with the project. But since this is cars, we are both learning.
And I forgot to clarify that the surface is a hard-tile floor or carpet, so I agree that a 4wd rally chassis will do just fine.
Thank you for the quick reply and the sound advice! =)
12m is 3 seconds needs an acceleration of 2.7m/s^2, which is not crazy fast, only 0.27g so most
rubber tyres on most surfaces will work.
The 3cm tracking requirement is vastly more of a problem. Expect this to be really hard to measure,
let alone achieve a control loop that works.
Maybe i misunderstood. I was assuming you need to get to speed and back down again.
Someone correct my physics here if i am wrong but if you are accelerating to the mid point then slowing to stop.
So 6 meters int the first 1.5 seconds and 6 meters in the last 1.5.
d = 1/2at^2
2d/t^2 = a
a = 5.45m/s^2
v = a * t so
this means you need will accelerate up to 8.175 m/s at the mid point (30kph). In 1.5 seconds. Its not unreasonable but its faster than i can run. Then back down. If it is a brick wall slowing you down at the end then yes, you can accelerate much more leisurely.
The rate of acceleration is not terribly high but you will have to accelerate slowly to minimize slip and bounce between tiles and over imperfection and the such. A hard rubber tire will make the problem worse between tiles on the floor.
There is no brick wall at the end. I will have to decelerate manually. However, it doesn't have to be an even acceleration/deceleration throughout the whole run.
And now, that I think about it, perhaps I will look to street-car/drifting tires rather than buggy tires in order to get maximum traction.
Thinking about this a bit more, a standard IMU (or just single axis gyro) might be the way to go for
steering control loop, but your control loop will have to take the current speed into account if its to
be stable for the whole acceleration. And you'll need to align the car to an accuracy of 8 minutes of
arc at the start (3cm in 12m is very tough problem without visual feedback - which tends to mean
a camera and fast computer running OpenCV? Or can you get away with line-following?)
MarkT, thank you for the advice on the IMU. I didn't know such a useful device existed! And as far as aligning goes, I think that in 3 seconds, there won't be much that I can do to change steering in real-time. I think I will use a laser-pointer attached to the front of the vehicle to get it as close to dead center as possible and then hope the vehicle doesn't veer too far.
And just for context, this project is for the Science Olympiad Electric Vehicle Competition, so measuring and aligning devices that are not on the vehicle have to be removed before the vehicle starts its run, so camera and OpenCV are not possible, but the IMU is still viable if I need it.
Thanks again for the recommendations!
I think you will need to have some steering control if you want to be right on your target. Marks advice about the gyro is spot on.
However if you remove the steering components you remove most of the chance for deviation as the steering links and tolerances will be a part of the problem. Without steering you can have have a very tight and well aligned setup there wont be much that can change your course. 3cm is a very small margin though.
I had a thought when looking at my mouse.
If you kept the steering aligned with a gyro and then had something like a big mouse ball / or optical system to track your deviation and distance. Think giant computer mouse!
sounds like great fun, make your vehicle as large as you can (looks like 35x60, a pretty decent size)