I'm approaching a Rover like a dog.
What does a dog do? (no bad jokes please)
- track (not the same thing!)
Take any one of those words, and you can probably expand on it - with various sensors and useful ideas.
It's a mechanical take on why we have kept those flesh beasts around us, all these years...
It's only one approach.
Maybe make a word list for cars - and see where that takes you.
The thought had just occurred to me that an autonomous car/robot could take elevation measurements as part of a topographical survey. The GPS readings would not be terribly accurate, given that the onboard AP2 is accurate to a few feet, I think, vertically, but it might be an interesting experiment for someone to try and tackle.
I don't see much use for RC cars. With a boat or plane the GPS is accurate enough since you don't have to follow an exact path or avoid anything.
With a car your 3-5m GPS accuracy (when you get even that) isn't going to be quite good enough to keep you in a road lane or on a path. You also would have to give a large berth to any areas you must avoid. So you won't be dodging potholes or weaving through cones.
One cool thing I did read about once was a little rolling robot that drew chalk designs on the sidewalk. It was kind of like a printer and I think it dumped different colored chalk dust onto the sidewalk in whatever pattern you wanted. (it was more like a printer than a plotter)
I thought it was slick since you could have the robot rolling around leaving messages or advertising or whatever. Since it's just chalk it's as legal as drawing on a sidewalk with chalk, which people do all the time. You could have it draw directions or advertising or lay a trail for hashing, etc..
A few simple sensors and you could have that thing rolling around like a roomba marking out whatever you want. GPS would be good, but I don't think you'd get much out of the other sensors. A magnetometer might help the GPS with your direction, but I don't see what the others could do. There's a lot of work that's been done on these sort of things, so there's a lot of designs for sensors and pathfinding.
I say make the ultimate advertising robot. It would truck up and down mainstreet sidewalks covered in advertising, harassing peds, and laying chalk. All that should be legal as long as it doesn't run over/into someone.
Just cross a roomba with a dalek.
I want to implement a 6 (maybe 9) axis IMU in order to tune my F1 RC Car during races. I could use it to calculate my traction circle, slip angles, etc. and then use that data to optimize my camber and caster, tire compound, etc.. Just like a real F1 car. :)
Given that Google has a full size car that has logged tens of thousands of miles I'm going to have to say that lane following is entirely within reach but, yeah, not with GPS alone.
What's needed is a video system and better obstacle detection sensors & software to augment the Rover's accuracy / following capabilities. Not many of us can afford current LIDAR systems.
My rover, Data Bus, is able to achieve approximately 1m precision/accuracy on parking lot pavement over a ~30 second, 270m run by adding nothing more than low end gyro, wheel encoders, and a rather simple Kalman Filter. Better is possible. That level of accuracy over longer periods...? Don't know about that...
I like the dog analogy. I've been trying to figure out what autonomous rovers are good for. The only thing I've come up with is going to the grocery store for me... but I don't know how it'd get stuff off the shelf. :D
We created a robot that can recognize, track, and follow a specified color object as well as have the ability to avoid obstacles in its way. In a nutshell, we installed an Android device onto a RC car and used an IOIO board to control it autonomously. I hope might help you with some ideas: