Automated Cart

I have an idea for an Arduino-controlled robotic cart that would follow a tag at a certain distance or respond to commands sent from a cell phone. This thread is just to assess the feasibility of the idea and point out any glaring errors.

I'm planning to use something like a wagon as the base. I would use a decent sized electric motor for the drive powered by a large enough battery (probably car, motorcycle, or boat). Steering would be accomplished with a bidirectional motor mounted vertically that would be attached to a gear. The gear would be attached to a frame that would turn the front wheels.

I know there is a SIM shield available, so I plan to use that for the cell commands.

The main part I'm not sure about in terms of feasibility is the tag and receiver. My current idea is to use an IR LED or a cluster of them mounted on a backpack or shoe so that the robot only has to scan on one plane. What I'm not sure about, however, is whether or not a receiver would be able to pick up the IR LEDs and how to keep the robot at the same distance, especially when the person it's following turns.

Is this idea feasible, and what things am I forgetting to think about? (I'm not worried about the programming yet; I'll deal with that later, as long as I know it's possible)

It sounds as if it would be possible to get something like that working as a proof-of-concept.

Getting it working reliably enough to do anything useful in the real world is another matter, and I think that would be quite tough.

Given your description I'm envisaging something big enough to be a nuisance (and/or perhaps even dangerous) if it ran amok so reliability is probably an important requirement.

Yes. You are correct in that it would be fairly large.

Do you have ideas on how to put together a tag-sensor assembly that would work in the real world?

There have been numerous post in the past for object tracking projects, such as a golf caddie that follows the golfer or a camera that follows a person. None of them appeared to go anywhere. You could do a google search of this and the old forum for key words of your project.

Permanewb:
Yes. You are correct in that it would be fairly large.

Do you have ideas on how to put together a tag-sensor assembly that would work in the real world?

I’ve never tried to do it myself, and I have no idea how difficult it would be to get working, but I’d try using some sort of beacon on the target (perhaps a light source or IR source) that could be detected by a scanning sensor on the cart. By putting two sensors a known distance apart you can use triangulation to estimate the range. I’m pretty sure you could get something working. I’m much less sure that you could get something working reliably enough to be trusted out in the wild.