Society of Robots - Robot Forum
General Misc => Misc => Topic started by: nerdmeister on April 18, 2010, 09:01:59 PM
-
Basically it's the feasibility of my senior design project. this year we're doing group projects and we generally do pick our own projects but i got put into a group that had already picked and what I'm worried about is whether or not such a project is possible to put together in less than a years time with acceptable results that stays within our budget.
The project is to make a fully autonomous car that can follow another car in front of it. I know it's really easy to make something to directly follow what's right in front of it, but here's the tricky part: The car needs to follow at a distance of about 1-1.5 meters and it can't just go straight at the leading car, it has to take the same path, because we're designing this to be autonomous for highway roads. Additionally the following car is FULLY autonomous and receives NO input from the lead car and the cars should be going about 3-5km/h.
Personally, from my limited knowledge, this project seems incredibly difficult, if not impossible. My team is made of me (biomedical engineer), a computer engineer, and an electrical engineer. None of us have a background in robots, only circuits and some programming. One of my groupmates has also taken a microcontroller class.
I would like to hear what you all think of the feasibility of this project and how you think it would be possible to implement it, if it is in fact possible. Rather than just stating that it is feasible or not, please also explain why/why not.
By the way, total funds are probably around $1000 tops.
Thank you for any help or advice!
-
Following at 1 to 1.5 meters, there will be little difference between following the same path and just following straight at the car. If you can just follow straight at the lead car, it should be feasible if you have a full year.
-
While I appreciate your input, there's the inherent problem of this: (not great, threw it together really quick in paint, but gets the point across) (http://imgur.com/biHUS.jpg)
This is for my senior design project. I unfortunately can't just follow it straight and say "close enough" it has to actually do what we intend it to do, which is follow a lead car's path. On a real highway, it would be a bad idea to just to follow the lead car along a straight line, for obvious reasons.
-
Just an idea, I dont actually know how to do this, but if you could use an IR sensor to get an integer value for the lead cars position and store it in an array, then you could have the lead cars position saved in, for example, index five while your car is still moving towards index 0. That way your car will be moving towards a "shadow" of where the lead car used to be. It might be a good idea to use a 2D array to lay out your surroundings as a grid or something of the like...
Good luck, sounds really interesting!
-
I drew up a little diagram that shows what Im talking about a little more clearly I hope. Hope this helps!
-
But, are you using full size cars? If so, I dont think they would deviate from the path as much as your picture would suggest. But the "following the ghost car" by Warhawk sounds like a great idea.
-
@Warhawk87 - Thanks man, great suggestions, I will definitely look into that.
@Xyver - I'm sorry I should have written that out better. We will be trying to use models that are around 1/8 scale and they will probably have a top speed around 5km/h
-
I guess you will need an array of distance sensors (IR or narrow US, probably 3-5 sensors) mounted on the front of the car, just like the line following sensors, but facing forward. The sensors will read the distance to the followed car and the left-center-right position. You store the positions in an array say every 10 cm of travel. Then, after 15 positions you can follow the "shadow". You tend to keep the car in the center, so if the position goes left, you need to turn left, etc. Also, you need to measure the distance and keep it constant, that is if the front car slows down to take a curve, the second should too, but there should be some hysteresis.
Good luck!