Line following robot [average speed: min. 1.5 m/s]

Hey,

We are three engineering students, from the Technical University of Denmark, who wants to participate in a “robocup”. Our goal is to build a line following robot, which can keep an average speed of at least 1.5 m/s (preferable more) through curves and sense obstacles.

We are pretty much newbies, so the question is:
What do we need to buy?

We have enough time, and want to learn a lot, but to keep it manageable we want to buy some sort of kit. We have been considering the 4WD rover from lynxmotion, but we are afraid it’ll be to slow with a tracking device. What is the most optimale line tracking device?

We are thankful for all tips and advices!

:slight_smile:

Perhaps you should let us know what your skills and limitations are.
Can you build, test, and debug your own electronics?
Can you program a microcontroller?
Do you understand the different types of sensors and how they work?
How much money can you spend?
Etc.

Instead of using a “rover” kit, you could consider buying a radio-control car/truck, and then modifying it to add your own sensors and a microcontroller.

Pete

[Perhaps you should let us know what your skills and limitations are. ]
Can you build, test, and debug your own electronics?
Hmm, no, we can’t, but we hope to learn by doing…

Can you program a microcontroller?
We all have some programming skills, and are hoping it won’t take to long to understand.

Do you understand the different types of sensors and how they work?
Yes, but we don’t know how all sensors work in detail.

How much money can you spend?
Around 500$

For raw beginners, I recommend the “what’s a microcontroller” book from radioshack for around $60. This comes with the book, basic stamp II microcontroller board, a servo, and a handful of simple electronic components to complete the exercises in the book. It’s a very helpful starting point.

Once you got that complete, it makes transitioning to the atom micro a lot easier. There are several options and paths one can take when starting out, but for me personally, the book mentioned above was a fabulous learning experience that really got the wheels moving in my head in a short amount of time.

Just a suggestion.

I’d follow up on that with The Robotics with The BOE-bot book and kit:
parallax.com/detail.asp?product_id=28132 In fact if you have some minimal programing skills, you might want to start there. (The book is very straightforward) Add a Line-Follower AppKit: parallax.com/detail.asp?product_id=28108 and you can have a basic line-following robot for around $175 (buying directly from Parallax), much less shopping around. I got my BOE-bots for about $75 each on eBay.

This is an okay base, but the continuous rotational servos on the bot can’t get anywhere near 1.5 m/s. However, you could switch from a BOE chasis to a Lynx one easily enough for the competition, keeping all the electronics and sensors. You’d waste some money, going down this route, since you’d have a leftover BOE-bot body, but it might be worth the time and frustration saved by making learning easier.

Hope this was helpful, and good luck!

Can anyone give some more concrete advices about what to by, in order to get a line following robot that can travel at least 1.5 m/s?

I’d like to simply start naming motors, wheels, and chassis that would work to produce those speeds, but there’s quite more than a few that would.
For instance, a 207rpm motor with 5" tires would do the trick.
But, so would a 344rpm motor with 3" tires.
Good news, though.
The LM wheel calculator helps you determine just that:
lynxmotion.com/images/html/wheelspd.htm

Skid steering (one motor per wheel) isn’t the most efficient means when one needs to maintain high speeds while turning.
Skid steering is nice if you have to turn on a dime (without much positional accuracy), but it doesn’t sound like you’ll need that.

In that case, you might want to go with an Ackerman steered (one drive motor, one steering motor) vehicle.
A car is an example of this.

There aren’t any kits on the LM site that’d would fit that setup (you’d need a right angle gearbox to translate the motor rotation), but almost every RC car would.

If you’re enamoured of using the 4WD design, then calculating the maximum speed is simple.
Use the LM wheel calculator program (see above), and you can calculate the top speed of your bot.
Keep in mind that the motors usually specify their no-load speed.
When you make them drag around the weight of your bot, they’ll be considerably slower so go with motors that are faster than you’d actually need.
Once you know the tiresize and motors, you can get a HUB that mates them.
Then you can chose the speed controller that suits the current draw of your motors.
Then you can decide which micro you’d like to use.
And then batteries and a charger.
I’d say… leave the chassis to the end.
Deciding on one now will limit you to using what fits within it.

As for tracking that solid line, how about a camera?
You can get a cheap wireless camera from geeks.com.
Link the reciever to a TV tuner card, and you’ll be able to wirelessly process the video stream using the free RoboRealm software.
Roborealm is VERY easy to use, as it lets you simply point and click to do most of it.
There’s even a linefollowing tutorial on their site:
roborealm.com

Then you just need something like a bluetooth module (wibotics.com) to wirelessly send info back to your microcontroller.

Instead of doing all that, you could use something like the CMUcam which is an integrated camera and processor package.
That way, all the processing would be done right on your robot.
The disadvantage to that is that you’d need to understand more about the hardware and software, and you don’t get to view what the robot is seeing, as you would with the roborealm software.

It’d probably be a good idea to suppliment the camera with old-school sensors, like phototransistors.
These are great for detecting lines at low speeds, but can easily get confused at higher speeds and with bad light conditions.

I don’t think either just the camera or just the phototransistors would work alone at those speeds.
The chances of error are very high.
But, I think you could get good results out of using both.

Being engineering students, you need to specify the design limits you have to work within. How big can the bot be? How wide is the line that is to be followed? What is the minimum radius of the curves in the line (assuming the line isn’t straight)? A bot the size of a riding lawn mower following a line that is .5 meters wide at 1.5 meters/sec might be fairly easy.