Let's Make Robots!

Wall Bot (WallB)

Follows Walls and People

WallB is my first attempt at creating a robot that can carry a load down a hallway; specifically a small tool box. I'm constantly carrying my briefcase in one arm and a tool box in the other. The tool box weighs roughly 10lbs and its bulky shape makes it annoying to lug around. The idea came to me after reading about Steve Norris's "RoboStool". My project is split into 3 steps detailed below.     

 

Step 1: Test Motors and Sensors

 6/02/09

WallB takes its first steps using the Parallax Motor Mount Kit postion controllers to pulse the HB-25's (Parallax DC Motor Controller). Test code written to take forward distance from wall and calculate how many rotations to move until appox. 50cm away; WallB then moves to that distance and continues to keep that distance. First test run went great! 

See for yourself: WallB Test 1(YouTube)


 Step 2: Make it Follow Walls

6/24/09

WallB's preliminary code completed. HB-25's directly pulsed from the SUMO BS2; Motor Mount Kit position controllers not in use. Test results were great!

See for yourself: WallB Test 2(YouTube)

-------------------------------------------------

7/5/09

This weekend I took WallB out to a school with long hallways. Minor tweaks in the code made it go straight and smooth at home but when the environment changed, it went back to fish tailing. This was due to the IR sensors sensitivity to different color walls. I'm going to swap the corner IR sensors out for some more Ping sensors. Hopefully this problem will go away. 

See for yourself: WallB Test 3(YouTube)

-------------------------------------------------

***UPDATE***

7/29/09

WallB has received some upgrades in the brain department. I switched out the BS2's for a Propeller. I finally sat down and learned the basics of Spin, which was enough to make WallB work. Right away I saw an improvement in overall reaction to sensor data. I also figured out what was making WallB fishtail so much. Turns out the HB25 motor controllers aren’t putting out the same voltages. The right wheel is receiving .3v less than the other. I think its my wiring but since this is a prototype setup I decided to fudge the pulse numbers to make it go straight.

 

See for yourself: WallB Test 4(YouTube)

 


Step 3: Make it Follow Me

***UPDATE***

7/29/09

I wasn’t too excited to start this part of the project after reading other peoples troubles with this subject, but I seemed to have stubbed into a simple method for solving this. It happened while I was testing WallB's wall following capabilities when it ran out of wall and started to follow my leg instead. It was surprising how well it worked. I actually just put all my gear onto it and had it follow me to the car. It was like my sidekick. Unfortunately I didn't have anyone to hold the camera so its at a weird hard to see angle. The concept is simple, the bots senses my legs and tries to keep a certain distance. When my legs go too far it keeps turning until it finds them again(pretty much how it follows walls).

 

 See for yourself: WallB Follow Test 1(YouTube)

 



 Future:

Upgrade Chassis


 

For similar project see TheGrue's "TOBI"

 

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

How did you write your wall following algorithm? I am intrigued. Would it be possible to have a look at your code or flow diagrams?

I was just wondering how your robot project is coming along?

That "Follow Me" test is WAY COOL! Do you get good readings with the Ping))) at 45 Degrees like that? Can you post you Propeller code? I am still wresling with Spin even after going through the Prop manual that came with the Demo Board. I only have one Ping))) right now and need 2 more and a Sharp IR for my forward sensor array. Looks like you are well on the way. I am stalled with trying to recieve serial data from my remote control and the lack of money for my sensors.

Again I say it: Way Cool and VERY inspiring to me to get off my butt and get TOBI on the road!

Well like I said in my post, this was not on purpose. I was really shocked at how well it worked when I first discovered it. I'm currently in the process of writing a code that will react and match the speed of my legs. That’s the only problem it has so far; it just keeps going if I stop. As for the Ping sensors, they’re not perfect. The one that is doing most of the work is the corner. I think if I installed more(A rounded off front with about 5 in 180 degree pattern) it would be more reliable. From time to time I have to give it a kick in the butt to keep it away from obstacles, too many blind spots.

What kind of IR sensors did you use?

Basic Panasonic 3 pin receivers you find on Parallax's web page or RadioShack. I know I know there are much better ones I could use that are more accurate but after field testing I found that IR fluctuates too much to be used as a constant. If I were using them as just a object detector they would work great, but when trying to keep distance they fail. If you can argue my opinion please do; I'm not always right.

 BTW, did you see this: http://forums.parallax.com/forums/default.aspx?f=10&m=337693 

Check my Blog on the Propeller Expo and you see it right after the sharp edged steel scorpion.

200906271425_00038.jpg

After spending only a short time playing around with a new sharp distance indicator, I find them very unreliable in discernng distance. In my opinon there are too many things that need to be perfect - most noteably the colour of the surface, the level of ambient light (solar AND domestic flourescent lights), and more frustratingly, the angle of the reflecting surface. I use home-brewed sensors (largely identical to the circuits posted by oddbot) for the REALLY short-range cliff detectors - in my case I'm simply detecting when the ground isn't there...

but ping sensors are tooooo expensive...

 

 

"but ping sensors are tooooo expensive..."

 Agreed :/  

Did you go to the Expo at Parallax this last weekend in Sac?