Wall Edge Detector

Interested in Robotics? Here's the place to be.
Post Reply
sambo_commando
Posts: 5
Joined: Wed May 28, 2003 1:01 am
Contact:

Wall Edge Detector

Post by sambo_commando » Tue Sep 30, 2003 9:48 pm

I am looking for ideas on a lightweight low power wall edge detector for a wall crawling robot. Possibilities I have considered so far are an IR detector or maybe some sort of sonic detector. This project is for my EE senior design class. I am complete novice in the robotics world so any direction would be greatly appreciated. Thanks.<p>Sam

hlreed
Posts: 349
Joined: Wed Jan 09, 2002 1:01 am
Location: Richmond, TX
Contact:

Re: Wall Edge Detector

Post by hlreed » Wed Oct 01, 2003 9:35 am

Sam, I assume your robot is on the floor and needs to see a little bit.
To detect differences in light you need att least two light sensors. Call them Sl and Sr and arrange them a distance s apart. You need to run the sensors through an ADC to create two integers which you can still call Sl and Sr.
Your robot will need two motors, one for steering and one for motion forward/backward.
Call them Ms and Mg for steering and go. You will need an HBridge controller on each motor, such that they accept an integer and run the motor proportionately. With all that you can now write an algebra that is the robot.
Let Ms = Sl - Sr ;Steering seeks Sl = Sr
Let Mg = Ms ; keep it simple.
It takes a lot of machinery to do even this simple vision system, but if you have this foundation it can be expanded without too much effort.
Draw all this on paper before you build anything.
A light sensor produces a change in resistance which produces a change in voltage which you can convert into an integer. (-128 to 0 to 127)
Draw a sensor connected to an ADC. That is Sl
Make another for Sr.
Connect Sl and Sr to a node that subtracts Sr from Sl. Connect that to an HBridge controller which is connected to a motor. That is Ms. Run that connection over to another HBridge controller and motor. That is Mg.<p>Let me know if you have questions.
Harold L. Reed
Microbes got brains

sambo_commando
Posts: 5
Joined: Wed May 28, 2003 1:01 am
Contact:

Re: Wall Edge Detector

Post by sambo_commando » Wed Oct 01, 2003 5:16 pm

Will that system work if the terrain is not all one color? The robot will currently drive around on a brick wall on the side of a building. I have to make sure that it doesn't run off the wall and plummet to the ground. Thanks for the info Harold.<p>Sam

hlreed
Posts: 349
Joined: Wed Jan 09, 2002 1:01 am
Location: Richmond, TX
Contact:

Re: Wall Edge Detector

Post by hlreed » Thu Oct 02, 2003 9:09 am

Sam, the system I described will not do what you want your robot to do, but it shows how you go about designing it.
I get the idea you want your robot to crawl on a brick wall. This is a mechanical problem. Insects can do that because they are light enough that attractive forces will hold them.
You need detection. (Am I on the wall or is the wall there?.)
and propulsion. (to crawl.)
You are starting correctly. The first thing you do is decide what capabilities your robot has. Then you impliment those capabilities.<p>How are you hanging on the wall?
Harold L. Reed
Microbes got brains

sambo_commando
Posts: 5
Joined: Wed May 28, 2003 1:01 am
Contact:

Re: Wall Edge Detector

Post by sambo_commando » Thu Oct 02, 2003 10:30 am

I didn't design the existing robot (I am adding to it) so I don't know the specific details, but it essentially creates a low pressure area underneath the robot such that atmospheric pressure pushes it against the wall. I am not creating a robot from scratch. I just have to add sensors to it that make it smarter. The primary considerations for sensors are: 1. Correct sensing, 2. Low power, and 3. Lightweight. The first consideration is a given, and the last two directly relate to extending the length of use or increasing the available weight for a possible payload.<p>What would really be useful to me is if someone directed me to a resource were I could find some example projects of table edge detection. I looked around on the internet and I tried in the school library and I haven't found much of anything. I keep getting all of this stuff about Canny Edge Detection which is an altogether different project than what I am working on. Some books or magazine articles would be great. Also, the sensors will be tied back to a Basic Stamp 2sx chip so if the material relates to the stamp that would be great too. Thanks.<p>Sam

hlreed
Posts: 349
Joined: Wed Jan 09, 2002 1:01 am
Location: Richmond, TX
Contact:

Re: Wall Edge Detector

Post by hlreed » Thu Oct 02, 2003 1:36 pm

Sam,
Table edge detector still requires two sensors.
Use the setup I described before, but put the sensors in line in the direction of motion. When the sensors see the wall, the difference will be near zero. When the lead sensor goes off the table a large difference will be seen.
The motion forward is independent of this and should be directed to stop or do something when a value appears in the difference stream.
Your stamp can do that.
Harold L. Reed
Microbes got brains

bwts
Posts: 229
Joined: Tue Jun 11, 2002 1:01 am
Location: britain
Contact:

Re: Wall Edge Detector

Post by bwts » Fri Oct 03, 2003 4:15 am

What U need is a "feeler" a light weight arm infront of ur robot (may B with a roller on the end) that tracks along the surface in front when the arm (spring loaded of course ensuring that there is not too much tension in the spring) moves in this case down too far (could be connected to a pot) then an edge has been encountered. Remember surfaces r rarley smooth so there should B a range of allowable movement for the feeler.<p>B)<p>[ October 03, 2003: Message edited by: Mr Bwtz ]</p>
"Nothing is true, all is permitted" - Hassan i Sabbah

Colinr
Posts: 33
Joined: Wed Oct 23, 2002 1:01 am
Location: UK
Contact:

Re: Wall Edge Detector

Post by Colinr » Fri Oct 03, 2003 5:28 am

A possable way would be to use ulta sonic range detection do a search for PICAXE (a low cost basic stamp look alike) Where thier is an ultra sonic module that is easy to interface. fit this ubove the robot angled so that the beam hits the wall about 6" infront of the robot. the distance will tell you that the wall is present also you will be able to identify possable obsitals infront of the robot eg window sills or moldings. <p>A further inprovement will be to rotate the module using a RC servo this will alow you to build up a complete map of the tarain around the robot.<p>
Colin

sambo_commando
Posts: 5
Joined: Wed May 28, 2003 1:01 am
Contact:

Re: Wall Edge Detector

Post by sambo_commando » Sun Oct 05, 2003 10:47 pm

Mr Bwtz, I thought about the sping loaded feeler and it seems to fit the low-power/light-weight requirement but I wasn't sure about how it would affect the initial transition from the ground to the wall. Is there a place where I could buy this sort of feeler?<p>Colin, I also considered an ultra-sonic type device figuring that I wanted my detection to be independent of visible light (i.e. it operates in daylight or at night). Does the ultrasonic device provide any advantages over an infrared detector? I would assume both are independent of lighting so my concern lies with which is lighter and consumes less power.<p>Thanks for all the help everyone. You have really got my wheels turning.<p>Sam

bwts
Posts: 229
Joined: Tue Jun 11, 2002 1:01 am
Location: britain
Contact:

Re: Wall Edge Detector

Post by bwts » Mon Oct 06, 2003 5:12 am

Making one would probably B cheeper!! How does the robot mount a wall? or is it placed on the wall its going 2 climb? The feeler could have a differnebt function depending on wether the robot was verticle or horizontal (a mercury switch could B used to detect this).<p>B)
"Nothing is true, all is permitted" - Hassan i Sabbah

Colinr
Posts: 33
Joined: Wed Oct 23, 2002 1:01 am
Location: UK
Contact:

Re: Wall Edge Detector

Post by Colinr » Mon Oct 06, 2003 5:36 am

Infra red devices will generally give you a yes / no decision of a object being present(the wall in yoyr case) to reflect the light it can be made immume to other light sources by modulateing the drive to the Infra red emmitter and looking for a modulated signal from the detector. <p>The ultra sonic approach will provide a distance to the object and from memory this is in the range of 3cm to 3m this gives you the advantage of being able to detect other obisticals. <p>A note of warning though both of the above options will only work if thier is enough reflection off the object from the emmitter to the detector. If the wall is smooth then the ultra sonic option with the module angled down so that it hits the wall just in front of the robot may not work.

Daggs
Posts: 8
Joined: Tue Oct 07, 2003 1:01 am
Location: Where ever the AF sends me
Contact:

Re: Wall Edge Detector

Post by Daggs » Tue Oct 07, 2003 11:33 am

<blockquote><font size="1" face="Verdana, Helvetica, sans-serif">quote:</font><hr>Originally posted by Sambo:
Is there a place where I could buy this sort of feeler?<p>
Sam
<hr></blockquote><p>Make one with a roll-over switch and a long wire that activates the switch. Go to an arcade and look at a pinball machine to see how these work.
Casual Hero
Combat Gynecologist

Post Reply

Who is online

Users browsing this forum: No registered users and 0 guests