IR prox sensor

Thread Starter

Salgat

Joined Dec 23, 2006
218
For my microcontrollers class, our last project involves a robot that is programmed to avoid hitting a wall (just randomly roaming around). My question is, what is the best setup for an IR proximity sensor, or is there a better option? I want to avoid limit switches and anything that involves physical contact with the wall.
 

Thread Starter

Salgat

Joined Dec 23, 2006
218
Please answer the question, I know I can google IR proximity sensor, but I want to know all my options and appreciate the experience on these forums much more than scouring Google in hopes of finding my best option (I especially appreciate the little tips and secrets to getting it working good since I have friends who had issues with their IR prox sensor).
 

GS3

Joined Sep 21, 2007
408
For my microcontrollers class, our last project involves a robot that is programmed to avoid hitting a wall (just randomly roaming around). My question is, what is the best setup for an IR proximity sensor, or is there a better option? I want to avoid limit switches and anything that involves physical contact with the wall.
Is the wall defined or can it be any wall? Because if the wall is defined then it is much easier. Walls can be very reflective to almost not reflective of IR (like any other wavelength).

Then, do you need to measure distance to the wall or is a yes/no detection enough? If you can do with a (wall/no wall) detector that is much simpler than if you need a range of measurements.

If the wall and ambient conditions are defined and all you need is a yes/no detection then I think the project is doable with relative ease. If the wall and ambient conditions are not pre-defined and you need a range of distances then the project can go from very difficult to almost impossible and it could be that Ultrasound would work better.

Assuming defined ambient conditions and yes/no detection then I would design a pulsed IR transmitter and an IR receiver tuned to that frequency. When reception level passes a set threshold it's time to turn. Using a pulsed signal tends to surmount problems of ambient IR noise and/or fixed level.

Tell us more.
 

Thread Starter

Salgat

Joined Dec 23, 2006
218
Well, I was considering a pulsed IR sensor, and my main worry is the fluorescent lighting in the school. I'm assuming that it will be placed in a hallway, so all I need to worry about is walls with white paint (if that matters). The only requirement is that it stops and goes another direction before hitting the wall. Is there specific LEDs I should use, or would normal IR LEDs work? How many should I have for each side of the robot.
 

GS3

Joined Sep 21, 2007
408
For each side of the robot? It moves in all directions? Why not make it turn? It seems to me that making something which can move in any direction is much more complex than making something which moves only forward and can turn.

You don't say at what distance from the wall you need to detect it but if we are talking like 25 cm then I think one LED is more than enough. Put it at the end of a pipe so it sends all its beam quite focused forward and put the receiver at the bottom of another (almost) parallel tube so it only gets light from the same point forward. That should work.

Then receive and amplify the pulsed signal and then detect the envelope and put it through a comparator. Be careful with filtering the amplifier's power supply if it will be common with the transmitter.
 

Thread Starter

Salgat

Joined Dec 23, 2006
218
For each side of the robot? It moves in all directions? Why not make it turn? It seems to me that making something which can move in any direction is much more complex than making something which moves only forward and can turn.

You don't say at what distance from the wall you need to detect it but if we are talking like 25 cm then I think one LED is more than enough. Put it at the end of a pipe so it sends all its beam quite focused forward and put the receiver at the bottom of another (almost) parallel tube so it only gets light from the same point forward. That should work.

Then receive and amplify the pulsed signal and then detect the envelope and put it through a comparator. Be careful with filtering the amplifier's power supply if it will be common with the transmitter.
Thank you very much :D
 

GS3

Joined Sep 21, 2007
408
You could have the proximity signal cause the vehicle to turn. Suppose you have just one front wheel which normally points ahead, then the proximity signal turns it sideways until the signal disappears. If the vehicle is in a confined space then it will keep turning until it has a clear path ahead.

If the vehicle has two driving wheels then making one turn faster or slower than the other (or in different directions) will make the vehicle turn.
 
Top