Sharp IR VS Ultra SONIC for Object detection

[wany]

Hey
I am new to this forum but not new to programming MicroController. I have programmed alot of HCS12 with code warrior.
As a final year project for my UNI. I have to build something and present it to the UNI.

I was think i would built a Collision avoidance robot. That the basis i may upgrade it as i get more and more stuff work.

Basically i am stuck with a problem Sharp IR VS UltraSonic for object detection.
I notice alot of people have built robots to do this on this forum. So i wondering which one did you find better i notice alot people are using the SHARP IR how good does it work?
Also i am not so familiar with Sharp IR how does it out put Signal is it linear i am guessing it only linear within a certain range?

With that said i have also seen alot of Ultrasonic device on the market but i was thinking if i would use ultra sonic i would just build a ultrasonic device. Do you think i should do this VS buying it.
Because the mirco that i have can definitely drive the Ultrasonic device that i build, but i am a bit uncertain about the accuracy. Could anyone give me advise on good ultrasonic transceiver

any info would be much appropriated

[waltr]

Also i am not so familiar with Sharp IR how does it out put Signal is it linear i am guessing it only linear within a certain range?

The Sharp data sheet has a voltage verse distance curve. Did you actually look at the data sheet? The answer is in there and the main purpose of University projects is for you to learn how to learn on your own, that is why it is called research.

Same with the Ultra sonic sensors. Start by reading the data sheets then then the app notes on the manufacture's web sites.

[wany]

Yea i have seen the datasheet.  But i just wanted to know what people think about IR vs Ultrasonic. When it comes to object detection and avoidance?

[macdad-]

Pros of IR:

• Uses noticeably less power than Ultra-Sonic.
• Very flexible as far a Transmission frequencies(As long as the IR unit is adjustable).
• Easy to find parts for(TV remotes, set-top boxes, etc.).
• Relatively cheap, even for Sharp IR units.

Cons of IR:

• Can be interfered easily unless unique data stream is sent to differentiate between other IR Sources.
• Is sometimes ineffective at detecting dark objects or IR-absorbing material.

Pros of Ultra-Sonic:

• Much more resilient to interference.
• Can detect more objects than IR.

Cons of Ultra-Sonic:

• Hard to find spare parts(Ultra-Sonic Transducers are kinda a item that you would only find in mail-order.
• Somewhat more expensive than IR.
• Use considerably more power than IR.

In my opinion I would stick with IR due to its availability and ease of use. I have not used Ultra-Sonic but have seen datasheets on them and they can become power-hungry. Another important factor that you want to take into consideration is the objects that you expect to be detected. But there's the facts for your decision.

So i wondering which one did you find better i notice alot people are using the SHARP IR how good does it work?
Also i am not so familiar with Sharp IR how does it out put Signal is it linear i am guessing it only linear within a certain range?

Sharp IRs come in a variety between Serial output or Analog Voltage output, same with some Ultra-Sonics.

,Nick(macdad-)

[wany]

Well since i am doing object avoidance ultrasonic sounds a bit better but yea i also have the problem with power consumption.
If i was going to do ultrasonic i was thinking of building my own ultrasonic device as it would be far cheaper than buying one. I found schematic on building them. But then when i build it i would have to see accurate it is and of course it won't be near as good as just to buy one.
What i was thinking is that i could probably buy the sharp IR and build the ultra sonic and use both of them. That is only use ultra-Sonic at certain time intervals therefore use less power overall and i get a better idea of what really around the robot.
So therefore if ultrasonic pick up something it would take priority over the IR code for controlling the robot.
But then again i could also have the IR scan a certain range instead of just point in one direction.

I've even though about use an array of IR  that i could build and make them look forward then i would get a better understand of what in front the robot.

[macdad-]

Integration of both devices is a good way to solve it. Prioritizing the Ultra-Sonic is an excellent way to cut power draw. You could introduce a basic learning algorithm for your bot by having it do a test run, going back and forth from IR to Ultra-Sonic. Then the robot can compare the number of detects by the IR to the number of detects by the Ultra-Sonic thereby prioritizing the sensors for the robot to use in that particular environment.

[madsci1016]

I wanted to add a few to the list from personal experience.

Both are subject to interference. My Senior design class was to build robots that travel over imitation grass (AstroTurf) and all the groups that used Ultra Sonic failed miserably because the plastic grass would reflect sound back and distort all distance readings even with the tightest FOV sensor.

So I think this statement

Pros of Ultra-Sonic:

    * Much more resilient to interference.

is a little misleading. All sensors have some form of interference.

And on the flip side, IR sensors have issues with natural sunlight and other forms of IR emitters like security motion detectors.

[wany]

UMM good idea. I was thinking more along the lines of. Very few seconds just take a snap shot of whats around the robot using the Ultrasonic device and compare it to what the robot sees at the same time it uses the IR. IF the ultrasonic sees something that the IR doesn't the Ultrasonic device would take priority to tell the robot what to do something different. That way it uses less power because the ultra sonic only comes on every few seconds.
 
Another adaptive look i was thinking about was using a photo sensor if the lighting is above a certain value that i would would affect the IR a-lot i would just turn off the IR and use the ultrasonic. (but i might not even bother with this)

I was thinking of using dumper switches in-case something goes wrong and it does hit a wall it could just reverse turn around and start over.

[wany]

And on the flip side, IR sensors have issues with natural sunlight and other forms of IR emitters like security motion detectors.

Well security motion detectors not really a problem i would say, don't see my self bumping into those . Well that why i was thinking if the lighting at a certain degree i could just not use the IR at all. But then again my robot would not be going out doors. So don't think i'll have to worry about this

[wany]

On another note. I was thinking why not build an Autonomous helicopter. I won't actually build the helicopter but rather. Buy a RC electrical controlled helicopter and used the same idea as before.

[Soeren]

Hi,

I was thinking why not build an Autonomous helicopter.

Perhaps to not make life any harder than it has to be 
It will take so much more and it's better with a finished 'bot, than high flying thoughts that don't come off the ground.

[wany]

Yea but i like to challenge my self. I like doing things that i have no idea how to do that is to say that if i can figure out how stuff going to work before actually doing it then i am not really challenging my self.

[knossos]

To add another comparison to the ultrasonic/ir debate, the classic door/chair leg problems.  An ultrasonic may miss a doorway that it should go through due to its wide FOV, and a IR may miss a chair leg due to its narrow FOV.

[wany]

Could you explain how it would Miss the doorwa. Because if the ultrasound pass through the door and doesn't bounce back means it should go through. Unless you mean it will bounce off the side of the door and make it think there a wall there.

[rbtying]

Could you explain how it would Miss the doorwa. [...] Unless you mean it will bounce off the side of the door and make it think there a wall there.

Exactly right - ultrasonic rangefinders usually have an angle of detection of about 15 degrees, and doorways aren't that wide - thus, at a distance, it can easily detect the side of a doorway rather than the open space.  Remember that the sensor will return the range to the closest object inside the sensing area, whether that be the ground, a wall, or an obstacle.

[wany]

Well if you use IR and UltraSonic won't you get better resolution of what you seeing if you have the IR scan an angle. And i am guessing after trouble shooting lots of it you can come up with a algorithm to defeat both the chair leg and the door.

[rbtying]

Your issue with scanning is speed - either you stop, scan, and start again, or you have to deal with a moving reference point for the scans (involving some more complex mathematics to translate).  And yes, you could (and should) come up with an algorithm.  Build your robot, post up some psuedocode, and we'll see what we can help you with then.