914 PC Bots Forum  


<< Start < Prev 1 2 Next > End >>
jamesbruton
Admin

Admin
Posts: 683
graph
Karma: 8  
Sonar Scanner - 2007/07/14 02:07 Another idea for a project:

It would be great to get hold of a laser scanner/range finder to basically scan the room and essentially 'easily' map the environment, however these are quite expensive.

I spotted these - there are various different versions/makes/model available, but this one is 'new' and has a long range/low power. It struck me that although it only measures distance in one direction, you could physically rotate it (or an array of them) to scan a 180 degree, or further, sweep of the environment.

As long as you knew where the thing was pointing when you got the reading you could map out where objects are in the room. Perhaps even just using a simple R/C servo to rotate it one step at a time, you'd be able to get some useful data provided the robot carrying it was stationary during the reading - basically it 'stops and has a look around'

If you had a known map of the environment and perhaps 4 of these looking forward/backward/left/right you could use them for complete navigation - then rotate/scan the whole assembly 90 degrees to fill in the gaps.

Has anyone taken this 'scanning' approach with Sonar, or any other form of longer distance range finder?

Post edited by: jamesbruton, at: 2007/07/13 18:08
  | | The administrator has disabled public write access.
aibo_rescue
User

Platinum Boarder
Posts: 155
graphgraph
Karma: 2  
Re:Sonar Scanner - 2007/07/14 02:18 You can use the inexpensive IR sensors or cheap sonar rangers to scan an average room with up to 1 cm precision. I would think that expensive laser scanners/rangers are best for outdoors and longer distances. As we know Ernie was/is working on mapping and location functionality within WBOS using built-in IR sensors.

KISS ... ... even the most expensive sensor/scanner/gizmo is not going to help without practical/integrated/working SOFTWARE.

Post edited by: aibo_rescue, at: 2007/07/13 18:26
Best ... Richard
  | | The administrator has disabled public write access.
jamesbruton
Admin

Admin
Posts: 683
graph
Karma: 8  
Re:Sonar Scanner - 2007/07/14 02:31 Yep - agreed, obviously though the IR sensors do 80cm Max whereas the sonars do 6m - so you can get lot more data about the room in one go without having to move about bit by bit.

Also, if you use sensors fixed onto the robot then you have to rely on rotating using the wheels, dead reckoning, and compensating for wheel slip etc. But making a accurate scanner that works when the robot is stationary would remove all these issues - you could even use it to help rotate the robot accurately based on it's readings of the nearest object (obviously the same is partly true for fixed IR sensors if the object is near enough and big enough).

Post edited by: jamesbruton, at: 2007/07/13 18:44
  | | The administrator has disabled public write access.
jamesbruton
Admin

Admin
Posts: 683
graph
Karma: 8  
Re:Sonar Scanner - 2007/07/14 03:13 In terms of software I don't think it would be too hard to poulate a 2D array with 1's or 0's depending if the sonar sees something there.

Of course supporting software is required for any hardware accessory/project - if it were an arm project then the task would also be quite challenging, unless it were for telepresence only.
  | | The administrator has disabled public write access.
c6jones720
User

Platinum Boarder
Posts: 350
graphgraph
Karma: 3  
Re:Sonar Scanner - 2007/07/14 19:29 You guys hit the nail on the head there. At uni they always said dont rely on only ultrasonic or infrared, instead use the two together. What you're saying about scanning areas mechanically with sonar is exactly what I did here:

http://www.robotics.com/robomenu/mrhead.html

Exactly like you said the robot outputted a 2 dimensional array of ones or zeros. Its is preferable to have an accurate map of analogue distance instead though.

theres a book called "Robots Androids and Animatrons" by Robert Iovine where he uses a PIC programmed in BASIC (just like picaxe) and has written a fuzzy logic algorithm to use a servo to scan the area with a single ultrasonic sensor. It is exactly what you said. the only thing thats missing is a way of getting the information from the sensor processor into the computer - USB/serial/parallel etc.

The system you gys proposed would still work quite well in pure open loop, because as long as the servo can move and its load is only a tiny sensor, you could reasonably assume that the position you drove it to was accurate.
  | | The administrator has disabled public write access.
jamesbruton
Admin

Admin
Posts: 683
graph
Karma: 8  
Re:Sonar Scanner - 2007/07/14 19:53 I was thinking about building a motorised thing that pops out of a drive bay like a a CD ROM drive tray. Mounted on that would be a rotating sonar sensor that does 180' scan in the smallest steps you could get out of an R/C servo.

Based on what it can map (populate a 2D array) like that you could work out roughly how far you could move forward and where to rotate the Bot 90' or some other angle to scan some more of the room. If the room is smaller than the range of the sonar then you would probably only have to do the scan at 3 points (from 3 corners for instance) to pretty much map the whole thing, of course depending on what obstacles there are in there.

IR sensors could also be used for getting the Bot's rotational angles correct based on proximity to a wall/corner/object as it rotates.
  | | The administrator has disabled public write access.
<< Start < Prev 1 2 Next > End >>