0

Sorry if I am not following guidelines.

I am a noob trying to build a Robot for learning purposes - I want to play with navigation algorithms using particular measurements - distance to the obstacle along a straight line. I know there are many other ways to skin the cat, but I want it this way. I have been looking online for a sensor that would help me, but there are so many options and none of the are ideal. My requirements:

  • Can sense distance between 1 foot (30 cm) and and 15 feet (5m); ideally between 10 cm and 10 meters (I'll accept infinity as the upper bound :) ).
  • Is directional, not shotgun-like. Ideally, at 10 feet (3 meters), the sensor would bounce back if the bounding sphere the size of a tennis ball would contain a wall. (Hopefully this makes sense - this translates to a cone with a certain angle).
  • Is small - no bigger than a human eyeball including casing.
  • Is low voltage/power - so that I can make it draw power from a usb cable, perhaps adding some sort of converter, but hopefully not.
  • Functions in complete dark, or near-dark.
  • Is under $20 per piece (I might buy 10 in bulk).
  • has distance resolution of at least 10 cm (4 inches); higher is better.

It does not have to be sample the data very fast; even 10 times per second will do. Hopefully this question is withing the guidelines. Again sorry, I do not have previous robotics experience.

Job
  • 101
  • 3
  • What have you tried? Ultrasonics probably OK except for wide angle. 4" resolution is very "low". Should be easy :-). OTS tape measures MAY be an easy path. – Russell McMahon Apr 13 '13 at 04:36
  • I'd vote ultrasonics too, but size might be a problem: most US modules use separate transmitter and receiver, each ~ 1 cm in diameter, with some distance in between. And US tend to be rather wide-beam. Some sort of barrel around sender and/or transmitter might help. – Wouter van Ooijen Apr 13 '13 at 10:36
  • An alternative, that still won't meet your requirements, could be the Sharp range of IR sensors. They come in different flavours, but could be worth considering. – Outsider Apr 13 '13 at 12:35
  • For a 10 centimeter object detection precision (larger than the tennis ball mentioned) at a distance of 3 meters, you require an s-angle of 6 degrees. That is rather narrow for ultrasonic sensors, are you sure you need that precision? Laser based LIDAR or even DIY hobby-laser-based sensors could achieve this, but the cost goes up significantly. See [this project](http://coolarduino.wordpress.com/2011/12/07/arduino-laser-3d-tracking-range-finder/) for ideas. – Anindo Ghosh Apr 13 '13 at 13:03

1 Answers1

1

I think you might have to compromise on at least one of those requirements. The normal way of doing this sort of thing on small robots is ultrasound, e.g. http://dx.com/p/ultrasonic-sensor-distance-measuring-module-138563

That fails your "directionality" and probably "size" requirement, but is cheap and effective. There are smaller, narrower ones: https://www.sparkfun.com/products/639

If directionality is really important, I don't think there's any substitute for doing it with a laser. There's a discussion here: Any published designs for time-of-flight laser distance measurement , but the engineering involved is not simple and I can't find any modules implementing it. The "time of flight resonance" technique looks best to me, although the round-trip time for 15cm is 1ns => 1GHz signal. That's going to make for tricky PCB design and isn't feasible for a beginner.

Community
  • 1
pjc50
  • 46,540
  • 4
  • 64
  • 126
  • There's the option of laser-based angular measurement, very similar to the process used in the Sharp IR proximity sensors... The link on my other comment has a project implementing this by laser + webcam. – Anindo Ghosh Apr 13 '13 at 13:46
  • Thank you! The one from sparkfun.com looks promising. There is also the EZ4 version that supposeddly has a narrower beam - https://www.sparkfun.com/products/8504 Price is the easiest option to compromise on. – Job Apr 13 '13 at 16:21