sensing a wooden block's position

How would one go about sensing a small wooden block's position (1"x1"x1") with a robot (robot dimensions will be less than 1'x1'x1') ? Initially, I was thinking I could use some type of ultrasonic transducer and process its output. However, after reading a few datasheets, it seems that the field of view of an ultrasonic transducer is much bigger than the actual block. This would make it almost useless to try and locate several of these blocks. I am just looking for general ideas on how this could be done or if it is even possible. I can flesh these ideas out later, just looking for a quick brainstorming type of session on these boards.

Thanks.

-Whiskers (cheapest, smallest, simplest, most reliable, and can't be fooled)

-a 235kHz sonar (shorter range, narrower FOV, and...if the angle of approach relative to the surface of the block is too great the echo will reflect in such a way that it will never reach your receiver, rendering it invisible (with lower frequency sonars that have wider FOV, the echo will still reach your receiver since the angle is wider and more likely to intersect with your receiver).
https://www.robot-electronics.co.uk/htm/srf235tech.htm

SHarp Single Beam IR Sensors (15 degree angle?)
-Sharp GP2D12 (10-80cm)
-Sharp GP2D120(4-30cm)

SHarp Multi-beam IR Sensors (5 beams each with a 5 degree FOV)
-GP2Y3A001K0F (4-30cm)
-GP2Y3A002K0F (20-150cm)
-GP2Y3A003K0F (40-300cm)
THese sensors also cost 5x as much as the single beam IR sensors but have a narrower field of view per beam

Is your robot going to be approaching the block? Because a wide FOV isn't so bad...it allows you to tell if a block is somewhere in front of you and moving closer towards it will increase your accuracy, whereas a narrow FOV sensor might miss it altogether if it is far away (depending on your pan-tilt mechanism and other factors of course). In which case....
https://www.robot-electronics.co.uk/htm/srf02tech.htm

All the sensors mentioned, except for whiskers, though have a dimension that is a bit larger than 1".

Just thought I'd add my two cents based on my experience with the cheaper Sharp IR sensors. If you use these, you'll want to be aware of what these can do and what you need to find your blocks.

These sensors have useless readings if they're within 5cm to 10cm (depending on the range of the sensor) of the object they're trying to detect. This can be overcome by placing the sensor far enough back on the robot so that the useless range doesn't "overhang."

Will your blocks all be on a single plane, i.e. on a flat floor that your robot will be rolling along? If so, then I would go for one of the Sharp sensors with the 5cm to 30cm range, especially given the size of your blocks. The smaller the detection range of these sensors, the better the resolution. You could point the sensor at the ground, extending it out on an arm from the robot kind of like a mindsweeper. The field of vision for these sensors is cone shaped, I think to about 10cm diameter at their maximum range. If you can get the sensor positioned in that sweet spot above the floor just outside the useless range mentioned above, then the field of vision would be reduced to, I'm guessing here, no greater than an inch or so. Provided the blocks are more than an inch away from each other, you should be able to find them individually. For extra coolness, you could mount the sensor on a spinning arm rotating around the robot.

By the way, the Sharp sensors' refresh rates are approximately 20Hz, which is fast enough for most robots unless time is a serious competitive factor.

Sorry, I'm making some presumptions about the environment in which you're required to find these blocks - can you describe that in more detail?

Hank Fletcher said:

The field of vision for these sensors is cone shaped, I think to about 10" diameter at their maximum range.

Click to expand...

I just edited a typo: it's about 10cm, not 10".

Yes the blocks will be placed on a level playing field with respect to the robot. These posts are great so far. I will look into these options. Thanks.

To answer your original question then: yes, this can be done. Your cheapest solution then is to use the Sharp GP2D120 using something similar to the method I've already described. They cost about $12 each, and are quite straight-forward to connect to an mcu.

There's actually a fairly abrupt change in the output signal when an object enters the field of vision cone of these sensors (provided the object is within 90% or so of the maximum range). I'd wager that you could get by quite well with a robot design using just one of these sensors, but if you wanted you could use three sensors arranged so that the centre of each cone forms the corner of an equilateral triangle (like three spotlight circles overlapping). You could adjust the position of the sensors into a tight triangle, so that the space where the cones all overlap is quite small. Using this method, you'd be able to quite easily detect individual blocks, even if there were blocks that were positioned closer than 1" together.

Sounds like a cool project! Please let us know how you get along, and if there are any other details we can help you with! Just curious: what are you planning on using for the brain (e.g. mcu) of your robot?