That's also an algorithm problem. You need to write code to detect stuckness and exploratory (blind) algorithms to get unstuck.
What I want to know is when is someone going to commercialise my 3-sensor sonar? With a transmitter in the centre and two receivers 5cm each side, I can resolve the angular position of the echo to better than 3 degrees with ease. So when your robot is trundling along and sees the leg of a chair, it knows that it's off to the left a bit and can steer around it to the right. It's really *very* effective.
No single sensor will provide valid data in all instances. Far too many variables in the physical world. This is why we humans have five senses, and why most of the better robots incorporate proximity, distance, and touch detection as basic means to navigate an environment.
While I'm not saying the EZ1 isn't a welcome addition to the collection of inexpensive ultrasonic rangers available for amateur robots, wouldn't you say the product might also not exist had your daughter's robot also included a simple IR proximity sensor (Sharp OPIC type) or even whisker switch?
You mention "wouldn't you say the product might also not exist had your daughter's robot also included a simple IR proximity sensor (Sharp OPIC type) or even whisker switch?"
True, if my daughter's robot had one of these other sensors I may not have had the passion to put this effort into this project. I could have pulled the sensor away from the front of the bot, I could have switched to IR, or I could have even used a simple bump sensor. I just felt that it was possible to detect to the front sensor face, even pressing against the front sensor face. To me, wining robot contests was not the challenge here, but making a better sensor was.
What is funny, is that it was a design goal (hobby based) to detect to the front face, even pressing against a wall. This was not possible with dual sensor models, but with dual sonic sensors I built, I was able to detect closer than 1mm, but when I moved closer, the sensor stopped detecting the presence of the wall. My repeated measurements of all the ultrasonic sensors mentioned and others showed that 1mm was much better than the others, but I was doing this as a hobby, and as a design challenge to myself, so I then moved back to the single sensor design. I wanted to detect up to and touching the front sensor face. The math and circuit simulator showed this was possible for a single sensor model.
History (cutting out previous to 2004) In early 2004 I built up the first single sensor PCB. In early 2004, I found it very difficult to get useable up close data from a single sensor model. From spring of 2004 though February 2006 I built up about six separate dual sensor models at a cost upwards of a few thousand dollars. (I ordered PCBs, sensors, more sensors, design tools, ect.) Each dual sensor model was working better than the previous. This initial work was done as a hobby. I am a high end electronics and software engineer in my day job, where I both write design instructions and participate and do the actual design/prototype work, so I can do this type of thing as a hobby and actually enjoy it! (Even buzz, like this form, is kind of fun.) The MaxSonar-EZ1 came out of a six week stay away from my family (for day job related reasons) when I would work this problem, every night from about 5:30 to midnight. It was here during midnight hours that I cracked the door to the actual problems and found solutions. Not easy, but kind of fun, in a challenging sort of way.
It's your decision for your product, but I'd rather have a blind spot than the required startup distance. I can measure distances that are right near the robot with a switch.
That does make sense, even though it requires an output pin to deal with that, though that is much easier than the circuitry that I was thinking of. For some reason I thought that the calibration happened at power-up.
Clifford, I found your web page. Your circuit and observations are very cool! I did something like this in 1994 & still have the electronics and paperwork sitting around somewhere. At the time the cost of using a
68HC11 and multiple opamps (without using spice to design) made the design somewhat costly. Lindsay Kleeman's work was (and still is) very good reference material. His very important work showed that this
3D stuff worked and could be very accurate.
PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.