range of PING vs the IR detectors for finding objects

gncguy

I am modifying a 30 mph rc car to be a robot using a BOE.· At these speeds I need to see objects a long way off to have time to avoid them. Which is longer range: the PING or the IR detectors such as come with the BOEBOT kit.· Also, do you know what the range of the PING is?

Thanks,

gncguy

Mike Green

Have you looked at the manual for the PING))) or even the webstore page for the product? The information is there.

I don't think either of these (IR or PING) will work for what you want. Neither of them work reliably at the kind of range and speed you need.
The PING's maximum specified range is only 10 feet and it just doesn't work well at that range. Objects would have to be fairly large and planar to get an adequate echo return and there would be a lot of reflections from other objects and the ground (and ceiling or roof if any). IR wouldn't be any better.

This is a difficult problem and is often solved (or attempted) with the use of stereoscopic video and sophisticated image analysis, usually with something like a PC.

You might be able to do it with either ultrasound or IR, but you'd need more power and focused emitters and detectors to produce narrower beams than you could get with the PING))) or simple IR emitter / detector.

Post Edited (Mike Green) : 11/13/2008 4:30:45 AM GMT

Badger

Mike

i may be really of my rocker tonight but www.sparkfun.com has a cmos ir camera under there biometrics listing would that be any good with this problem, or is the speed of processing a road block here. no pun intented

Badger trying to help

Phil

gncguy

Thanks. I am surprised by the short range of the PING. I seem to remember that before lasers got cheap I had an ultrasonic "tape measure" that could measure 20 or 30 feet. It was based on the Paloroid technology used in the SX70 (which I also had). that was of course before the current autofocus technology was developed.

Mike Green

The Polaroid ultrasonic sensor was (is) really quite good at measuring distances. It sweeps through a range of ultrasonic frequencies (called a "chirp"), then analyzes the return echo. You can still get these along with the circuitry to produce a signal proportional to the distance to a people-size object at 20-30 feet. Remember that the actual distance didn't matter that much beyond 10-20 feet because of the depth of field at that distance. An accurate distance reading over the 1 to 10 foot range was much more important and the closer the more accurate it needed to be.

SRLM

30 mph = 44 feet per second

Question: do you really want you robot going 30 mph? Unless you are in a very open place (a parking lot perhaps) you run the risk of doing serious damage. I mounted a ping on a servo for one of my projects, and found that the maximum effective speed to take 7 distance readings of about 160 degree field of view took about a second of time. Quite a distance in a second.

So, if you mount a sensor that can see 44 feet in a pulse, you'll just finish taking measurements when you've covered the distance before you have to take more...

So, say you mount a battery of sensors (or array if that's what you want to call it). The ping takes up to 40 ms to make a measurement, so times that by 4.5 and you 180 ms. Therefore, your vehicle has traveled 180/1000 * 44 feet, or just ~8 feet. You still have to get the other distances and process all the data...

I'm not trying to dissuade you from doing this: rather, the only system that will work at those speeds and distances is light based: either a camera or a laser. The camera requires lots of processing power (see DARPA Urban Challenge) while the laser requires lots of money or hard work to build the module.

gncguy

Ok. You have convinced me not to let the throttle go up to 30 mph. I agree that vision processing requires a PC but a PC requires something the size of the quad rover. I have seen numerous lab based commerical robots that use an array of ultrasonic sensors. They seem to go 10 mph and they are about the cost and size of a quad rover. Unfortunately, the DARPA grand challenge folks aren't giving out their vision processing software so I am stuck. I was actually thinking of running at 5- 10 mph in a big empty parking lot with simple obstacles like traffic cones and paper boxes and have it follow GPS waypoints that I pre-select to be obstacle free. Then the sensor would stop the vehicle if an obstacle appeared in the field of view.

SRLM

Some of the best algorithms are simple: one of the DARPA teams (Stanford if I'm not mistaken) used lasers to determine where objects were, and a pair of cameras to find the color of clear ground: then it knows it can drive on that color.

10 mph is still pretty fast through... Try riding a bike through cones at that speed, and see how much processing power that takes. Still, it's possible with tightly integrated code.

allanlane5

LIDAR units ($3,000 each? Not sure. More than $100, anyway) are commonly used on each corner of a robotic vehicle, like a car. These scan, using laser range-finding, hundreds of feet out in front of the vehicle.

The LIDAR units on a full-sized robotic car are simply ONE input -- usually the MAIN input is some kind of video camera, digitized in real time, with 'features' of the road digitized and recognized. And even then, most navigation is done through GPS 'way points'.

Oh, note the FIRST CMU robotic vehicle prototype had a top speed of like 15 miles per hour -- it couldn't digitize the picture and make decisions any faster than that.

SRLM

Somewhere in my junk box I've got a mirror mounted on a stepper motor from an old printer. Sometime, I'd like to make my own laser ranger that scans on a line. Just an idea though...

gncguy

The ones I looked at were about $5K for robotics applications

SRLM

That's why I want to make my own [noparse]:)[/noparse]

I might be able to get it down to about $50 using cheap parts. Commercially, I've found some for about $2k

Cole Logan

I wonder if something like this could be modified to work.

http://www.amazon.com/CST-Stanley-77-910-Tru-Laser-Measurer/dp/B000BDIRYC/ref=pd_cp_hi_1?pf_rd_p=413863601&pf_rd_s=center-41&pf_rd_t=201&pf_rd_i=B000T7LISM&pf_rd_m=ATVPDKIKX0DER&pf_rd_r=0PNKWR6GGZA663WPAZSX

Its one of those laser range finders that builders use. This one says its acuriate to 1/4 at 100 feet

gncguy

That is a good suggestion.· Anybody have any ideas how to hack into it.· I also wondering about using an IR laser pointer as an illuminator for IR sensors.· My idea is to be looking straight ahead and if something is detected, stop and think about what to do.· This of course would only work in a relatively uncluttered world such as an empty paking lot.· But that is good enough for me.

SRLM

Take a look at this thread. It's exactly what you proposed.

Laser Ranger