Optical range sensor

[Ben Mitchell]

Has anyone tried the sensor out?

Is it realiable?

Is it accurate? (and to what degree?)

How fragile is it? (would violent shaking/impacts break it?)

Can it be used to "track" goals?

How often does it refresh range information?

Anything interesting to note about it.

Any ways to "spoof" or jam the sensor? (beside blocking it with a robot, duh)

How quickly can it track a goal (degrees per second?) fast, very fast? slow??

Any useful information would be a great help.

Thanks for anyone with any info.

--Ben Mitchell

[Dan 550]

Watch the mpeg in the main directory of the CD given with the kit for a good demo of goal tracking. It's only the speed of the motor that should limit tracking speed.

[Patrik]

Originally posted by Dan_550
Watch the mpeg in the main directory of the CD given with the kit for a good demo of goal tracking. It's only the speed of the motor that should limit tracking speed.

I dunno I would wager that the the new black motors run faster than a control loop can take readings. The specs are at www.bannerengineering.com I think these are known listed as world-beam sensors or something of the like. As for being fragile, they seem pretty sturdy though I wouldn't take a hammer to them.

[s_alaniz]

Having worked with other optical sensors while I was at HP, I know from experience that these kinds are VERY reliable and VERY durable. The rest of the machine will fall down and the optical sensors will still work... HOWEVER!!!! I havn't tested them yet and I'm wondering if there could be:
1) A problem with cross emitter false readings... that is... instead of tracking the reflection from the tape, tracking the emitted light from someone elses sensor. So instead of finding a goal, you find someone elses robot! This depends on how narrow and how polorized the beam can be. I'm checking...
2) (You didn't think there would only be ONE problem did you?) How good they are at rejecting ambiant light sources like sunlight coming through a window in late afternoon. Sensors are usually used in fixed locations with known light conditions, so the wilds of FIRST tents could be a problem... looking into this also....


Best of luck!

Steve Alaniz


"Yeah those robots are cool...but if you can't play Nintendo on them, what good are they?" - Hilary Forth

[Patrik]

Dean or Woody mentioned at kickoff that it was hard to get the light beam in such a way that you could read light output from another sensor (in that case he was referring to jammers).
I think these are IR sensors/emitters ( I didn't notice a strong enough output of visible light on the device to be for detecting). So as long as no one is pointing IR emitters at the field ambient light shouldn't be a problem. All the fields I have been on are well shaded except for maybe main field at nationals last year, but even then I'd have problems believing there'd a be a problem with ambient light. Of course I haven't worked with these sensors except for fooling around with them this year and skimming the specs so I dunno.

[s_alaniz]

Forgot to comment on this one.... the time it takes for tracking is a function of your program which runs a loop about 40 times a second (25 ms) for updates... the latency of the sensor is probably negligible. just remember, (last I heard,) if you exceed 5 delta_t in your program... your robot's gonna shut down anyway so you can't take readings at a longer interval than 125 ms and still have a running robot.
I could be mistaken on the amount of time the program runs... if anyone has different numbers I'd love to hear them.


Best Wishes

Steve Alaniz

"You're saying that we can look forward to robots behaving badly and demanding higher salaries." - Sally Forth

[smokescreen]

It is entirely possible that some decidedly evil team would take some high powered lights, say a car headlamp unit or something, and partially cover the lens with an infrared filter. This would have the effect of the blasting the field with both visible and infrared light. This could screw up the sensors. Of course, none of this would be done intentionally. Said lights would be for special effects and would be tied into a flashing special effects light sequencer.

-smokescreen

[Gauge, Tim E.]

Yes, the optical sensors are quite good this year. Obviously from the demonstration during the main playing field presentation was to generate excitement about these sensors.

However, by running a few simple tests you can figure out how easily they can be distracted.

For instance, if some one were to use some sort of sandcasted mirror, and buffed it to a highly reflective shine. If then these mirrors were mounted on a robot in several places it would cause havoc with all optical sensors being used, by either side.

Also we don't know if any other form of red light will interfere, being from small LEDs say on some random person's yo-yo going up and down on the side of the field, or just simply some other form of red filtered light emitter.

Just some things to think about.

Tim Gauge

*NOTE: Red filtered and inferred lights have quite interesting affects especially in low light situations. Try it, just don't lose what ever the lights are attached to.*

[Patrik]

Mirrors have opposite incident angle reflection, so unless the mirror is perpindicular it fails and only then it will be unlikey that the the mirror will be perp. for more than a few fractions of a second and hence the system will not mistake it for the retroreflective tape that is on the goals. I am sure the program will not be ruled by a one or two false readings.

I don't think that the red light will affect the infrared sensor. I admit I haven't tested against this but I am willing to bet that the output is polarized in such a way that it should prevent most false outputs (except for other sensors). I don't think the "Dean and The Crew" would have let sensors that could be affected by the safety lights (these are red remember).

I could believe heavy sunlight would affect it, it has a lot of light in every wavelength. I know some of the lazer tag sets (IR guns and sensors) are affected by sun, but I still have a hard time believing either of the mentioned scenarios would affect the sensors to any great amount.

[Andy A.]

I'm guessing that the only good way to mess with some one's sensors would be to get in their way.

The things are real good at seeing the retro tape, but not much else. As for stray IR from the sun, lights and video cameras, it would make sense that they are calibrated to a very limited range of input, so they should ignore most of all that. Remember, their range isn't anything spectacular, as we saw at the kick off demo.

Also, a good solid program and quick drivers can overcome even bots trying to get in the way. Remember that you have a lot of control over how the sensors behave, and what they lock onto. Simply pressing a button could set it to track for a diffrent target and cycle through till you get it locked on the one you want. It's all in how you utilize them, get a good mounting, good program, and skilled drivers and you shouldn't have any problems.