27 replies to this topic

[Pip]

Hi,

Here's a new pressure sensor that looks interesting ..

http://www.elektor.c...m_campaign=news

http://www.epsontoyo.../2010/0824.html

3cm resolution (measurable) is very good really I thought.

[Bani Greyling]

do you think it is good enough for automated landing, or would one rather have height reading? Environmental pressure should not change too much in 10 minutes...

wonder what it would cost though

[CheBuzz]

If the 3cm resolution turns out to be true, it would definitely be good enough for landing. Especially if you chose to go with a "differential setup", where you had some groundstation hardware that had the same sensor. This would allow you to have a very accurate altitude reading that could be used for an autoland feature.

[peabody124]

CheBuzz, on 03 September 2010 - 11:36 AM, said:

If the 3cm resolution turns out to be true, it would definitely be good enough for landing. Especially if you chose to go with a "differential setup", where you had some groundstation hardware that had the same sensor. This would allow you to have a very accurate altitude reading that could be used for an autoland feature.

Or just before take off have the home location save the altitude that way. I don't know how similar their readings would be - they're probably good for measuring 3cm changes, but not good for absolute level.

[osnwt]

peabody124, on 03 September 2010 - 12:45 PM, said:

Or just before take off have the home location save the altitude that way. I don't know how similar their readings would be - they're probably good for measuring 3cm changes, but not good for absolute level.

Right, but in differential (two-way) setup the GCS sensor will play the same role as a ground station for DGPS. It knows that it is at zero height, and that correction can be sent to the model to update onboard value taking into account ambient pressure change, temperature change, etc. Their absolute readings may be different, but at least readings on the near-the-same height may be similar (+/- delta which depends on ambient pressure). Or at least we only need to know that "absolute measured on the ground height from the time of take-off was decreased by 137cm". Sending that correction onboard will help to update local readings.

Maybe someone who has two mainboards can check how two sensor readings correlate in time.

BTW, maybe bosch sensor may be used the same way. We only have to check how it is temperature-dependent. As I see, powering-on the main board increases measured temperature to 3-5 degree almost in seconds.

[peabody124]

I see - you mean watch the GCS delta to see what the ambient temperature effects were and take that out of the Flight delta.

[osnwt]

Yes, but not only temperature effects - local pressure changes as well.

[Andrew]

Has pipbee hardware been finalised?
That could provide a convenient platform for a gcs pressure sensor and an in flight one as well...
And without having an extra hardware module to worry about just for pressure corrections.

[Pip]

There's no room left on the modem board for extra components unfortunately.

I really wanted to keep the signal line RF filters on the board but had to remove them and use both sides of the board for components to fit the modem PCB onto the same footprint as the AHRS and GPS boards. It's not a lot of space really to work with, A bigger footprint for all the boards would have been nice (especially for prototype work) but say-la-vee.

[John]

Hi,

Also note that the pressure rapidly goes up close to the ground in case of a helicopter due to the downwash "hitting" the ground. This completely puts your height estimate off when you go close to the ground. This renders landing based on pressure difficult on a helicopter (if not impossible). Of course this effect depends on the size and weight of the helicopter etc.

Exactly for this problem I use an ultrasound sensor for landing on my own HW setup (which is a completely different setup). +- the last 5m I base the height on the ultrasound sensor.

Regards,

John.

[Andrew]

John, on 07 September 2010 - 07:24 PM, said:

Exactly for this problem I use an ultrasound sensor for landing on my own HW setup (which is a completely different setup). +- the last 5m I base the height on the ultrasound sensor.

I was looking at this sonar range finder the other day wondering if it would be useful.
http://www.pololu.co...0_Datasheet.pdf
It has a pwm output option which could make it compatible with a servo input, but I haven't looked at the minimum pulse timings that OP can deal with to confirm.
It does however only output distance greater than 20 cm. which could make the final landing of a heli a little bit of a bump... or it may be in the noise...

Andrew

[Pip]

The single transducer ones do have the 20cm minimum distance.

You can use the twin transducer ones (one transmit transducer and one receive transducer), they don't have the 20cm minimum limit.

[Coptaire.com]

The air turbulence created by the props and specifically close to the ground could make very difficult, or inefficient, to conceive an auto-landing scheme based on ultra sensitive pressure sensor.
By example, on a MK, the pressure sensor, with is far from being as sensitive, must be covered like a mike windshield, to lower the pressure washing.
Hovering on dead leaves under a tree is interesting to visualize this, where leaves are sucked massively at the bottom center of the kopter, not far from the pressure sensor.

Would prefer a more realistic approach is the obstacle avoidance technique using ultrasonic TX/RX pair to measure distance (like a bat). Several distance ranges exists on easy to find robotic shop.

[Andrew]

Coptaire, on 08 November 2010 - 08:00 PM, said:

Would prefer a more realistic approach is the obstacle avoidance technique using ultrasonic TX/RX pair to measure distance (like a bat). Several distance ranges exists on easy to find robotic shop.

I have been looking at this one, http://www.robot-ele...m/srf05tech.htm
it claims ranges from 1cm to 4m.

it has a pulse output which is a bit shorter than a servo pulse, but I think OP will be able to cope with it.
This might allow it to be be plugged directly into a spare RC input channel with the correct scaling values this could be tested very quickly.

It needs an external trigger, I was originally thinking to plug it into a servo output and let that trigger it, but it can only handle a 20Hz rate, not the 50Hz min of OP..
but a quick 555 circuit should suffice.

Andrew

[osnwt]

I bought 6pcs of almost same sensors (with separate Tx and Rx as well).

One was for MK (wanted to add front obstacle avoiding feature but in that project I had no time to add to control loop, so ended with just OSD distance display).
Remaining 5 pcs were ordered for playing with OP-based quad (still wait for Pulso motors which are in transit now).

It really works quite well from 1cm up to 3m (declared range is up to 5m in one place, up to 3m in other, but 3m seems to be maximum reliable, 4m is so-so, 5m is... hmm). And 20Hz maximum rate limitation is due to speed of sound, at 5m it is exactly lowest possible period to get the echo.

But I think that separate MCU with I2C could be better interface for it.
In addition, up to 5 units can be used (4 directions and downwards).

[Coptaire.com]

This idea paves the way for a 5 sensors module I2C capable, a collision avoidance plugin, quite useful.
prolonging the superb work of FrederickG on the MK-Serial plugin, and the I2C plugin project, this could be a real good idea that some of us join effort to develop this addon?

Except if David has a specialty surprise (like yesterday CoptaireControl, sorry, CopterControl board!) concerning this kind of application !

[Andrew]

Do you have a link to the specs for your hardware?

[Andrew]

Andrew, on 09 November 2010 - 02:08 AM, said:

I have been looking at this one, http://www.robot-ele...m/srf05tech.htm
it claims ranges from 1cm to 4m.

it has a pulse output which is a bit shorter than a servo pulse, but I think OP will be able to cope with it.
This might allow it to be be plugged directly into a spare RC input channel with the correct scaling values this could be tested very quickly.

It needs an external trigger, I was originally thinking to plug it into a servo output and let that trigger it, but it can only handle a 20Hz rate, not the 50Hz min of OP..
but a quick 555 circuit should suffice.

Andrew

I purchased one of the above,
connected its trigger input to an OP output (50Hz trigger is above spec but tried it anyway...)
Connected the output pulse to OP input ch 7 and set manual control so Accessory1 pointed to that input...

Ran up the GCS and looked at the current value for Accessory1, and sure enough it was proportional to the aircraft height above the floor....

This shows promise for an easy approach to get height above ground into OP and maybe automated landing...

[Lew Payne]

Coptaire, on 08 November 2010 - 08:00 PM, said:

The air turbulence created by the props and specifically close to the ground could make very difficult, or inefficient, to conceive an auto-landing scheme based on ultra sensitive pressure sensor.

Establish a "rate of descent" while still far away enough from the ground that the pressure sensor remains accurate. Adjust that "rate of descent" to ensure a soft landing (again, prior to getting close to the ground). Continue with the same thrust parameters, and touch down. Your accelerometers will likely notice the landing, or you can compute rate of descent and subtract from prior altitude at the time you established the descent vector.

Am I missing something here?

[jesboxx]

The pressure sensor could be used to detect the ground effect and help staying close to ground. Low-altitude sniffing

Well, the absolute altitude and speed is hard to tell without sonar or something. Smooth landing might be hard to achieve.

Should the pressure sensor be on a periscope-like structure above rotor turbulence for better accuracy? Or below to sense the ground effect?