Basically here is the site it is a short range inferred radar that could be used for a failsafe in case your uav gets close to the ground. However it is slightly bulky but that can be easily modified because it was made for parking cars. This includes a servo with an inferred transceiver or receiver and transmitter separate, i haven't really looked into it very far. If you are board and want to serf the web go to "Hackaday.com". Thats where i found the range finder and is a great place to hack if you want to.
@Chris: I'd like to update on tests I made today, as well indoor and outdoor. I used an Arduino Pro Mini and an MaxSonar EZ1 from Maxbotix. My code was a simple reading of the AN pin of the EZ1 (values 0-511) through an analog arduino port (values 0-1023).
I made a basic empirical indoor calibration at 3.5m, 3m, 2.5m, 2m, 1.5m, 1m and 0.5m distances to find out a good conversion constant for distances in centimers.
I then applied the empirical calibration result: float distance = rangeVal / 2 * 2.6. This basically cuts the analog arduino reading in half to recover the 0-511 value range from the sensor, and multiplies it to get centimeter distance. I didn't use the default 2.54cm/inch conversion which did not give an accurate cm distance in my opinion. At the end, my readings were ±5cm correct for the 0-3.5m range. They were better in the 1.5-3.5m range with ±1-3cm error than in the lower range 0-1.5m with ±4-6cm error. In the future, I 'll calculate my distance as an integer to keep processing low.
Finally, I took my laptop in one hand and sensor on arduino connected to ftdi breakout board out in the garden to check out different dry surface readings. Everything seemed fine and correct on short grass (snow melted only two weeks ago), gravel, brushes, cobbles, bare earth, wood, evergreen tree branches (tuya), etc. The biggest problem during this outdoor test was reading the laptop screen and the cold!
For now I am overall impressed with the accuracy of this sensor. I did not have noticeable noise readings or other strange data and the sensor range (after distance conversion) is 12-652cm. i definitely plan to do more test when grass grows taller, but I am already quite convinced my EZ1 sensor could be implemented in an auto landing project on many different surfaces.
A question that came to my mind is whether the MaxSonar you had could maybe have something wrong.
Thanks for the info, Chris. I read you had not so good results (with an EZ3, if I remember). I am also concerned about the grass readings and I'll log EZ1 data coupled to GPS readings in manual landings to check this out. For now, I do not feel in investing in the 4x more expensive sensor. For me, it is as much about learning from the process than actually auto landing on grass! And even if I loose a little time on this, I have fun doing and trying. Maybe I'll finally only make tests for auto landing on concrete strip.
An intermediate question I need to solve is proper placement of the sensor in the airframe. Initially, I wanted it under the nose, between motor and landing gear. Could the proximity of the swinging prop and brushless/propeller noise be a problem for the sensor? I'll probably log some data in different settings before fixing it.
Reto, I think you'll have trouble with the EZ-1 for this purpose. In our testing, it doesn't give reliable results on grass, ground or even carpets--basically anything other than a hard floor. I think you will find it useless outdoors. You might have better luck with the long distance one, but it's expensive and big.
This is an old post, but I want to mention I received a Maxbotix EZ1 from Sparkfun recently (0-255 inches range) and made some basic tests with Arduino and the EZ1 analog output. Works well for now.
I intend to build it into my airframe for automated landing. I'll mix two methods: GPS and range finder.
A landing descent would be preprogrammed with at least 7 waypoints:
- wp993: descent initialization point at 30m agl (GPS altitude control)
- wp994: approach axis begin at 30m agl (GPS altitude control)
- wp995: middle approach at 15m agl (GPS altitude control)
- wp996: final approach at 6m agl (GPS horizontal location, range finder takes over altitude control at 6.45m agl)
- wp997: landing strip begin at 2m agl (GPS horizontal location, range finder altitude)
- wp998: throttle down and rounding point at 0.5m agl (GPS horizontal location, range finder altitude)
- wp999: touchdown and throttle idle (GPS location)
Since GPS altitude is not accurate, I'll implement two "emergency" scenarios for bad approaches:
1) if GPS approach makes descent too short, range finder is holding 6m altitude until waypoint 997 is reached and then follows normal procedure
2) if GPS approach makes descent too long (overshoot of waypoint 996 with no range finder signal or out of range signal), then airframe would gain altitude again (following a few preprogrammed take off waypoints) and go to first landing approach waypoint 993 for a new descent.
Maybe there are some people who already have experienced this kind of automatic landing approach. Any remark/objection/input would be welcome. I still have to figure out the optimal horizontal distances between approach waypoints. I'll be able to figure them out only in a few weeks, when my Kadet Senior is ready for maiden.
Marcus, I just wrapped a project of using a ultrasonic sensor by maxbotics that has a 255 inch range. Works pretty well but required me to program a filter to get rid of some noise variance.
I'm trying to work with an ultrasonic sensor that speaks I2C.
This should be much smaller and even have more range
then the IR-system.
Pro: It is independent of light and reflection -properties
Con: ...ultrasonic and dogs. Works inly in one direction at a time
well i have been reading that sometimes the sound waves will bounce in different directions and give messed up results
so i think inferred is more reliable but it is only for short range. like an emergency
Comments
I made a basic empirical indoor calibration at 3.5m, 3m, 2.5m, 2m, 1.5m, 1m and 0.5m distances to find out a good conversion constant for distances in centimers.
I then applied the empirical calibration result: float distance = rangeVal / 2 * 2.6. This basically cuts the analog arduino reading in half to recover the 0-511 value range from the sensor, and multiplies it to get centimeter distance. I didn't use the default 2.54cm/inch conversion which did not give an accurate cm distance in my opinion. At the end, my readings were ±5cm correct for the 0-3.5m range. They were better in the 1.5-3.5m range with ±1-3cm error than in the lower range 0-1.5m with ±4-6cm error. In the future, I 'll calculate my distance as an integer to keep processing low.
Finally, I took my laptop in one hand and sensor on arduino connected to ftdi breakout board out in the garden to check out different dry surface readings. Everything seemed fine and correct on short grass (snow melted only two weeks ago), gravel, brushes, cobbles, bare earth, wood, evergreen tree branches (tuya), etc. The biggest problem during this outdoor test was reading the laptop screen and the cold!
For now I am overall impressed with the accuracy of this sensor. I did not have noticeable noise readings or other strange data and the sensor range (after distance conversion) is 12-652cm. i definitely plan to do more test when grass grows taller, but I am already quite convinced my EZ1 sensor could be implemented in an auto landing project on many different surfaces.
A question that came to my mind is whether the MaxSonar you had could maybe have something wrong.
An intermediate question I need to solve is proper placement of the sensor in the airframe. Initially, I wanted it under the nose, between motor and landing gear. Could the proximity of the swinging prop and brushless/propeller noise be a problem for the sensor? I'll probably log some data in different settings before fixing it.
I intend to build it into my airframe for automated landing. I'll mix two methods: GPS and range finder.
A landing descent would be preprogrammed with at least 7 waypoints:
- wp993: descent initialization point at 30m agl (GPS altitude control)
- wp994: approach axis begin at 30m agl (GPS altitude control)
- wp995: middle approach at 15m agl (GPS altitude control)
- wp996: final approach at 6m agl (GPS horizontal location, range finder takes over altitude control at 6.45m agl)
- wp997: landing strip begin at 2m agl (GPS horizontal location, range finder altitude)
- wp998: throttle down and rounding point at 0.5m agl (GPS horizontal location, range finder altitude)
- wp999: touchdown and throttle idle (GPS location)
Since GPS altitude is not accurate, I'll implement two "emergency" scenarios for bad approaches:
1) if GPS approach makes descent too short, range finder is holding 6m altitude until waypoint 997 is reached and then follows normal procedure
2) if GPS approach makes descent too long (overshoot of waypoint 996 with no range finder signal or out of range signal), then airframe would gain altitude again (following a few preprogrammed take off waypoints) and go to first landing approach waypoint 993 for a new descent.
Maybe there are some people who already have experienced this kind of automatic landing approach. Any remark/objection/input would be welcome. I still have to figure out the optimal horizontal distances between approach waypoints. I'll be able to figure them out only in a few weeks, when my Kadet Senior is ready for maiden.
This should be much smaller and even have more range
then the IR-system.
Pro: It is independent of light and reflection -properties
Con: ...ultrasonic and dogs. Works inly in one direction at a time
so i think inferred is more reliable but it is only for short range. like an emergency
I dont understand though, what does this do that a much lighter and smaller ultrasound rangefinder won't?
-
1
-
2
of 2 Next