ScubaBob
Well-Known Member
I don't trust any sensors when over water. I trust my eyes and what the camera is showing me.
Over water flight
- Thread starter mstef66
- Start date
black_magic100
Well-Known Member
- Joined
- Oct 20, 2017
- Messages
- 503
- Reactions
- 146
- Age
- 27
So far as far as flying in Cancun what do I need to know. Do I tell authorities at the airport its just a toy or just a camera that flies. If I do any flying at my resort I am assuming I should talk to any of the authorities on duty to make sure as I assume flying near resorts is most likely not allowed. Does your drone have to be in LOS in cancun?
Just this morning i flew mine about 1/3 mile across the bay at 14'. Regular GPS mode with sensors. No issues at all.
ScubaBob
Well-Known Member
I have never had a problem bringing my MP into Cancun or bringing ones for others so long as I only bring one in at a time. Otherwise, should you be questioned by the customs officer (highly unlikely) they may think you are selling the second unit. I just put the MP into my carry bag and have no problems. I even brought back Goggles last week without a problem. No one I know tells any governmental official in Mexico that they have a drone. No need to do so as far as I know. I fly on the beach in the Hotel Zone almost every day and have not had any problems. I send my MP over 1,000 meters away so it is clearly no LOS. The only area where the drones are restricted are in the marine parks, especially where there are whale sharks (during May through October). You need a special permit from Semarnat for that and I am working through the process to get that. It is a 5 day permit for 200 pesos (about $12USD). If they catch you flying a drone over the whale sharks they will confiscate the drone. But on the other side of the coin, the marine police are rarely out supervising anything.
Too bad the manual doesn't address this issue well.
I would think that moving waves, ripples and floating debris could trick optical positioning into inducing the Mavic to follow the drift and discount GPS accuracy. Of course GPS can't lock to the inches or feet accuracy that visual can and a conflict would not register immediately. Inertial sensors may more readily be doubted if the optical position is assumed correct.
I had always surmised that relying on optical precision altitude height over water could be tricked by the Mavic seeing the bottom below the surface of the water and interpreting its altitude as being in reference to that rather than the surface. If the Mavic is four feet above ground and flies over water six feet deep where it can see objects at the bottom of the water, the visual system will want to fly it down to two feet below the surface.
Ultrasonic altitude should still work as should the barometer, but the barometer is incremental (like GPS vs visual precision and probably 'smoothed') and there is no way of knowing how DJI implemented their sensor priority between visual and ultrasonic.
If the manual were more clear it would be easier to make better decisions.
I would think that moving waves, ripples and floating debris could trick optical positioning into inducing the Mavic to follow the drift and discount GPS accuracy. Of course GPS can't lock to the inches or feet accuracy that visual can and a conflict would not register immediately. Inertial sensors may more readily be doubted if the optical position is assumed correct.
I had always surmised that relying on optical precision altitude height over water could be tricked by the Mavic seeing the bottom below the surface of the water and interpreting its altitude as being in reference to that rather than the surface. If the Mavic is four feet above ground and flies over water six feet deep where it can see objects at the bottom of the water, the visual system will want to fly it down to two feet below the surface.
Ultrasonic altitude should still work as should the barometer, but the barometer is incremental (like GPS vs visual precision and probably 'smoothed') and there is no way of knowing how DJI implemented their sensor priority between visual and ultrasonic.
If the manual were more clear it would be easier to make better decisions.
ScubaBob
Well-Known Member
It is pretty hard for a human to see the bottom of a river or lake if they are in a canoe or boat and the water is over 2 meters deep. I am not sure the Mavic sensors are more sensitive than a human eye. Nonetheless, when I am flying low over water I always rely on visual observations rather than anything the drone sensors tell me.
The reason there are two forward facing obstacle avoidance cameras set some distance apart and two downward facing cameras also set some distance apart is to allow the images from each pair to be compared. If a clear feature is discernable in each image and is found in the same position in both cameras, the object is far away. If that feature is 'x' pixels to the right of that in the other sensor, that object is closer. If it becomes twice as many pixels in one sensor than the other, the object is calculated to be even closer. Objects at great distances remain aligned and the closer objects get, the further out of alignment they become. Theoretical accuracy can improve with increasing the distance between each of a pair of cameras. The cameras cannot determine the distance to something undifferentiated along the axis between the cameras. If the Mavic is flying level and approaches a completely plain flatly illuminated wall with a 2" wide horizont black stripe, it cannot determine the distance to the wall. If the stripe is vertical, the Mavic can easily detect the distance and warn a user.
Drone sensors don't tell us anything. They are inputs into an algorithm that balances, prioritizes and determines which reading upon which to rely and the degree of reliance. Given a clear edge image, optical rangefinding is highly precise and accurate and not subject to drift like barometric based altitude determinations. I'm not as clear on ultrasonic ranging but suspect it might be subject to being confused by angular reflections and surfaces that absorb rather than reflect the acoustic pressure wave. I do know that if I was landing a Mavic on a glass table top 3 feet over a floor, I'd want the ultrasonic sensors and not the optical ones to be determining altitude.
There are many places where you may not see items with distinct edges at the bottom of water while canoeing. I suspect the Mavic altitude sensors not to be readily tricked into determining a false altitude in that situation. If the surface is visually indiscrimate, the optical sensors cannot provide meaningful input and only the barometer and ultrasonic sensors can be considered. If there is a vertical accelerometer, ic can inform as to a change in rate of vertical displacement but not altitude.
I use a kayak and paddleboard and I find there are many streams and lakes where the bottom is visible for a distance. Unlike a person, the Mavic is looking straight down and reflections at a perpendicular angle through the shadow of the Mavic at a low height would be minimized. A silver lure might be visible at a great depth.
Have you ever seen the lane markings at the bottom of a swimming pool? It is a scenario where water will cause one of the inputs in the altitude determination algorithm to be inconsistent with the others identical to the glass table example.
Drone sensors don't tell us anything. They are inputs into an algorithm that balances, prioritizes and determines which reading upon which to rely and the degree of reliance. Given a clear edge image, optical rangefinding is highly precise and accurate and not subject to drift like barometric based altitude determinations. I'm not as clear on ultrasonic ranging but suspect it might be subject to being confused by angular reflections and surfaces that absorb rather than reflect the acoustic pressure wave. I do know that if I was landing a Mavic on a glass table top 3 feet over a floor, I'd want the ultrasonic sensors and not the optical ones to be determining altitude.
There are many places where you may not see items with distinct edges at the bottom of water while canoeing. I suspect the Mavic altitude sensors not to be readily tricked into determining a false altitude in that situation. If the surface is visually indiscrimate, the optical sensors cannot provide meaningful input and only the barometer and ultrasonic sensors can be considered. If there is a vertical accelerometer, ic can inform as to a change in rate of vertical displacement but not altitude.
I use a kayak and paddleboard and I find there are many streams and lakes where the bottom is visible for a distance. Unlike a person, the Mavic is looking straight down and reflections at a perpendicular angle through the shadow of the Mavic at a low height would be minimized. A silver lure might be visible at a great depth.
Have you ever seen the lane markings at the bottom of a swimming pool? It is a scenario where water will cause one of the inputs in the altitude determination algorithm to be inconsistent with the others identical to the glass table example.
Don’t shut any sensors off. Turn off landing protection only. See how that does for you.
It prioritizes ultrasonic when below 30 feet, but also incorporates IMU to ensure accuracy in the movement adjustments, along with vps.
I've reviewed the manual and specs and am unable to find the basis for these assertions. The vertical vision sensors have a range of 1 to 43 feet while the ultrasonic range appears to be undocumented.
The manual warns about altitude sensing when flying over 'water or transparent surfaces' which validates what I have proposed and mentions ultrasonic sensor readings are compromised over acoustic absorbent materials or angled reflective surfaces. Waves and ripples present angled surfaces that could set up standing acoustic waves or reflect the signal away from the receiver. In typical DJI fashion, there is no mention of a barometric altitude determination, typically used in full scale aircraft, nor of GPS altitude determination. GPS altitude is generally low in accuracy and barometric readings are subject to drift. I have read of DJI Go4 reports of barometer errors and suspect it likely that the Mavic would have such a sensor, particularly because the Go4 app reports stable altitudes beyond the documented vertical vision range and the likely, but undocumented, ultrasonic range. Most available UAV 40 khz ultrasonic sensors appear to have a maximum range of between 300 and 400 centimetres which is a fair bit shy of 30 feet.
Curiously, the distance between the forward facing optical sensors, at about 60mm is less than the vertical sensors at about 82mm so, assuming they use identical hardware, the vertical optical sensors should be capable of greater range and accuracy yet they are documented to be identical.
I cannot understand how anyone can come up with a reasonable explanation of how an IMU could possibly ensure the accuracy of an ultrasonic altitude measurement. It can certainly be used to help distinguish which of the 4 probable altitude inputs might be suspect but it is incapable of providing information about absolute vertical positioning. It is incapable of reporting vertical speed or position.
Kindly provide sources for your assertions.
The information came from DJI engineer on another forum that they actively post information like this. It was quite a while ago, I think about a year ago that this was posted. The IMU can detect movement in ANY direction of the drone. So when the drone goes higher or lower, the IMU can detect this movement. The IMU is used to constantly measure and make small adjustments to stabilize the drone on all axis, and is incorporated into the other systems to assist in overall accuracy and stability.
How the drone works on a technical level isn't going to be in some manual somewhere... It's beyond the normal scope of what a user manual would provide.
CamoGreg
Well-Known Member
- Joined
- Jul 7, 2017
- Messages
- 1,092
- Reactions
- 506
- Age
- 43
People ask questions on forums because they want to know how to safely fly their drones and meaningful information is not available in the manual. Then they are bombarded with inarticulate, uninformed and inaccurate conjecture occasionally backed up with anecdotal 'evidence' and even misinformation provided by a 'DJI engineer'. No engineer with some knowledge of IMUs would assert that an IMU can detect movement unless they also believe in a stationary flat earth.
Forward precision measurement range is given as 2 ft (0.7 m) to 49 ft (15 m); detectable range: 49 ft (15 m) to 98 ft (30 m). The manual says that the Downward Vision System works over a range of 1 to 43 feet and uses both optical and ultrasonic sensors. This conflating of optical and ultrasonic sensing is thrown out the window when they state that the operating environment for the Downward Vision System is surfaces with clear patterns and adequate lighting. They call it Vision even though glass is opaque to an ultrasonic sensor and does not in any way depend on light.
They would have been better off simply saying that altitude sensing can assist in maintaining altitude in certain conditions within a range of altitudes.
When I seek information and find YouTube videos and forum posts that assert that ultrasonic sensors have problems with water, I simply note to myself that ultrasonic sensors are commonly used to detect fluid levels in containers for industrial processes. When I see a claim that barometric altitude measurements are skewed by prop wash/ground effect over water, I note the poster conveniently forgets that it's called 'ground effect' and the pressure effects are identical over ground and the Mavic appears only to be confused by over water. I've even seen a claim that an IMU can detect movement.
While the OPs query was probably adequately answered, there are those who may want to operate in circumstances where a bit more accurate information would truly be helpful.
The information I provided is correct. The DJI reps that post on the "other forum" are not just blowhards who spout inaccuracies.
Are you saying that IMUs, Inertial Measurement Units, which are accelerometers and gyros inside of a drone - are incapable of detecting movement?
That's werid, because when it gets dark, my mavic will tell me that there isn't enough light for the vision position system to work.
The information I provided is correct. The DJI reps that post on the "other forum" are not just blowhards who spout inaccuracies.
Are you saying that IMUs, Inertial Measurement Units, which are accelerometers and gyros inside of a drone - are incapable of detecting movement?
Yes. A stationary IMU relative to an observer will provide the same reading as one travelling at a constant
velocity. Accelerometers detect change in speed or direction, which is change in velocity, aka acceleration. Not movement, or change in position.
I was more interested in the claim of prioritization of ultrasonic sensors below 30 feet when the manual stipulates vertical detection over a range of 1 to 43 feet but if the movement detection claim is from the same source...
Similar threads
DJI Drone Deals
1. Mini 2
2. Mini 3 Pro
3. Mini 4 Pro
4. Air 2s
5. Air 3
6. Avata 2
7. Mavic 3 Pro
8. Mavic 3 Classic
2. Mini 3 Pro
3. Mini 4 Pro
4. Air 2s
5. Air 3
6. Avata 2
7. Mavic 3 Pro
8. Mavic 3 Classic
Forum statistics
