RealSense Obstacle Avoidance and Sun Glare

[Barton]

I realize that a significant portion of the 'regular crew' on the forum have little use for obstacle avoidance so this post is targeted for those that will be utilizing the RealSense obstacle avoidance feature of the H Plus.

This morning was a beautiful day, sunny with no clouds. As I was practicing being a pilot, I was running through some of the features available to me. The H+ was at a reasonable height over an open field (30 - 40 feet), no trees within about a 100 feet of flight direction. I turned on OBS AVOID (RealSense) and continued to putter around. At one point I ended up flying directly towards the morning sun (about 8:40 AM, Oregon). To my surprise, the H+ did its Obstacle Detected Dance, halted forward movement and flashed the 'radar' arc on the ST. The H+ would not go directly forward. A few taps of the rudder to rotate the H+ and off it went again. A few more tests runs seemed to confirm to me that the RS system will be tricked or blinded by reasonably early (and probably evening) sun glare. It appears that the blinding situation exists for a very small flight path heading, maybe 5 degrees or so...

This seems to me to be an extreme corner case for most pilots but thought I would mention it before my aging brain cells eliminate this tidbit of information.

 

[Eagle's Eye Video]

It had clearly been shown in the H version. that either large reflective surfaces or facing directly into the sun will cause erroneous or reduced effectiveness of RS... The H+ is supposed to have a newer enhanced version, but it will be months before enough lab rats (err, us) are able to put up side by side comparisons of both units operating, to tell what those improvements actually do for us.

Purchased the H Pro in Oct. 2016... never used the RS for really anything... this time I'll wait for the non RS version, TYVM.

 

[Barton]

It had clearly been shown in the H version. that either large reflective surfaces or facing directly into the sun will cause erroneous or reduced effectiveness of RS...

I figured that was probably the case. My reason for posting in the H+ forum was that some, like me, will be newbies not coming from an H and may not be reading or searching old posts on the H. My purpose was mostly 'heads up' alert and maybe to start a discussion on the H+ forum.

 

[rdonson]

I appreciate you’re sharing your experience with the newest version of RealSense on the H Plus. Yuneec and Intel have been mum about its capabilities.

If you’re going to continue to test RS keep in mind there are NO sensors on the top, sides or back of the H Plus. U

One other thing learned with the H is that if OBS is active with RS the H won’t land with the landing gear up. That should be a low risk thing to test with the H Plus.

Correction: on the H if RS is activated in the ST16 the OBS does NOT need to be activated for it to refuse to land with the gear up. chapeau to @wings for catching this

 

[wings]

I appreciate you’re sharing your experience with the newest version of RealSense on the H Plus. Yuneec and Intel have been mum about its capabilities.

If you’re going to continue to test RS keep in mind there are NO sensors on the top, sides or back of the H Plus. U

One other thing learned with the H Pro is that if OBS is active with RS the H won’t land with the landing gear up. That should be a low risk thing to test with the H Plus.

With the H Pro the RS has to be set to active but the OBS switch on the ST16 does not have to be on. I have tested this many times and the H Pro will not land with the gear up and the ST16 Obs/Avoid switch off. Do not know if t he H Plus is different. Is it?

 

[Barton]

I appreciate you’re sharing your experience with the newest version of RealSense on the H Plus. Yuneec and Intel have been mum about its capabilities.

If you’re going to continue to test RS keep in mind there are NO sensors on the top, sides or back of the H Plus. U

One other thing learned with the H is that if OBS is active with RS the H won’t land with the landing gear up. That should be a low risk thing to test with the H Plus.

Ron, yes, I intend to keep testing RS and am aware that I only have OBS looking forward. One of my curiosities about RS is how big a 'map' of 3D space it creates and how long it retains this map. One test I will eventually make is to fly under the branch canopy of a tree and then very gingerly fly vertically (with RS ON) to see if it recognized and remembered the tree canopy. Given that the rate slider combined with the joy stick seems to provide very precise (and very slow, if desired) speeds, I don't think this test will be too risky.

One of the reasons I chose the H+ over other UAVs was RS and the 360 degree gimbal. My thought was that even though the H+ only had forward looking OBS, this might be sufficient because the H+ could always be flying forward and the camera could be pointing in the desired direction because of the gimbal (sideways, backwards, etc). Other UAVs require the additional sensors if OBS is desired because the UAV HAS to fly sideways for panning shots, etc. Time will tell if I was delusional about this.

While I have not specifically tried the 'landing gear up, OBS ON' test, today I did try RTL with landing gear up and confirmed that the H+ automatically lowers the landing gear before landing. My recollection is that the landing gear was lowered when the H+ was maybe 15-20 feet off of the ground. I also tested the concept of using RTL to bring the H+ back close to home and then aborting the landing by switching RTL OFF - worked like a charm. Let the H+ 'land' to about 6' or so, switched off RTL and the H+ immediately went into its stable 'hover' mode and I had full control from that point on.

Next time I fly (hopefully tomorrow morning), I will do the landing test with the OBS ON and landing gear UP.

 

[Barton]

With the H Pro the RS has to be set to active but the OBS switch on the ST16 does not have to be on. I have tested this many times and the H Pro will not land with the gear up and the ST16 Obs/Avoid switch off. Do not know if t he H Plus is different. Is it?

If I understand the H+ at this point, the difference between your H PRO and the PLUS is that there is no way to activate RS without also turning on OBS. The only way, I think, to turn on RS on the H+ is by using the OBS switch. I have not seen any 'buttons' on the control screen that activates RS. The test mentioned above should help resolve the question.

 

[rdonson]

With the H Pro the RS has to be set to active but the OBS switch on the ST16 does not have to be on. I have tested this many times and the H Pro will not land with the gear up and the ST16 Obs/Avoid switch off. Do not know if t he H Plus is different. Is it?

You are correct about the H with RS. I will amend my post with the correction. Thanks for catching this.

I have no first hand knowledge of the H Plus with RS.

 

[rdonson]

@Barton unless they’ve changed RS radically between V1 which is on the H and V2 which is on the H Plus there should be two downward facing sonar cameras at the rear of the RS module. This is how on the H with RS it detects the ground for IPS and knows it shouldn’t land with gear up.

On the H, RS was most valuable in “Watch Me” and “Follow Me” modes.

 

[Barton]

OK, results of the H Plus landing test with the OBS ON and landing gear UP : The results surprised me. After takeoff, I flew to about 15 feet elevation, turned on OBS, retracted landing gear, made sure the rate slider was in the slowest position and slowly dropped elevation. At about 5 feet got a warning that OBS was disabled below 5 feet. Continued to descend. At less than 1 foot I stopped descent. Landing gear still UP. No apparent attempt was made (warnings on display, etc) to keep me from tanking the camera by going lower - bummer. I reviewed the video from the HD recorder just to make sure I didn't miss anything during the test.

On the positive side, at the end of the flight, I decided to continue to fly (close to the launch area) until low battery triggered the landing. At 21% battery, RTL triggered and landing was without any drama...

 

[wings]

OK, results of landing test with the OBS ON and landing gear UP . The results surprised me. After takeoff, I flew to about 15 feet elevation, turned on OBS, retracted landing gear, made sure the rate slider was in the slowest position and slowly dropped elevation. At about 5 feet got a warning that OBS was disabled below 5 feet. Continued to descend. At less than 1 foot I stopped descent. Landing gear still UP. No apparent attempt was made (warnings on display, etc) to keep me from tanking the camera by going lower - bummer. I reviewed the video from the HD recorder just to make sure I didn't miss anything during the test.

On the positive side, at the end of the flight, I decided to continue to fly (close to the launch area) until low battery triggered the landing. At 21% battery, RTL triggered and landing was without any drama...

Are you talking about the H Plus?

 

[Barton]

Are you talking about the H Plus?

Yes, will update my post to make this clear. The only UAV that I own is the PLUS.

 

[robport]

From my understanding of a long discussion with Yuneec CS, Realsense does absolutely nothing unless the craft is running in one of the self-control modes. So unless it is in follow, watchme, cablecam, orbit, or return home, it does nothing. If you are flying it yourself, the only thing that is working is the sonar.

 

[Barton]

From my understanding of a long discussion with Yuneec CS, Realsense does absolutely nothing unless the craft is running in one of the self-control modes. So unless it is in follow, watchme, cablecam, orbit, or return home, it does nothing. If you are flying it yourself, the only thing that is working is the sonar.

Interesting... This would not explain why it appeared that RS was stopping forward motion due to sun glare - something stopped the motion and I doubt that sonar would do that.

Not sure what test one could perform to confirm/deny the CS proposition about RS as in most flight situations when NOT being in self-control modes, the sonar and RS responses to an obstacle would probably appear to be the same (stopping forward motion). What is lacking with sonar is the ability to 'go around' an obstacle. The 'go around' only make sense when one is flying with one of the self-control modes. It might be a rhetorical question anyway as I don't really care what stops forward motion if an obstacle is detected as long as it does stop forward motion (when I am flying with the expectation of OBS assistance).

 

[rdonson]

It would appear there are few correlations between RS on the H and RS on the H Plus. Then again I haven’t seen any Yuneec documentation for RS on the H Plus. It’s largely a mystery for me anyway.

 

[robport]

Interesting... This would not explain why it appeared that RS was stopping forward motion due to sun glare - something stopped the motion and I doubt that sonar would do that.

Not sure what test one could perform to confirm/deny the CS proposition about RS as in most flight situations when NOT being in self-control modes, the sonar and RS responses to an obstacle would probably appear to be the same (stopping forward motion). What is lacking with sonar is the ability to 'go around' an obstacle. The 'go around' only make sense when one is flying with one of the self-control modes. It might be a rhetorical question anyway as I don't really care what stops forward motion if an obstacle is detected as long as it does stop forward motion (when I am flying with the expectation of OBS assistance).

I'm not sure if he was just talking about having the system react or simply sensing. Like you, I thought that realsense runs to detect and puts up another bar on the display when not self flying (one for sonar and one for realsense, cause I've seen two in obstacle avoidance mode), but only drives it away from obstacles when the craft is in an auto mode. It would be nice to find out if both will stop if from going forward or if just the sonar does that.

 

[egodyla]

Hello, I am new to the typhoon, and have a question about Obstacle Avoidance (OA) with RS: If there are only front facing sensors for OA, then isn't it possible that the H crashes into something when maneuvering to avoid an obstacle in front of it?