Various RTL improvements (mainly VTOL specific) #16377

sfuhrer · 2020-12-11T15:23:13Z

Replaces #14728.

Describe problem solved by this pull request

currently landing on the landing pattern doesn't seem to work (land pattern = whatever of the mission is after the DO_LAND_START marker)
high transition altitude (see RTL: For VTOL first descend before transition #14728 for context)

Describe your solution
RTL logic with this PR, changes to current logic in bold
(assuming RTL_TYPE is set to 1 (VTOL and FW default, using mission landing if available))

if close to RTL landing position and in hover, use cone RTL logic (if mission landing is planned that one is used)
if VTOL and in hover mode, use normal MC RTL strategy (climb to RTL_RETURN_ALT, come back and land at RTL landing position)
if DO_LAND_START is set (landing using mission landing):
- if in FW mode, go to DO_LAND_START marker in FW mode and from there on follow mission landing
- if in MC mode, go to the Land position and land there (don't follow landing pattern as in FW mode)
if DO_LAND_START is not set
- if VTOL and in FW mode, climb to RTL_RETURN_ALT, fly to home, loiter down until RTL_DESCEND_ALT in FW mode, transition to hover and land

Test data / coverage
SITL tested with Standard VTOL and multicopter.

Additional context
For VTOL, it is recommended to always fly with a mission landing planned, such that this one can be used during the RTL. Landing approaches using mission landing can be optimized for current wind direction (transition into the wind), and result in less hover time (vehicle transitions over hover, instead of while loitering around home)

Antiheavy · 2020-12-11T23:35:52Z

is this:

For VTOL, it is recommended to always fly with a mission landing planned, such that this one can be used during the RTL. Landing approaches using mission landing can be optimized for current wind direction (transition into the wind), and result in less hover time (vehicle transitions over hover, instead of while loitering around home)

consistent with this?:

if close to home position and in hover, use cone RTL logic

if VTOL and in hover mode, use normal MC RTL strategy (climb to RTL_RETURN_ALT, come back and land)

Have you considered if there should be a parameter that allows/avoids RTL at the Home location if in MC mode? I wonder how many VTOL people would always want to use the mission landing if at all possible? The "return to home" fall-back in the RTL_TYPE=1 logic was originally designed for the case when the mission landing accidentally gets deleted or gets overwritten with an infeasible landing pattern (which unfortunately is something that can easily happen).

From the way this PR is currently written you intend the VTOL vehicle will perform different Return mode behaviors depending on the context. I wonder if that is how 100% of VTOL users will want it to work. I'm not a VTOL expert so I'm not advocating one way or another, just curious.

sfuhrer · 2020-12-15T10:34:45Z

Have you considered if there should be a parameter that allows/avoids RTL at the Home location if in MC mode?

Thanks for pointing this out. I actually wrote it down wrongly in the description. It will always use the mission landing for RTL if RTL_TYPE is 1. What changes is that it now won't follow the landing pattern (all that is after the DO_LAND_START) in hover, but directly go to Land and descend there.

dagar · 2020-12-17T00:58:53Z

src/modules/navigator/mission.cpp

@@ -73,6 +76,12 @@ Mission::on_inactive()
 	 * is used for missions such as RTL. */
 	_navigator->set_cruising_speed();

+	// if we were executing an landing but have been inactive for 2 seconds, then make the landing invalid
+	// this prevents RTL to just continue at the current mission index
+	if (_navigator->getMissionLandingInProgress() && (hrt_absolute_time() - _time_mission_deactivated) > 2_s) {


I'm not sure I follow, how would this fail if the mission landing progress was set to false immediately on first inactivation?

@dagar This prevents the situation where e.g. you are on a mission landing and then RTL kicks in. You don't want to start from the beginning but continue the mission. However, if e.g. the user takes over in manual during a mission landing and then switches back, you probably want to start he mission landing from scratch, as you have no clue where exactly you are and if you are still aligned with the path.

I should have been more specific, the part I don't follow is the magic 2 seconds.

@dagar A long study over many years has calculated this number to be the one and only reasonable value. It's almost like pi....

Why not do it immediately on the actual mission deactivation?

Let's track this in a separate issue - we shouldn't be blocking an incremental move towards a much better behavior.

Issue to track that: #16488

@dagar I think I might need to explain this a bit better:
When you are executing a mission landing and RTL kicks in due to low battery, then the mode switches from mission to RTL and back to mission (assuming you are executing a mission landing, which is the preferred choice almost always).
Two the second delay (it actually needs way less) allows the mission to just continue the landing rather than restarting the landing pattern.

@dagar Sorry, I noticed I told you this before. What is the issue here?

@dagar Sorry, I noticed I told you this before. What is the issue here?

Nothing necessarily, I just wanted to make sure there wasn't a potential hole for a hard to reproduce bug down the round. Eg dataman chokes reading the sdcard and there's a one off delay that's much longer than expected.

src/modules/navigator/rtl.cpp

src/modules/navigator/rtl_params.c

moreba1 · 2020-12-29T12:58:36Z

in the master version, VTOL has a dangerous behavior when it reaches VTOL Land.
@sfuhrer
https://logs.px4.io/plot_app?log=9df156fc-233e-4b54-ad23-e82017ff47f4

desirable behavior is that VTOL transition location be near to land location to minimize power consumption, and the drone doesn't have that much tilt angle and oscillation in MC mode.

…CEND_ALT before reaching home Signed-off-by: RomanBapst <[email protected]>

…a transition Signed-off-by: RomanBapst <[email protected]>

NAV_CMD_LOITER_TO_ALT - this allows vtol to track the loiter circle during the transition instead of trying to fly to the landing position Signed-off-by: RomanBapst <[email protected]>

- descend to RTL descend altitude - transition - move to land waypoint - loiter and then land Signed-off-by: RomanBapst <[email protected]>

Signed-off-by: RomanBapst <[email protected]>

…nding - previously the decision of being on a landing pattern was taken by just looking at the current mission index. However, even if the current mission index indicates a landing pattern the vehicle could be at an arbitrary location, far from being established on the pattern. Signed-off-by: RomanBapst <[email protected]>

Signed-off-by: RomanBapst <[email protected]>

…item instead of using the previous waypoint as landing target - the previous waypoint could be miles away Signed-off-by: RomanBapst <[email protected]>

…ilable Signed-off-by: RomanBapst <[email protected]>

Signed-off-by: RomanBapst <[email protected]>

- this fixes a race condition which happens when an RTL is triggered during the final approach of a mission landing. In that case the mission inactive method is never called. Signed-off-by: RomanBapst <[email protected]>

- if vtol and in rotary wing mode then don't execute the mission landing because it's designed to be flow as a fixed wing - if vtol and in rotary wing mode and mission land is available then fly directly to landing point and don't go home! Signed-off-by: RomanBapst <[email protected]>

… landing - especially when there are strong winds it's better to just go straight to the landing point instead of trying to follow the planned path Signed-off-by: RomanBapst <[email protected]>

Signed-off-by: Silvan Fuhrer <[email protected]>

…TO_ALT for FW Do not use LOITER_TO_ALT for rotary wing mode as it would then always climb to at least MIS_LTRMIN_ALT, even if current clib altitude is below (e.g. RTL immediately after take off) Signed-off-by: Silvan Fuhrer <[email protected]>

dagar · 2020-12-29T17:13:23Z

Rebased on master.

ghost · 2020-12-31T03:42:06Z

I also noticed that the standard_vtol back transition is behaving dangerously. it's on master and also in this pr,
I noticed this on my flight on the real hardware as well.
just after the back-transition it tries to do a roll maneuver in most of the times,
But every once in a while, it does a ok back-transition, without trying to roll .

same mission, two instances, same-pr (firmware), back-transition behavior is different.
https://review.px4.io/plot_app?log=1e1bb5d2-5260-40da-bd7e-46fadf55fc6c ---> rolling
https://review.px4.io/plot_app?log=12468162-8610-4965-831c-dd25c8edfeda ---> without roll

dangerous.back-transition.mp4

LorenzMeier · 2020-12-31T10:27:09Z

Thanks for reporting - @RomanBapst @sfuhrer could you please follow up?

LorenzMeier · 2021-01-04T08:48:34Z

@yuthikasagarage Have you reviewed your flights? It looks like the vehicle is overshooting the transition waypoint (so your plan is "not doable") and hence correcting. There is nothing wrong about the control performance, this seems to be more a planning UX / tuning issue than anything else. That still needs to be solved, but not necessarily in the VTOL controller.

ghost · 2021-01-04T09:29:15Z

i will look into the flights again, and will try to tune the back transition and check whether the issue persists.
however, with a bad planning let's say it does a back-transition in some case, i think all the back-transitions need to be on level flight even if it overshoots, just my two cents.

LorenzMeier

Reviewed, approving after intense flight testing.

sfuhrer · 2021-01-04T12:40:56Z

@moreba1

in the master version, VTOL has a dangerous behavior when it reaches VTOL Land.

This should be fixed with this PR, please re-test.

desirable behavior is that VTOL transition location be near to land location to minimize power consumption, and the drone doesn't have that much tilt angle and oscillation in MC mode.

With the code from this PR in, it will transition at a distance of the loiter radius from home if no mission landing is planned and the waypoint before the landing waypoint isn't at the transition altitude (RTL_DESCEND_ALT). I'm aware that this isn't optimal for minimized hover time, but at least it should be an improvement to before where it could transition 100m above home and then descend in hover.

moreba1 · 2021-01-28T16:12:01Z

* if in FW mode, go to DO_LAND_START marker in FW mode and from there on follow mission landing

Is it possible that VTOL first climbs to RTL_RETURN_ALT and keep this attitude until it reaches to landing pattern?

sfuhrer · 2021-01-28T16:52:58Z

Is it possible that VTOL first climbs to RTL_RETURN_ALT and keep this attitude until it reaches to landing pattern?

That should be the desired behavior. If you noticed something else then let me know, that would indicate a bug.

VTOLDavid · 2021-02-10T18:34:07Z

i will look into the flights again, and will try to tune the back transition and check whether the issue persists.
however, with a bad planning let's say it does a back-transition in some case, i think all the back-transitions need to be on level flight even if it overshoots, just my two cents.

I having this problem with a tailsitter with very dangerous turns during transition. Is it possible to disable the navigation during transition? The sugested solution is to give more time to the backtransition?

sfuhrer · 2021-02-11T14:48:08Z

I having this problem with a tailsitter with very dangerous turns during transition. Is it possible to disable the navigation during transition? The sugested solution is to give more time to the backtransition?

I don't think this is directly related to this specific PR - would you mind creating a new issue, and include some flight logs in the description? Thanks!

hamishwillee · 2021-03-24T06:26:02Z

@sfuhrer @Antiheavy Are there any docs changes related to this for PX4v1.12? My guess is yes, though it is unclear with the long thread, what delivered/what the landing behaviour is (?).

Looking at the initial post, it is mostly about what should happen if RTL_TYPE is set to use a mission landing (RTL_TYPE=1) and there was no mission landing pattern defined - in which case it should fly down in FW mode to the descent altitude and then switch to MC and descend "as per usual".

The description is a little flawed because it does not cater for the behaviour with rally points.

Docs for vtol are here:
https://docs.px4.io/master/en/flight_modes/return.html#vtol
https://docs.px4.io/master/en/flight_modes/return.html#mission-landing-rally-point-return-type-rtl-type-1

sfuhrer requested review from dagar and RomanBapst December 11, 2020 15:23

sfuhrer added the Hybrid VTOL 🛩️🚁 Multirotor + Fixedwing! label Dec 11, 2020

sfuhrer force-pushed the pr-vtol-rtl-fixes branch 4 times, most recently from 7061aff to 0ef2f6e Compare December 11, 2020 20:52

mrpollo added the [ARCHIVED] v1.12-Release-Notes label Dec 16, 2020

dagar reviewed Dec 17, 2020

View reviewed changes

src/modules/navigator/rtl.cpp Outdated Show resolved Hide resolved

dagar reviewed Dec 17, 2020

View reviewed changes

src/modules/navigator/rtl_params.c Show resolved Hide resolved

RomanBapst previously approved these changes Dec 18, 2020

View reviewed changes

sfuhrer dismissed RomanBapst’s stale review via 14735ee December 21, 2020 11:36

sfuhrer force-pushed the pr-vtol-rtl-fixes branch from 214f214 to 14735ee Compare December 21, 2020 11:36

RomanBapst added 13 commits December 29, 2020 12:13

rtl: if very close to home then do not descend if higher than RTL_DES…

6cb4956

…CEND_ALT before reaching home Signed-off-by: RomanBapst <[email protected]>

rtl: for vtol first desend down to RTL descend altitude before doing …

e1dbd66

…a transition Signed-off-by: RomanBapst <[email protected]>

rtl: for fixed wing (also vtol) make the descend state of RTL use type

087e211

NAV_CMD_LOITER_TO_ALT - this allows vtol to track the loiter circle during the transition instead of trying to fly to the landing position Signed-off-by: RomanBapst <[email protected]>

RTL: implement proper RTL sequence for VTOL

30f7bfc

- descend to RTL descend altitude - transition - move to land waypoint - loiter and then land Signed-off-by: RomanBapst <[email protected]>

rtl: set proper acceptance radius for move_to_land item

680fe11

Signed-off-by: RomanBapst <[email protected]>

reworked RTL cone implementation

dbadf2d

Signed-off-by: RomanBapst <[email protected]>

mission: fixed bug in computation of landing altitude

d8b3e41

Signed-off-by: RomanBapst <[email protected]>

mission: when land start marker present, look for loiter to altitude …

c92247b

…item instead of using the previous waypoint as landing target - the previous waypoint could be miles away Signed-off-by: RomanBapst <[email protected]>

rtl: fixed computation of return altitude when mission landing is ava…

e3bc901

…ilable Signed-off-by: RomanBapst <[email protected]>

rtl: use math namespace

e0bf0b9

Signed-off-by: RomanBapst <[email protected]>

various fixed for bugs encountered during testing

d8e66c7

Signed-off-by: RomanBapst <[email protected]>

mission: reset work_item_type during mission inactivation

59e2224

- this fixes a race condition which happens when an RTL is triggered during the final approach of a mission landing. In that case the mission inactive method is never called. Signed-off-by: RomanBapst <[email protected]>

RomanBapst and others added 5 commits December 29, 2020 12:13

navigator: deactivate line following after backt-ransition and before…

899078e

… landing - especially when there are strong winds it's better to just go straight to the landing point instead of trying to follow the planned path Signed-off-by: RomanBapst <[email protected]>

RTL: use NAV_CMD_LOITER_TO_ALT for Climb state

772ff51

Signed-off-by: Silvan Fuhrer <[email protected]>

FW POS: remove setting of curr_sp to LOITER if next wp invalid

9eaef07

Signed-off-by: Silvan Fuhrer <[email protected]>

dagar force-pushed the pr-vtol-rtl-fixes branch from 14735ee to 0bc54f2 Compare December 29, 2020 17:13

sfuhrer mentioned this pull request Jan 4, 2021

RTL: For VTOL first descend before transition #14728

Closed

LorenzMeier approved these changes Jan 4, 2021

View reviewed changes

LorenzMeier merged commit 92634e7 into master Jan 4, 2021

LorenzMeier deleted the pr-vtol-rtl-fixes branch January 4, 2021 10:58

This was referenced Jan 4, 2021

RTL: Perform transition towards home location #14780

Merged

RTL follow up #16488

Closed

sfuhrer mentioned this pull request Jan 18, 2021

RTL: descending before having reached mission land point #15410

Closed

mrpollo mentioned this pull request Mar 22, 2021

v1.12 Updates: DNF updated with Gyro FFT PX4/PX4-user_guide#1136

Closed

hamishwillee mentioned this pull request Mar 24, 2021

Add more release notes from past PRs PX4/PX4-user_guide#1122

Merged

ThomasRigi mentioned this pull request Sep 21, 2021

RTL going to wrong Landing location #18271

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Various RTL improvements (mainly VTOL specific) #16377

Various RTL improvements (mainly VTOL specific) #16377

sfuhrer commented Dec 11, 2020 •

edited

Loading

Antiheavy commented Dec 11, 2020 •

edited

Loading

sfuhrer commented Dec 15, 2020

dagar Dec 17, 2020

RomanBapst Dec 18, 2020

dagar Dec 18, 2020

RomanBapst Dec 21, 2020

dagar Dec 21, 2020

LorenzMeier Jan 4, 2021

sfuhrer Jan 5, 2021

RomanBapst Jan 29, 2021

RomanBapst Jan 29, 2021

dagar Jan 29, 2021

moreba1 commented Dec 29, 2020 •

edited

Loading

dagar commented Dec 29, 2020

ghost commented Dec 31, 2020 •

edited by ghost

Loading

LorenzMeier commented Dec 31, 2020

LorenzMeier commented Jan 4, 2021

ghost commented Jan 4, 2021

LorenzMeier left a comment

sfuhrer commented Jan 4, 2021 •

edited

Loading

moreba1 commented Jan 28, 2021

sfuhrer commented Jan 28, 2021

VTOLDavid commented Feb 10, 2021

sfuhrer commented Feb 11, 2021

hamishwillee commented Mar 24, 2021

Various RTL improvements (mainly VTOL specific) #16377

Various RTL improvements (mainly VTOL specific) #16377

Conversation

sfuhrer commented Dec 11, 2020 • edited Loading

Antiheavy commented Dec 11, 2020 • edited Loading

sfuhrer commented Dec 15, 2020

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

moreba1 commented Dec 29, 2020 • edited Loading

dagar commented Dec 29, 2020

ghost commented Dec 31, 2020 • edited by ghost Loading

LorenzMeier commented Dec 31, 2020

LorenzMeier commented Jan 4, 2021

ghost commented Jan 4, 2021

LorenzMeier left a comment

Choose a reason for hiding this comment

sfuhrer commented Jan 4, 2021 • edited Loading

moreba1 commented Jan 28, 2021

sfuhrer commented Jan 28, 2021

VTOLDavid commented Feb 10, 2021

sfuhrer commented Feb 11, 2021

hamishwillee commented Mar 24, 2021

sfuhrer commented Dec 11, 2020 •

edited

Loading

Antiheavy commented Dec 11, 2020 •

edited

Loading

moreba1 commented Dec 29, 2020 •

edited

Loading

ghost commented Dec 31, 2020 •

edited by ghost

Loading

sfuhrer commented Jan 4, 2021 •

edited

Loading