Pitch roll and barrel roll?

Blue Robotics Software ArduSub
1

I see 2 months ago an official video was posted showing the BR2 heavy doing barrel rolls and holding downwards pitch at nearly 90’ down.

I can not get my unit to go past about 60’ any more even at 100% thruster power, Acro mode doesnt hold.

on old firmware versions when the BR2 heavy was just launched ( 3.5 I think) the control was frame referenced to the vehicle (FRD) rather than world ( NED)- so doing sustained liek shpwn in the video was manageable, then firmware 3.5.4 you could do a full barrel roll NED but could not hold sustain pitch/roll, then after that in firmware the pitch/roll angle limits were set via firmware.

I am using a DVL so using 4.5.0 and 4.5.7 firmware and blueOS, but with newest firmware I can not get the vehicle to go further than 45-60’ pitch/roll at 100% gain.

Considering this is a fairly recent video- does anyone know how to get the ROV to do the full 6DOF control as shown in the video?, or operate via BlueOS as ROV body reference rather than NED/world reference.

Looking to do this via cockpit/BlueOs/Qgroundcontrol.

If i go back to old old firmware and Qgroundcontrol I loose the DVL position hold features.

anyone?

2

Hi @WileyP,

That video is seemingly the same ROV movement shots from the original BlueROV2 Heavy Kit launch video (i.e. recently posted video, old content):

I have limited experience with Acro mode, but if my understanding is correct it should hold attitude components while commanded with a zero change rate for them.

I’ll try to follow that up internally.

What is your ATC_ANGLE_MAX parameter set to? The current firmware limit is set to 80°.

There have been multiple discussions over the years about how best to manage the various frames of reference people may want to be able to control the vehicle in, and there is some ongoing work which I believe moves us towards being able to properly support that - currently focused on validation tests (like this one) so we have a solid foundation of current/expected behaviour established that implementations can be developed over.

3

Ok, so to clarify, Acro’s attitude control is only rate-based, so commanding a zero change rate for an attitude axis (assuming ACRO_TRAINER is disabled) results in the autopilot resisting rotations but not strictly maintaining a given lean angle. In practice that will generally maintain attitude angles, but if there is a stronger disturbance than the thrusters can compensate for (or a slower drift than the sensors can detect) then the commanded zero change rate will resettle at whatever angle the vehicle has been moved to, rather than returning to the initial angle before the rotation rate commands changed to zero.

Things that make that easier for the controller include tuning the vehicle’s buoyancy to be neutral in the water body being operated in, aligning the vehicle’s centre of mass and centre of volume (so it is not self-righting), and making sure the PID values are well-tuned for the vehicle dynamics.

As I understand it the behaviours shown in the video are all currently possible, but some of them are difficult to achieve, and could be made more practical with better application of the autopilot’s control systems. To that effect I’ve just raised this issue to describe our thoughts so far around the desirable feature set. Feel free to comment there (or here) if there’s anything we’ve missed that you think would be valuable to include :slight_smile:

4

ATC_ANGLE_MAX is at 8000cdm, have tried ANGLE_BOOST- disabled/enabled

I will have a check to see if ACRO_TRAINER is enabled or not.

In current setup buoyancy is fairly balance but still ‘top heavy’ so self rights.
in Acro mode: Holding pitch/Roll gets to about 60’ then starts spinning about randomly and rotating but not doing a barrell roll or maintaining pitch/roll. then when let go , reverts back to level so Rate goes back to zero.

Application aiming for is sustained and maintained controlled pitch and roll control. we have to be able to look close to 90’ up or down. which is why the early firmware on my first BR2 heavy was great as 3.5 could hold unrestricted angle and control was FRD/body orientated rather than NED control. could hold pitch and roll, and also do clean smooth barrel rolls.

With current firmware- can not even do a barrel roll to show off in the water.

I think if we had the option of choosing between body frame -FRD, or world frame-NED , it would make all the difference in being able to orientate the ROV as we want.

FRD would also be easier when doing pitch/roll combinations as would be operating like an FPV drone. In current NED body frame reference, if for example the ROV is pointed 35’ down and roll 20’ , and you are relying on camera only it is not intuitive as it is not behaving in relation to the camera view and you have to try to work out what combination of lateral, forward, vertical thrusters you have to use to get into position. even doing this in a test pool or close enough to surface for you to be watching the ROV it is not easy to operate at any real pitch/roll angle.

NED frame reference only really works if the ROV is close to level.
FRD fame reference works well when in any pitch/roll combination .

when in Body Frame-FRD mode, all movements are in relation to the nose of the ROV, and with the Navigator board, looking at Mavlink inspector, it does not seem to suffer any gimbal lock issues. so if ROV was pointing 45’, driving forward would swim it down at the 45’ , vertical thrust would lift it at angle relative to the 45’ etc

This could also make diving to deeper depths quicker as you could point the ROV nose down and just power dive down.

We used to be able to do this on some of the original firmware when the BR2 Heavy was just launched way back in 2018

.