actuator: add support for MAV_CMD_DO_SET_ACTUATOR #16758

coderkalyan · 2021-02-04T21:51:58Z

Adds support for using the MAVLink command MAV_CMD_DO_SET_ACTUATOR to
update the actuator values on control group 3 aux{1, 2, 3}. A simple
deconfliction with rc_update is implemented: when a MAVLink command is
sent, RC is disabled for that channel until a major stick movement is
detected.

Closes what was attempted in #10320 and #15281.

The following MAVSDK script can be used to test (minor modification might be necessary based on your setup):

#include <mavsdk/mavsdk.h>
#include <mavsdk/plugins/mavlink_passthrough/mavlink_passthrough.h>
#include <mavsdk/plugins/info/info.h>
#include <chrono>
#include <cstdint>
#include <iostream>
#include <future>
#include <memory>

using namespace mavsdk;

void send_actuator(MavlinkPassthrough& mavlink_passthrough,
        float value1, float value2, float value3);

int main(int argc, char **argv)
{
    Mavsdk mavsdk;
    std::string connection_url;
    ConnectionResult connection_result;
    float value1, value2, value3;

    if (argc == 5) {
        connection_url = argv[1];
        connection_result = mavsdk.add_any_connection(connection_url);
        value1 = std::stof(argv[2]);
        value2 = std::stof(argv[3]);
        value3 = std::stof(argv[4]);
    } 

    if (connection_result != ConnectionResult::Success) {
        std::cout << "Connection failed: " << connection_result << std::endl;
        return 1;
    }

	bool discovered_system = false;
    mavsdk.subscribe_on_new_system([&mavsdk, &discovered_system]() {
		const auto system = mavsdk.systems().at(0);

		if (system->is_connected()) {
			std::cout << "Discovered system" << std::endl;
			discovered_system = true;
		}
	});

	std::this_thread::sleep_for(std::chrono::seconds(2));

	if (!discovered_system) {
        std::cout << "No device found, exiting." << std::endl;
        return 1;
    }

	std::shared_ptr<System> system = mavsdk.systems().at(0);
	for (auto& tsystem : mavsdk.systems()) {
		auto info = Info{tsystem};
		std::cout << info.get_identification().second.hardware_uid << std::endl;
		if (info.get_identification().second.hardware_uid == "3762846593019032885") {
			system = tsystem;
		}
	}

	auto mavlink_passthrough = MavlinkPassthrough{system};

	send_actuator(mavlink_passthrough, value1, value2, value3);

    return 0;
}

void send_actuator(MavlinkPassthrough& mavlink_passthrough,
		float value1, float value2, float value3)
{
	std::cout << "Sending message" << std::endl;
	mavlink_message_t message;
	mavlink_msg_command_long_pack(
			mavlink_passthrough.get_our_sysid(),
			mavlink_passthrough.get_our_compid(),
			&message,
			1, 1,
			MAV_CMD_DO_SET_ACTUATOR,
			0,
			value1, value2, value3,
			NAN, NAN, NAN, 0);
	mavlink_passthrough.send_message(message);
	std::cout << "Sent message" << std::endl;
}

coderkalyan · 2021-02-04T21:52:57Z

FYI @jinger26 @bkueng @hamishwillee

dagar · 2021-02-04T22:00:32Z

Hold tight, we're very close to finally having a good solution to this problem.

FYI @JacobCrabill

dagar · 2021-02-04T22:02:04Z

src/modules/rc_update/rc_update.cpp

 	actuator_controls_s actuator_group_3{};
+	// copy in previous actuator control setpoint in case aux{1, 2, 3} isn't changed
+	_actuator_controls_3_sub.update(&actuator_group_3);


It's almost always a bad idea to publish the same topic from multiple sources simultaneously.

Hmm ok. I implemented the deconfliction logic that Beat recommended but I see how it could cause problems.

And then you copy control group 6 here if published and changing and overwrite group 3? That would allow you to default the output of group 3 to the last active from either manual or external control. And someone just interested in external can map the mixer straight to group 6 and never have manual intervention.

coderkalyan · 2021-02-04T22:07:58Z

Hold tight, we're very close to finally having a good solution to this problem.

@dagar Ok 😄 Do you mean a good solution that will solve the deconfliction or something that circumvents my PR completely?

coderkalyan · 2021-02-04T22:09:09Z

I fixed the format issues but CI seems to be overflowing flash space on fmuv2. Is this something I need to account for, and how?

coderkalyan · 2021-02-04T22:12:00Z

Note before any merging can take place: I haven't checked how mission mode handles this command yet, if its different.

Adds support for using the MAVLink command MAV_CMD_DO_SET_ACTUATOR to update the actuator values on control group 3 aux{1, 2, 3}. A simple deconfliction with rc_update is implemented: when a MAVLink command is sent, RC is disabled for that channel until a major stick movement is detected.

LorenzMeier · 2021-02-04T22:36:53Z

src/modules/rc_update/rc_update.cpp

+	// if there is a major stick movement to re-activate RC mode
+	bool major_movement[3] = {false, false, false};
+
+	// detect a big stick movement


This will not detect a big, but a fast stick movement.

LorenzMeier · 2021-02-04T22:38:15Z

src/modules/mavlink/mavlink_receiver.cpp

+			}
+
+			if (updated) {
+				_actuator_controls_pubs[3].publish(actuator_controls);


How about you publish control group 6 (might need to be defined) here, so it's conflict-free? https://docs.px4.io/master/en/concept/mixing.html#control-group-6-first-payload

@LorenzMeier I think the rationale @bkueng had with using control group 3 was that it would work OOTB (mixers would not have to be modified by the user) and RC override is also possible. If group 6 is still preferable I can update

@coderkalyan I was suggesting to handle the vehicle_command MAV_CMD_DO_SET_ACTUATOR directly in rc_update.cpp to avoid conflicts and then it would also just work for navigator.

LorenzMeier · 2021-02-04T22:39:40Z

src/modules/rc_update/rc_update.cpp

 	actuator_controls_s actuator_group_3{};
+	// copy in previous actuator control setpoint in case aux{1, 2, 3} isn't changed
+	_actuator_controls_3_sub.update(&actuator_group_3);


And then you copy control group 6 here if published and changing and overwrite group 3? That would allow you to default the output of group 3 to the last active from either manual or external control. And someone just interested in external can map the mixer straight to group 6 and never have manual intervention.

LorenzMeier · 2021-02-04T22:40:26Z

src/modules/mavlink/mavlink_receiver.cpp

@@ -489,8 +489,39 @@ void MavlinkReceiver::handle_message_command_both(mavlink_message_t *msg, const
 			send_ack = true;
 		}

-	} else {
+	} else if (cmd_mavlink.command == MAV_CMD_DO_SET_ACTUATOR) {
+		// since we're only paying attention to 3 AUX outputs, the


This also needs to be added to navigator.

Yep, will do.

To avoid code duplication, you could move it to the commander. This handles commands in a mission as well as MAVLink commands. Compare with the code and successful tests in #14809

since we're only paying attention to 3 AUX outputs What is the rational behind this? This is very confusing, and is not obvious unless you dig into the code

JacobCrabill · 2021-02-05T06:24:00Z

I actually am working on a more general solution with @dagar that should solve this problem as a subset of the much broader control allocation / output mapping topic... will share more details once it's a little more complete, but as of right now I can arbitrarily map a DO_SET_SERVO command to any output you'd like, simply by setting a parameter on that output. More details to come soon!

coderkalyan · 2021-02-05T07:13:01Z

@JacobCrabill sounds cool! Keep me updated as soon as you have details/a PR 😄

JacobCrabill · 2021-02-10T02:20:41Z

@JacobCrabill sounds cool! Keep me updated as soon as you have details/a PR

It's still very much a WIP, but see here for how to control up to 8 arbitrary actuators via DO_SET_SERVO: #16808

LorenzMeier · 2021-02-14T13:33:22Z

@coderkalyan I expect the more general approach to take more time, so I would like to get this here in for 1.12. Please disregard my earlier comments regarding the control groups, I would keep this as-is at it seems like a good solution for 1.12. But could you please take note of the comment from @dayjaby regarding the command handling? His PR is indeed pretty good and should be included:
#14809

mrpollo · 2021-02-24T17:05:18Z

Closing in favor of #16808

LorenzMeier · 2021-03-02T10:56:09Z

@mrpollo This should not have been closed without the alternative being actually merged. I'm merging this now. @dagar @JacobCrabill we need to make sure to not impair the usage of PX4 by trying to do the right thing. This has been going on as a pattern, effectively pushing away users because it leads to six-month long delays where people can't use the project.

mrpollo · 2021-03-02T22:20:44Z

@LorenzMeier thanks for merge, over the last developer call we decided to close it in favor of #16808, at the time it seemed ready to be merged.

DronecodeBot · 2023-08-15T16:03:18Z

This pull request has been mentioned on Discussion Forum for PX4, Pixhawk, QGroundControl, MAVSDK, MAVLink. There might be relevant details there:

https://discuss.px4.io/t/using-do-set-actuator-to-releasing-the-payload/28686/5

coderkalyan mentioned this pull request Feb 4, 2021

Add support for actuator control via DO_SET_ACTUATOR #15281

Closed

dagar reviewed Feb 4, 2021

View reviewed changes

coderkalyan mentioned this pull request Feb 4, 2021

Make DO_SET_SERVO work #10320

Closed

coderkalyan force-pushed the pr-actuator-aux branch from 591553a to 45209f5 Compare February 4, 2021 22:06

coderkalyan force-pushed the pr-actuator-aux branch from 45209f5 to 2e5ad58 Compare February 4, 2021 22:14

LorenzMeier requested changes Feb 4, 2021

View reviewed changes

coderkalyan mentioned this pull request Feb 17, 2021

Configurable Actuator Output Control #16808

Closed

mrpollo closed this Feb 24, 2021

LorenzMeier reopened this Mar 2, 2021

LorenzMeier approved these changes Mar 2, 2021

View reviewed changes

LorenzMeier merged commit b257f9d into PX4:master Mar 2, 2021

Jaeyoung-Lim mentioned this pull request Mar 6, 2021

Unable to set Set servo mission item for index bigger than 5 mavlink/qgroundcontrol#9517

Closed

This was referenced Mar 7, 2021

Add package dropping for simulating drone delivery PX4/PX4-SITL_gazebo-classic#724

Merged

Add a payload drop model for drone delivery scenarios in SITL Gazebo #17047

Merged

Fix actuator controls timestamps for MAV_CMD_DO_SET_ACTUATOR #17048

Merged

mrpollo mentioned this pull request May 31, 2021

Add set_actuator mavlink/MAVSDK#1165

Merged

junwoo091400 added the Admin: Enhancement (improvement) 💡 label Aug 15, 2023

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

actuator: add support for MAV_CMD_DO_SET_ACTUATOR #16758

actuator: add support for MAV_CMD_DO_SET_ACTUATOR #16758

coderkalyan commented Feb 4, 2021

coderkalyan commented Feb 4, 2021

dagar commented Feb 4, 2021

dagar Feb 4, 2021

coderkalyan Feb 4, 2021

LorenzMeier Feb 4, 2021

coderkalyan commented Feb 4, 2021 •

edited

Loading

coderkalyan commented Feb 4, 2021

coderkalyan commented Feb 4, 2021

LorenzMeier Feb 4, 2021

LorenzMeier Feb 4, 2021

coderkalyan Feb 4, 2021

bkueng Feb 15, 2021

LorenzMeier Feb 4, 2021

LorenzMeier Feb 4, 2021

coderkalyan Feb 4, 2021

dayjaby Feb 5, 2021

Jaeyoung-Lim Mar 7, 2021

JacobCrabill commented Feb 5, 2021

coderkalyan commented Feb 5, 2021

JacobCrabill commented Feb 10, 2021

LorenzMeier commented Feb 14, 2021

mrpollo commented Feb 24, 2021

LorenzMeier commented Mar 2, 2021

mrpollo commented Mar 2, 2021

DronecodeBot commented Aug 15, 2023

actuator: add support for MAV_CMD_DO_SET_ACTUATOR #16758

actuator: add support for MAV_CMD_DO_SET_ACTUATOR #16758

Conversation

coderkalyan commented Feb 4, 2021

coderkalyan commented Feb 4, 2021

dagar commented Feb 4, 2021

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

coderkalyan commented Feb 4, 2021 • edited Loading

coderkalyan commented Feb 4, 2021

coderkalyan commented Feb 4, 2021

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

JacobCrabill commented Feb 5, 2021

coderkalyan commented Feb 5, 2021

JacobCrabill commented Feb 10, 2021

LorenzMeier commented Feb 14, 2021

mrpollo commented Feb 24, 2021

LorenzMeier commented Mar 2, 2021

mrpollo commented Mar 2, 2021

DronecodeBot commented Aug 15, 2023

coderkalyan commented Feb 4, 2021 •

edited

Loading