ArduPilot doesn't respect Mavlink version of port when sending MAV_CMD_DO_DIGICAM_CONTROL.

[pklapperich]

Bug report

Issue details

When running ArduCopter in simulator with a real serial port attached to the vehicle so I can test a camera on serial4/uartE I noticed that that MAV_CMD_DO_DIGICAM_CONTROL when sent from either clicking "Trigger Camera" within the GCS or if the autopilot generates this messages due to flying a mission containing MAV_CMD_DO_SET_CAM_TRIGG_DIST the messages are sent via Mavlink2 even though the port is set to Mavlink1 and the camera is sending heartbeats as a Gimbal via Mavlink1, the COMMAND_LONG.

For testing I'm using a Sentera Double 4k.

Steps to reproduce

1. Start the simulator via:

~/ardupilot/ArduCopter $ ../Tools/autotest/sim_vehicle.py -v ArduCopter -f quad -L Pyramid -j $(nproc) -A "--uartE=uart:/dev/ttyUSB1"

2. Connect a ground station and configure the port parameters:

SERIAL2_BAUD 115200
SERIAL2_OPTIONS 0
SERIAL2_PROTOCOL MAVLink1
SR2_EXTRA1 10hz
SR2_EXTRA3 2hz
SR2_POSITION 4hz

3. Restart the simulator using the same command as step 1
4. Using the rx line from a second serial port, watch the traffic coming out of the camera. Notice it's all heartbeats with a Mavlink1 delimiter:
   HEARTBEAT {type : 26, autopilot : 0, base_mode : 0, custom_mode : 0, system_status : 0, mavlink_version : 0}; delim: 0xFE
5. Using the rx line from a third serial port, watch the traffic coming out of the drone. Notice the messages all have a Mavlink1 delimiter:

ATTITUDE {time_boot_ms : 4642939, roll : 0.00841472763568163, pitch : -0.6049841046333313, yaw : 3.134385108947754, rollspeed : -0.0005203436594456434, pitchspeed : 0.011106032878160477, yawspeed : 0.0009611798450350761}; delim: 0xFE
SIMSTATE {roll : 0.010485770180821419, pitch : -0.6031168103218079, yaw : 3.1256189346313477, xacc : -5.167874336242676, yacc : 0.0917649194598198, zacc : -8.3803071975708, xgyro : -0.0013637746451422572, ygyro : 0.007096045184880495, zgyro : 0.0012050365330651402, lat : 468800607, lng : -967801686}; delim: 0xFE
AHRS2 {roll : 0.008754936046898365, pitch : -0.6044777631759644, yaw : 3.1205947399139404, altitude : 0.0, lat : 0, lng : 0}; delim: 0xFE
AHRS3 {roll : 0.00841472763568163, pitch : -0.6049841046333313, yaw : 3.134385108947754, altitude : 397.4499816894531, lat : 468800647, lng : -967801684, v1 : 0.0, v2 : 0.0, v3 : 0.0, v4 : 0.0}; delim: 0xFE
HEARTBEAT {type : 2, autopilot : 3, base_mode : 217, custom_mode : 3, system_status : 4, mavlink_version : 3}; delim: 0xFE
GLOBAL_POSITION_INT {time_boot_ms : 4643039, lat : 468800557, lon : -967801684, alt : 397450, relative_alt : 124993, vx : -1009, vy : 2, vz : -1, hdg : 17959}; delim: 0xFE
LOCAL_POSITION_NED {time_boot_ms : 4643039, x : -1.9576263427734375, y : -0.107635498046875, z : -124.9931411743164, vx : -10.09014892578125, vy : 0.025735659524798393, vz : -0.01863064430654049}; delim: 0xFE
ATTITUDE {time_boot_ms : 4643039, roll : 0.00804208219051361, pitch : -0.6060734987258911, yaw : 3.134498357772827, rollspeed : -0.0034370357170701027, pitchspeed : -0.02397013269364834, yawspeed : 0.0016968990676105022}; delim: 0xFE
SIMSTATE {roll : 0.01009929459542036, pitch : -0.6041584014892578, yaw : 3.1257967948913574, xacc : -5.6842360496521, yacc : 0.052538178861141205, zacc : -7.934842586517334, xgyro : -0.0029758054297417402, ygyro : -0.02649005688726902, zgyro : 0.0020633188541978598, lat : 468800516, lng : -967801686}; delim: 0xFE
AHRS2 {roll : 0.00830044411122799, pitch : -0.6055868864059448, yaw : 3.120851755142212, altitude : 0.0, lat : 0, lng : 0}; delim: 0xFE
AHRS3 {roll : 0.00804208219051361, pitch : -0.6060734987258911, yaw : 3.134498357772827, altitude : 397.4499816894531, lat : 468800557, lng : -967801684, v1 : 0.0, v2 : 0.0, v3 : 0.0, v4 : 0.0}; delim: 0xFE
SYSTEM_TIME {time_unix_usec : 1570831512604762, time_boot_ms : 4643039}; delim: 0xFE
AHRS {omegaIx : -0.0026554116047918797, omegaIy : -0.0026360841002315283, omegaIz : -0.002453599590808153, accel_weight : 0.0, renorm_val : 0.0, error_rp : 0.0016389929223805666, error_yaw : 0.009721661917865276}; delim: 0xFE
HWSTATUS {Vcc : 5000, I2Cerr : 0}; delim: 0xFE

6. Upload a mission that contains MAV_CMD_DO_SET_CAM_TRIGG_DIST. Watching the commands sent to the camera and observe the COMMAND_LONG are sent using a Mavlink2 delimiter while the CAMERA_FEEDBACK and other messages are sent using Mavlink1

AHRS3 {roll : 0.0051164585165679455, pitch : -0.6225802302360535, yaw : 3.1369504928588867, altitude : 397.4499816894531, lat : 468799545, lng : -967801683, v1 : 0.0, v2 : 0.0, v3 : 0.0, v4 : 0.0}; delim: 0xFE
COMMAND_LONG {'target_system': 0, 'target_component': 0, 'command': 'MAV_CMD_DO_DIGICAM_CONTROL (203)', 'confirmation': 0, 'param1': 0.0, 'param2': 0.0, 'param3': 0.0, 'param4': 0.0, 'param5': 1.0, 'param6': 0.0, 'param7': 0.0}; delim: 0xFD
CAMERA_FEEDBACK {time_usec : 1570831513800000, target_system : 0, cam_idx : 0, img_idx : 104, lat : 468799475, lng : -967801683, alt_msl : 395.8899841308594, alt_rel : 123.43000030517578, roll : 0.29999998211860657, pitch : -35.77000045776367, yaw : 179.72999572753906, foc_len : 0.0, flags : 0, completed_captures : 0}; delim: 0xFE
ATTITUDE {time_boot_ms : 4644239, roll : 0.005234807264059782, pitch : -0.6250613927841187, yaw : 3.1368649005889893, rollspeed : -0.0008950171759352088, pitchspeed : -0.025688666850328445, yawspeed : -0.000348108122125268}; delim: 0xFE

7. Use the manual trigger button in the GCS and observe the same behavior as step 6.

Version
Tested with ardupilot rev 05d1046

Platform
only tested on copter so far.

Airframe type
quad

Hardware type
SITL

Logs
See steps to reproduce above.

[rmackay9]

Thanks very much for this detailed report. I've discussed with @peterbarker who is more an expert in this area and I hope we can get to the bottom of it in the near future.

[pklapperich]

I will say that I haven't tried in simulator on ArduCopter 3.6, but on the drone at least ArduCopter 3.6 does behave correctly.

[rmackay9]

Peter Barker has created a fix for this issue here: #12576

I plan to test this fix this week.

[rmackay9]

@pklapperich, do you think you could re-test with this patch? If it makes it easier I can produce a binary for you especially if you tell me what type of flight controller you are using.

[pklapperich]

Sorry for the delay, yes I can test later today. I'm using software in the loop.

[pklapperich]

With the physical camera attached to ArduCopter SITL via a USB serial adapter to uartE things appear to be working fine now (rev 871305b4f). I tested both clicking the "Trigger Camera" button in the ground station as well as flying a mission that included MAV_CMD_DO_SET_CAM_TRIGG_DIST. In both cases the camera received the trigger command in a MAVLINK1 frame and snapped a photo.

But then I changed the SERIAL4_PROTOCOL parameter from MAVLink1 to MAVLINK2, and the camera still lit up green (indicating it received the GLOBAL_POSITION_INT messages it wanted) and also took pictures when I clicked the "Trigger Camera" button in the ground station. This camera device does not understand MAVLINK2, so I didn't expect this to work. I'm guessing that the patch is causing ArduPilot to ignore the SERIAL4_PROTOCOL parameter and instead communicates based on the heartbeats it's receiving. Is this expected behavior? I'm not complaining; I'm just not sure it was intentional.

When I changed the SERIAL4_PROTOCOL parameter to GPS the camera didn't work (as was expected).

[peterbarker]

@pklapperich We still start off by sending mavlink1 on a mavlink2 channel - if mavlink2 signing is disabled. If we receive a mavlink2 packet we start to send mavlink2.

So I think that's all good.