Remove actuators_pprz

[dewagter]

1. there are now 3 types of numbers of actuators and 2 types of control command arrays:

   • SERVO_NAME_IDX -> actuators[] -> must also be used in ABI messages
   • INDI_MATRIX_SEQUENCE -> actuators_pprz[] is called actuator but contains controller outputs which not always map 1 to 1 to actuators
   • SERVO_NAME no="" -> CAN: should never be used in code. I would like to remove the define from the airframe.h file to avoid errors
   • the sequence of commands defined in the airframe file is often used but not always.

2. there are now problems with adding virtual actuators when the number of commands becomes larger than the number of actuators and there are no checks if INDI_NUM_ACT is smaller than the actuator array size: see Rot mech fix pr #3180

3. actuators_pprz index numbers are often the same numbers as the actuators, so many users confuse them, but they are not always the same. One is the sequence in the INDI and the other is the list of servos.

4. SERVO_NAME_IDX and SERVO_NAME are sometimes confused in code: Hide actuator driver number from code #3206

5. Safety features on commands are OVERRULED (e.g. FAILSAFE) when using INDI: command-laws in xml must re-implement functions like kill

6. Command laws contain raw INDI indexes: not very readable + eff_sheduling relies on a specific order but cannot check this.

7. RC_COMMANDS and RC_AUTO_COMMANDS are broken with INDI

8. The simulator needs harmonized actuator commands (-MAX_PPRZ -> MAX_PPRZ), which required the actuators_pprz but this can be solved differently: actuators[ ] array of structs with pprz units and driver units #3205

9. some modules write PWM directly to actuators and ignore the servo MIN and MAX

• rotrocraft_cam
• switch_servo

10. there are unused commands in many airframe files: #Remove command_pitch and command_roll and command_yaw #3208

-> so I want to remove actuators_pprz: and instead do what was always the case before

control -> stabilization_cmd with the command sequence of the airframe file, and then set the actuators in the ActuatorsFromCommands block, just like the name says.

You can then:

• make as many (virtual) commands as you want by adding commands
• use default kill/failsafe functions
• read the order of commands in the airframe instead of the code
• forced to define which command goes where so no confusion with different numbering in actuator and actuator_pprz
• no double numbering in actuators and actuators_pprz
• RC_COMMANDS etc repaired...

[gautierhattenberger]

@dewagter I'm all in favor of removing actuators_pprz !

• In the "reference" design, the output of stabilization always goes to the commands vector, which is mapped to actuators with the command_law section of the airframe (and this is what you did). I also would like to be able to compile without the ROLL, PITCH, YAW command when they are not used. THRUST command is annoying because I thing it is used for the GCS display, but we could probably improve that as well.
• From what I can see, you are now feeding the stabilization_cmd array. I will be needed to copy it to the real commands array at some point, and set_rotorcraft_commands
• I'm not sure about the proper way to deal with the motors_on directly implemented in the airframe. It is really annoying and incompatible with dual MCU support (FBW doesn't have access to this variable).
• Last point, I would like to change the actuators array of pprz_t by an array of structures (with the type of actuators, feedback value if available, current value in pprz and actuators units, etc). This can be done later I think, but could also help to manage the different status (kill, failsafe, motor on/off, running, etc).

[dewagter]

Thanks a lot for the reply. I like the idea of actuators as an array of structs. In the meantime, I proposed an intermediate step to standardize it: #3205 but I like making structs even better.

SetRotorcraftCommands still needs to be done indeed. One option would be: https://github.com/paparazzi/paparazzi/pull/3203/files#diff-02e75cc9d71b8fb6997f5286311725c6d7cfc421e65da9e9b96f5e0a99ed46b2R133