zrzk
/
ardupilot
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
Browse Source

Copter: Autotune update

mission-4.1.18
Leonard Hall 10 years ago committed by Randy Mackay
parent
1900cc7582
commit
05103d9f9c
1 changed files with 16 additions and 17 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Unified View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 33
      ArduCopter/control_autotune.pde

33
ArduCopter/control_autotune.pde
Unescape View File

@ -52,12 +52,8 @@
#define AUTOTUNE_RD_STEP 0.05f // minimum increment when increasing/decreasing Rate D term #define AUTOTUNE_RD_STEP 0.05f // minimum increment when increasing/decreasing Rate D term
#define AUTOTUNE_RP_STEP 0.05f // minimum increment when increasing/decreasing Rate P term #define AUTOTUNE_RP_STEP 0.05f // minimum increment when increasing/decreasing Rate P term
#define AUTOTUNE_SP_STEP 0.05f // minimum increment when increasing/decreasing Stab P term #define AUTOTUNE_SP_STEP 0.05f // minimum increment when increasing/decreasing Stab P term
#define AUTOTUNE_RD_TEST_NOISE 0.05f // Rate D gains are reduced to 50% of their maximum value discovered during tuning
#define AUTOTUNE_RD_BACKOFF 4.0f // Rate D gains are reduced to 50% of their maximum value discovered during tuning
#define AUTOTUNE_RP_BACKOFF 1.0f // Rate P gains are reduced to 97.5% of their maximum value discovered during tuning
#define AUTOTUNE_SP_BACKOFF 1.0f // Stab P gains are reduced to 60% of their maximum value discovered during tuning
#define AUTOTUNE_PI_RATIO_FOR_TESTING 0.1f // I is set 10x smaller than P during testing #define AUTOTUNE_PI_RATIO_FOR_TESTING 0.1f // I is set 10x smaller than P during testing
#define AUTOTUNE_PI_RATIO_FINAL 2.5f // I is set 1x P after testing #define AUTOTUNE_PI_RATIO_FINAL 1.0f // I is set 1x P after testing
#define AUTOTUNE_YAW_PI_RATIO_FINAL 0.1f // I is set 1x P after testing #define AUTOTUNE_YAW_PI_RATIO_FINAL 0.1f // I is set 1x P after testing
#define AUTOTUNE_RD_MIN 0.004f // minimum Rate D value #define AUTOTUNE_RD_MIN 0.004f // minimum Rate D value
#define AUTOTUNE_RD_MAX 0.050f // maximum Rate D value #define AUTOTUNE_RD_MAX 0.050f // maximum Rate D value
@ -67,13 +63,16 @@
#define AUTOTUNE_RP_MAX 2.0f // maximum Rate P value #define AUTOTUNE_RP_MAX 2.0f // maximum Rate P value
#define AUTOTUNE_SP_MAX 20.0f // maximum Stab P value #define AUTOTUNE_SP_MAX 20.0f // maximum Stab P value
#define AUTOTUNE_SP_MIN 0.5f // maximum Stab P value #define AUTOTUNE_SP_MIN 0.5f // maximum Stab P value
#define AUTOTUNE_ACCEL_RP_BACKOFF 1.0f // back off from maximum acceleration
#define AUTOTUNE_ACCEL_Y_BACKOFF 0.75f // back off from maximum acceleration
#define AUTOTUNE_RP_ACCEL_MIN 36000.0f // Minimum acceleration for Roll and Pitch #define AUTOTUNE_RP_ACCEL_MIN 36000.0f // Minimum acceleration for Roll and Pitch
#define AUTOTUNE_Y_ACCEL_MIN 9000.0f // Minimum acceleration for Yaw #define AUTOTUNE_Y_ACCEL_MIN 9000.0f // Minimum acceleration for Yaw
#define AUTOTUNE_Y_FILT_FREQ 10.0f // Minimum acceleration for Roll and Pitch #define AUTOTUNE_Y_FILT_FREQ 10.0f // Minimum acceleration for Roll and Pitch
#define AUTOTUNE_SUCCESS_COUNT 4 // how many successful iterations we need to freeze at current gains #define AUTOTUNE_SUCCESS_COUNT 4 // how many successful iterations we need to freeze at current gains
#define AUTOTUNE_D_UP_DOWN_MARGIN 0.2f // The margin below the target that we tune D in #define AUTOTUNE_D_UP_DOWN_MARGIN 0.2f // The margin below the target that we tune D in
#define AUTOTUNE_RD_BACKOFF 1.0f // Rate D gains are reduced to 50% of their maximum value discovered during tuning
#define AUTOTUNE_RP_BACKOFF 1.0f // Rate P gains are reduced to 97.5% of their maximum value discovered during tuning
#define AUTOTUNE_SP_BACKOFF 1.0f // Stab P gains are reduced to 60% of their maximum value discovered during tuning
#define AUTOTUNE_ACCEL_RP_BACKOFF 1.0f // back off from maximum acceleration
#define AUTOTUNE_ACCEL_Y_BACKOFF 0.75f // back off from maximum acceleration
// roll and pitch axes // roll and pitch axes
#define AUTOTUNE_TARGET_ANGLE_RLLPIT_CD 2000 // target angle during TESTING_RATE step that will cause us to move to next step #define AUTOTUNE_TARGET_ANGLE_RLLPIT_CD 2000 // target angle during TESTING_RATE step that will cause us to move to next step
@ -495,7 +494,7 @@ static void autotune_attitude_control()
} }
break; break;
case AUTOTUNE_TYPE_RP_UP: case AUTOTUNE_TYPE_RP_UP:
autotune_twitching_test(rotation_rate, autotune_target_rate, autotune_test_min, autotune_test_max); autotune_twitching_test(rotation_rate, autotune_target_rate*(1+0.5f*g.autotune_aggressiveness), autotune_test_min, autotune_test_max);
autotune_twitching_measure_acceleration(autotune_test_accel_max, rotation_rate, rate_max); autotune_twitching_measure_acceleration(autotune_test_accel_max, rotation_rate, rate_max);
if (lean_angle >= autotune_target_angle) { if (lean_angle >= autotune_target_angle) {
autotune_state.step = AUTOTUNE_STEP_UPDATE_GAINS; autotune_state.step = AUTOTUNE_STEP_UPDATE_GAINS;
@ -578,13 +577,13 @@ static void autotune_attitude_control()
case AUTOTUNE_TYPE_RP_UP: case AUTOTUNE_TYPE_RP_UP:
switch (autotune_state.axis) { switch (autotune_state.axis) {
case AUTOTUNE_AXIS_ROLL: case AUTOTUNE_AXIS_ROLL:
autotune_updating_p_up_d_down(tune_roll_rd, AUTOTUNE_RD_MIN, AUTOTUNE_RD_STEP, tune_roll_rp, AUTOTUNE_RP_MIN, AUTOTUNE_RP_MAX, AUTOTUNE_RP_STEP, autotune_target_rate, autotune_test_min, autotune_test_max); autotune_updating_p_up_d_down(tune_roll_rd, AUTOTUNE_RD_MIN, AUTOTUNE_RD_STEP, tune_roll_rp, AUTOTUNE_RP_MIN, AUTOTUNE_RP_MAX, AUTOTUNE_RP_STEP, autotune_target_rate*(1+0.5f*g.autotune_aggressiveness), autotune_test_min, autotune_test_max);
break; break;
case AUTOTUNE_AXIS_PITCH: case AUTOTUNE_AXIS_PITCH:
autotune_updating_p_up_d_down(tune_pitch_rd, AUTOTUNE_RD_MIN, AUTOTUNE_RD_STEP, tune_pitch_rp, AUTOTUNE_RP_MIN, AUTOTUNE_RP_MAX, AUTOTUNE_RP_STEP, autotune_target_rate, autotune_test_min, autotune_test_max); autotune_updating_p_up_d_down(tune_pitch_rd, AUTOTUNE_RD_MIN, AUTOTUNE_RD_STEP, tune_pitch_rp, AUTOTUNE_RP_MIN, AUTOTUNE_RP_MAX, AUTOTUNE_RP_STEP, autotune_target_rate*(1+0.5f*g.autotune_aggressiveness), autotune_test_min, autotune_test_max);
break; break;
case AUTOTUNE_AXIS_YAW: case AUTOTUNE_AXIS_YAW:
autotune_updating_p_up_d_down(tune_yaw_rLPF, AUTOTUNE_RLPF_MIN, AUTOTUNE_RD_STEP, tune_yaw_rp, AUTOTUNE_RP_MIN, AUTOTUNE_RP_MAX, AUTOTUNE_RP_STEP, autotune_target_rate, autotune_test_min, autotune_test_max); autotune_updating_p_up_d_down(tune_yaw_rLPF, AUTOTUNE_RLPF_MIN, AUTOTUNE_RD_STEP, tune_yaw_rp, AUTOTUNE_RP_MIN, AUTOTUNE_RP_MAX, AUTOTUNE_RP_STEP, autotune_target_rate*(1+0.5f*g.autotune_aggressiveness), autotune_test_min, autotune_test_max);
break; break;
} }
break; break;
@ -633,16 +632,16 @@ static void autotune_attitude_control()
autotune_state.tune_type++; autotune_state.tune_type++;
switch (autotune_state.axis) { switch (autotune_state.axis) {
case AUTOTUNE_AXIS_ROLL: case AUTOTUNE_AXIS_ROLL:
tune_roll_rd = max(AUTOTUNE_RD_MIN, tune_roll_rd * (1.0f-AUTOTUNE_RD_BACKOFF*max(0.0f,(g.autotune_aggressiveness-AUTOTUNE_RD_TEST_NOISE)))); tune_roll_rd = max(AUTOTUNE_RD_MIN, tune_roll_rd * AUTOTUNE_RD_BACKOFF);
tune_roll_rp = max(AUTOTUNE_RP_MIN, tune_roll_rp * (1.0f-AUTOTUNE_RD_BACKOFF*max(0.0f,(g.autotune_aggressiveness-AUTOTUNE_RD_TEST_NOISE)))); tune_roll_rp = max(AUTOTUNE_RP_MIN, tune_roll_rp * AUTOTUNE_RD_BACKOFF);
break; break;
case AUTOTUNE_AXIS_PITCH: case AUTOTUNE_AXIS_PITCH:
tune_pitch_rd = max(AUTOTUNE_RD_MIN, tune_pitch_rd * (1.0f-AUTOTUNE_RD_BACKOFF*max(0.0f,(g.autotune_aggressiveness-AUTOTUNE_RD_TEST_NOISE)))); tune_pitch_rd = max(AUTOTUNE_RD_MIN, tune_pitch_rd * AUTOTUNE_RD_BACKOFF);
tune_pitch_rp = max(AUTOTUNE_RP_MIN, tune_pitch_rp * (1.0f-AUTOTUNE_RD_BACKOFF*max(0.0f,(g.autotune_aggressiveness-AUTOTUNE_RD_TEST_NOISE)))); tune_pitch_rp = max(AUTOTUNE_RP_MIN, tune_pitch_rp * AUTOTUNE_RD_BACKOFF);
break; break;
case AUTOTUNE_AXIS_YAW: case AUTOTUNE_AXIS_YAW:
tune_yaw_rLPF = max(AUTOTUNE_RLPF_MIN, tune_yaw_rLPF * (1.0f-AUTOTUNE_RD_BACKOFF*max(0.0f,(g.autotune_aggressiveness-AUTOTUNE_RD_TEST_NOISE)))); tune_yaw_rLPF = max(AUTOTUNE_RLPF_MIN, tune_yaw_rLPF * AUTOTUNE_RD_BACKOFF);
tune_yaw_rp = max(AUTOTUNE_RP_MIN, tune_yaw_rp * (1.0f-AUTOTUNE_RD_BACKOFF*max(0.0f,(g.autotune_aggressiveness-AUTOTUNE_RD_TEST_NOISE)))); tune_yaw_rp = max(AUTOTUNE_RP_MIN, tune_yaw_rp * AUTOTUNE_RD_BACKOFF);
break; break;
} }
break; break;

Write Preview
Loading…
Cancel
Save