Add tailsitter heading alignment phase during forward transition (Pr yaw alignment) - #39
Add tailsitter heading alignment phase during forward transition (Pr yaw alignment)#39Tarun24-airbound wants to merge 21 commits into
Conversation
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the You can disable this status message by setting the
✨ Finishing touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Tarun24-airbound
assigned
Tarun24-airbound,
botmayank and
atharva-airbound and unassigned
Tarun24-airbound
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 4
🤖 Fix all issues with AI agents
In `@ArduPlane/tailsitter.cpp`:
- Around line 924-972: The timeout start (align_phase_start_ms) is set
unconditionally on detecting a "new attempt" which can expire before alignment
actually becomes active; change the logic so align_phase_start_ms is only
initialized when alignment is actually active (should_run_alignment == true) —
e.g., on the new-attempt branch do not set align_phase_start_ms unconditionally
but only set it when should_run_alignment is true (or set it to 0/clear it when
should_run_alignment is false), and ensure you detect the transition from
inactive->active to start the timer (use last_run_ms and
alignment_completed_for_this_flight to gate this), updating the places that use
align_phase_start_ms, timeout_expired, and WAIT_DELAY_MS accordingly.
In `@libraries/AC_AttitudeControl/AC_AttitudeControl.h`:
- Around line 229-234: relax_attitude_controllers() currently initializes
_attitude_target and _attitude_ang_error but leaves _attitude_body
uninitialized, which can cause get_attitude_body_quat() to return garbage during
early startup; update relax_attitude_controllers() to set _attitude_body =
_attitude_target (matching the pattern used for _attitude_target and
_attitude_ang_error) so _attitude_body has a valid value before
attitude_controller_run_quat() runs and before get_attitude_body_quat() or
logging uses it.
In
`@libraries/AP_HAL_ChibiOS/hwdef/Pixhawk6C-bdshot/scripts/ab_door_auto_control.lua`:
- Around line 90-92: The OVERRIDE parameter (declared with
param:add_param(PARAM_TABLE_KEY, PARAM_IDX.OVERRIDE, "OVERRIDE", 0)) currently
only affects manual timeout but does not stop automatic behavior; fix by reading
the OVERRIDE param value wherever auto mode is triggered and skip calling
run_auto_mode() when OVERRIDE==1. Update both places that invoke run_auto_mode()
to first check the parameter (using the same PARAM_TABLE_KEY and
PARAM_IDX.OVERRIDE lookup method used elsewhere) and return/continue without
running auto logic if the override is set.
In `@modules/littlefs`:
- Line 1: Add a new submodule entry for modules/littlefs to the .gitmodules file
so CI can initialize the submodule; specifically, add a [submodule
"modules/littlefs"] section containing path = modules/littlefs and url =
<repository-url> (replace <repository-url> with the actual littlefs repo URL).
Ensure the entry exactly matches the submodule path "modules/littlefs" and
commit the updated .gitmodules so git submodule update succeeds.
| // Static Variables (State Tracking) | ||
| static uint32_t align_phase_start_ms = 0; | ||
| static uint32_t alignment_done_ms = 0; | ||
| static uint32_t last_run_ms = 0; | ||
| static uint32_t last_log_ms = 0; | ||
| // NEW: One-Shot Flag to prevent infinite loops | ||
| static bool alignment_completed_for_this_flight = false; | ||
|
|
||
| // --- DETECT NEW FLIGHT/TRANSITION ENTRY --- | ||
| // If this function hasn't run for >200ms, assume it's a new attempt. | ||
| if (now_ - last_run_ms > 200) { | ||
| align_phase_start_ms = now_; | ||
| alignment_done_ms = 0; | ||
| last_log_ms = 0; | ||
| alignment_completed_for_this_flight = false; // Reset flag for new transition | ||
| } | ||
| last_run_ms = now_; | ||
|
|
||
| // Check Validity & Only run if we haven't finished aligning yet | ||
| bool should_run_alignment = !alignment_completed_for_this_flight && | ||
| (plane.control_mode == &plane.mode_auto || plane.control_mode == &plane.mode_guided) && | ||
| (plane.nav_controller != nullptr); | ||
|
|
||
| if (should_run_alignment) { | ||
| // Disable weathervane during alignment | ||
| if (quadplane.weathervane != nullptr) { | ||
| quadplane.weathervane->allow_weathervaning(false); | ||
| } | ||
|
|
||
| int32_t target_bearing_cd = plane.nav_controller->target_bearing_cd(); | ||
| int32_t current_yaw_cd = quadplane.ahrs.yaw_sensor; | ||
| int32_t error_cd = wrap_180_cd(target_bearing_cd - current_yaw_cd); | ||
| Vector3f gyro = quadplane.ahrs.get_gyro(); | ||
| float yaw_rate_deg = degrees(gyro.z); | ||
|
|
||
| // LOGIC: Are we aligned right now? | ||
| bool is_aligned = (abs(error_cd) <= ALIGN_TOLERANCE_CD) && (abs(yaw_rate_deg) <= MAX_SPIN_RATE_DEG); | ||
|
|
||
| // TIMER LOGIC | ||
| if (is_aligned) { | ||
| if (alignment_done_ms == 0) alignment_done_ms = now_; | ||
| } else { | ||
| alignment_done_ms = 0; // Reset if we drift out | ||
| } | ||
|
|
||
| // EXIT CRITERIA | ||
| bool wait_complete = (alignment_done_ms != 0) && (now_ - alignment_done_ms >= WAIT_DELAY_MS); | ||
| bool timeout_expired = (now_ - align_phase_start_ms >= ALIGN_PHASE_LIMIT_MS); | ||
|
|
There was a problem hiding this comment.
Start the alignment timeout only when alignment is actually active.
align_phase_start_ms is set on entry regardless of should_run_alignment. If the vehicle enters TRANSITION_ANGLE_WAIT_FW in a non-AUTO/GUIDED mode and later switches into AUTO/GUIDED, the timeout can already be expired, causing alignment to be skipped immediately. Consider arming the timer when alignment becomes active (or clearing it while inactive).
💡 One way to gate the timer
- // --- DETECT NEW FLIGHT/TRANSITION ENTRY ---
- // If this function hasn't run for >200ms, assume it's a new attempt.
+ // --- DETECT NEW FLIGHT/TRANSITION ENTRY ---
if (now_ - last_run_ms > 200) {
align_phase_start_ms = now_;
alignment_done_ms = 0;
last_log_ms = 0;
alignment_completed_for_this_flight = false; // Reset flag for new transition
}
last_run_ms = now_;
// Check Validity & Only run if we haven't finished aligning yet
bool should_run_alignment = !alignment_completed_for_this_flight &&
(plane.control_mode == &plane.mode_auto || plane.control_mode == &plane.mode_guided) &&
(plane.nav_controller != nullptr);
+
+ if (!should_run_alignment) {
+ align_phase_start_ms = 0;
+ alignment_done_ms = 0;
+ } else if (align_phase_start_ms == 0) {
+ align_phase_start_ms = now_;
+ }🤖 Prompt for AI Agents
In `@ArduPlane/tailsitter.cpp` around lines 924 - 972, The timeout start
(align_phase_start_ms) is set unconditionally on detecting a "new attempt" which
can expire before alignment actually becomes active; change the logic so
align_phase_start_ms is only initialized when alignment is actually active
(should_run_alignment == true) — e.g., on the new-attempt branch do not set
align_phase_start_ms unconditionally but only set it when should_run_alignment
is true (or set it to 0/clear it when should_run_alignment is false), and ensure
you detect the transition from inactive->active to start the timer (use
last_run_ms and alignment_completed_for_this_flight to gate this), updating the
places that use align_phase_start_ms, timeout_expired, and WAIT_DELAY_MS
accordingly.
| // Return the body-to-NED error attitude used by the quadplane-specific attitude control input methods | ||
| Quaternion get_attitude_error_quat() const { return _attitude_ang_error; } | ||
|
|
||
| // Return the body-to-NED attitude used by the quadplane-specific attitude control input methods | ||
| Quaternion get_attitude_body_quat() const { return _attitude_body; } | ||
|
|
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
fd -t f "AC_AttitudeControl" | head -20Repository: AirboundInc/ardupilot
Length of output: 734
🏁 Script executed:
find . -name "*.h" -o -name "*.cpp" | xargs grep -l "AC_AttitudeControl" | grep -E "AC_AttitudeControl\.(h|cpp)$"Repository: AirboundInc/ardupilot
Length of output: 170
🏁 Script executed:
git ls-files | grep -i "attitude" | grep -E "\.(h|cpp)$"Repository: AirboundInc/ardupilot
Length of output: 1220
🏁 Script executed:
wc -l libraries/AC_AttitudeControl/AC_AttitudeControl.hRepository: AirboundInc/ardupilot
Length of output: 118
🏁 Script executed:
sed -n '220,240p' libraries/AC_AttitudeControl/AC_AttitudeControl.hRepository: AirboundInc/ardupilot
Length of output: 1262
🏁 Script executed:
sed -n '500,520p' libraries/AC_AttitudeControl/AC_AttitudeControl.hRepository: AirboundInc/ardupilot
Length of output: 1017
🏁 Script executed:
grep -n "AC_AttitudeControl::AC_AttitudeControl" libraries/AC_AttitudeControl/AC_AttitudeControl.cpp | head -5Repository: AirboundInc/ardupilot
Length of output: 47
🏁 Script executed:
grep -n "relax_attitude_controllers" libraries/AC_AttitudeControl/AC_AttitudeControl.cppRepository: AirboundInc/ardupilot
Length of output: 122
🏁 Script executed:
grep -n "attitude_controller_run_quat" libraries/AC_AttitudeControl/AC_AttitudeControl.cppRepository: AirboundInc/ardupilot
Length of output: 561
🏁 Script executed:
sed -n '168,190p' libraries/AC_AttitudeControl/AC_AttitudeControl.cppRepository: AirboundInc/ardupilot
Length of output: 840
🏁 Script executed:
sed -n '708,750p' libraries/AC_AttitudeControl/AC_AttitudeControl.cppRepository: AirboundInc/ardupilot
Length of output: 2304
🏁 Script executed:
grep -n "_attitude_body" libraries/AC_AttitudeControl/AC_AttitudeControl.cpp | head -20Repository: AirboundInc/ardupilot
Length of output: 104
🏁 Script executed:
sed -n '750,760p' libraries/AC_AttitudeControl/AC_AttitudeControl.cppRepository: AirboundInc/ardupilot
Length of output: 515
🏁 Script executed:
grep -n "get_attitude_body_quat\|get_attitude_target_quat\|get_attitude_error_quat" libraries/AC_AttitudeControl/AC_AttitudeControl.cppRepository: AirboundInc/ardupilot
Length of output: 47
🏁 Script executed:
grep -rn "get_attitude_body_quat" --include="*.cpp" --include="*.h" | head -20Repository: AirboundInc/ardupilot
Length of output: 307
🏁 Script executed:
grep -n "AC_AttitudeControl::" libraries/AC_AttitudeControl/AC_AttitudeControl.cpp | grep -E "(constructor|AC_AttitudeControl\()" | head -5Repository: AirboundInc/ardupilot
Length of output: 47
🏁 Script executed:
sed -n '1,100p' libraries/AC_AttitudeControl/AC_AttitudeControl.cpp | grep -A 20 "AC_AttitudeControl::"Repository: AirboundInc/ardupilot
Length of output: 907
🏁 Script executed:
cat libraries/AC_AttitudeControl/AC_AttitudeControl.h | grep -A 5 "class AC_AttitudeControl"Repository: AirboundInc/ardupilot
Length of output: 570
🏁 Script executed:
grep -n "Quaternion.*initialise\|Quaternion()" libraries/AC_AttitudeControl/AC_AttitudeControl.cpp | head -10Repository: AirboundInc/ardupilot
Length of output: 47
🏁 Script executed:
sed -n '1,50p' libraries/AC_AttitudeControl/AC_AttitudeControl.cpp | grep -A 30 "AC_AttitudeControl::" | head -40Repository: AirboundInc/ardupilot
Length of output: 1296
🏁 Script executed:
grep -n "AC_AttitudeControl.*{" libraries/AC_AttitudeControl/AC_AttitudeControl.h | head -5Repository: AirboundInc/ardupilot
Length of output: 190
🏁 Script executed:
sed -n '1,100p' libraries/AC_AttitudeControl/AC_AttitudeControl.h | grep -A 50 "AC_AttitudeControl("Repository: AirboundInc/ardupilot
Length of output: 2361
Initialize _attitude_body in relax_attitude_controllers() to ensure valid state before first use.
The getter get_attitude_body_quat() may be called before attitude_controller_run_quat() has executed (e.g., during early logging in startup), but _attitude_body is only assigned within attitude_controller_run_quat() at line 756. Unlike _attitude_target and _attitude_ang_error, which are explicitly initialized in relax_attitude_controllers(), _attitude_body is left uninitialized. The getter is already in use (ArduPlane/Log.cpp:27) and could return an uninitialized quaternion.
Add _attitude_body = _attitude_target; in relax_attitude_controllers() after line 172 to match the initialization pattern of other attitude members:
Suggested initialization
void AC_AttitudeControl::relax_attitude_controllers()
{
// Initialize the attitude variables to the current attitude
_ahrs.get_quat_body_to_ned(_attitude_target);
_attitude_target.to_euler(_euler_angle_target);
_attitude_ang_error.initialise();
+ _attitude_body = _attitude_target;📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| // Return the body-to-NED error attitude used by the quadplane-specific attitude control input methods | |
| Quaternion get_attitude_error_quat() const { return _attitude_ang_error; } | |
| // Return the body-to-NED attitude used by the quadplane-specific attitude control input methods | |
| Quaternion get_attitude_body_quat() const { return _attitude_body; } | |
| void AC_AttitudeControl::relax_attitude_controllers() | |
| { | |
| // Initialize the attitude variables to the current attitude | |
| _ahrs.get_quat_body_to_ned(_attitude_target); | |
| _attitude_target.to_euler(_euler_angle_target); | |
| _attitude_ang_error.initialise(); | |
| _attitude_body = _attitude_target; | |
| } |
🤖 Prompt for AI Agents
In `@libraries/AC_AttitudeControl/AC_AttitudeControl.h` around lines 229 - 234,
relax_attitude_controllers() currently initializes _attitude_target and
_attitude_ang_error but leaves _attitude_body uninitialized, which can cause
get_attitude_body_quat() to return garbage during early startup; update
relax_attitude_controllers() to set _attitude_body = _attitude_target (matching
the pattern used for _attitude_target and _attitude_ang_error) so _attitude_body
has a valid value before attitude_controller_run_quat() runs and before
get_attitude_body_quat() or logging uses it.
| -- Override parameter definition | ||
| --[[ @Param: DOOR_OVERRIDE, @DisplayName: Door Servo override, @Description: Enable/Disable for doors script auto logic, @Range: 0 1 --]] | ||
| assert(param:add_param(PARAM_TABLE_KEY, PARAM_IDX.OVERRIDE, "OVERRIDE", 0)) |
There was a problem hiding this comment.
OVERRIDE param doesn’t actually disable auto logic.
The parameter description says it enables/disables auto logic, but the implementation only bypasses manual-override timeout; auto mode still runs when the RC hasn’t moved. Either update the description or gate run_auto_mode() when OVERRIDE=1.
🔧 Suggested fix to honor DOOR_OVERRIDE by skipping auto mode
- if not check_manual_override() then
- run_auto_mode()
- end
+ local auto_disabled = (config.cache.man_override == 1)
+ if not check_manual_override() and not auto_disabled then
+ run_auto_mode()
+ endAlso applies to: 372-377
🤖 Prompt for AI Agents
In
`@libraries/AP_HAL_ChibiOS/hwdef/Pixhawk6C-bdshot/scripts/ab_door_auto_control.lua`
around lines 90 - 92, The OVERRIDE parameter (declared with
param:add_param(PARAM_TABLE_KEY, PARAM_IDX.OVERRIDE, "OVERRIDE", 0)) currently
only affects manual timeout but does not stop automatic behavior; fix by reading
the OVERRIDE param value wherever auto mode is triggered and skip calling
run_auto_mode() when OVERRIDE==1. Update both places that invoke run_auto_mode()
to first check the parameter (using the same PARAM_TABLE_KEY and
PARAM_IDX.OVERRIDE lookup method used elsewhere) and return/continue without
running auto logic if the override is set.
| @@ -0,0 +1 @@ | |||
| Subproject commit 34be692aafaaf4319a18e415b08aa4332697408b | |||
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
# Verify .gitmodules contents and submodule configuration
echo "=== Contents of .gitmodules ==="
if [ -f .gitmodules ]; then
cat .gitmodules
else
echo ".gitmodules file not found"
fi
echo ""
echo "=== Submodule paths ==="
git config -f .gitmodules --get-regexp '^submodule\..*\.path$' 2>/dev/null || echo "No submodule paths found"
echo ""
echo "=== Submodule URLs ==="
git config -f .gitmodules --get-regexp '^submodule\..*\.url$' 2>/dev/null || echo "No submodule URLs found"
echo ""
echo "=== Check modules/littlefs specifically ==="
git config -f .gitmodules --get 'submodule.modules/littlefs.url' 2>/dev/null || echo "No entry for modules/littlefs"
git config -f .gitmodules --get 'submodule.modules/littlefs.path' 2>/dev/null || echo "No path entry for modules/littlefs"Repository: AirboundInc/ardupilot
Length of output: 3603
Add modules/littlefs entry to .gitmodules to unblock CI submodule initialization.
The .gitmodules file is missing a configuration entry for modules/littlefs, but the submodule pointer in the working tree has been updated. This will cause CI to fail during git submodule update with "fatal: No url found for submodule path 'modules/littlefs' in .gitmodules". Add the following entry to .gitmodules:
[submodule "modules/littlefs"]
path = modules/littlefs
url = <repository-url>
Replace <repository-url> with the appropriate littlefs repository URL.
🧰 Tools
🪛 GitHub Actions: test chibios
[error] 1-1: Submodule update failed. Command 'git -c protocol.version=2 submodule update --init --force --depth=1 --recursive' reported: fatal: No url found for submodule path 'modules/littlefs' in .gitmodules.
🪛 GitHub Actions: test Linux SBC
[error] 1-1: fatal: No url found for submodule path 'modules/littlefs' in .gitmodules. This occurred during 'git submodule update --init --force --depth=1 --recursive'.
🤖 Prompt for AI Agents
In `@modules/littlefs` at line 1, Add a new submodule entry for modules/littlefs
to the .gitmodules file so CI can initialize the submodule; specifically, add a
[submodule "modules/littlefs"] section containing path = modules/littlefs and
url = <repository-url> (replace <repository-url> with the actual littlefs repo
URL). Ensure the entry exactly matches the submodule path "modules/littlefs" and
commit the updated .gitmodules so git submodule update succeeds.
| quadplane.pos_control->set_pos_target_z_from_climb_rate_cm(0.0f); | ||
|
|
||
| // Control | ||
| quadplane.pos_control->update_z_controller(); |
There was a problem hiding this comment.
We are manually calling update_z_controller() here but there is a comment in quadplane.cpp mentioning this shouldn't be done during transition
ardupilot/ArduPlane/quadplane.cpp
Line 1040 in a095c6c
Here is a relevant PR from when the comment was added: ArduPilot#19286
Needs investigation.
There was a problem hiding this comment.
Can try calling hold_hover(0) instead of running the position controllers to get the roll and pitch setpoints.
Will try this when I have time.
- Use the vertical accuracy along with the satellite count for switching between GPS instances - Use separate minimum and maximum thresholds for switching as a form of hysteresis - Check for VTOL/landing mission commands in Auto mode - Set the run interval to 1 second
- Check for the existance of both GPS instances before running the GPS instance switching logic
Replace mentions of hacc in the comments
Change yaw control disable thrust angle error limit from 60 to 90 Change linear scale yaw control thrust angle error limit from (45, 60) to (60, 90)
…airbound-4575-rc2
…s' into pr-airbound-4575-rc2
Add 4575 rc2 features given below: Restore desired pitch to 0 when actual pitch exceeds Q_A_RELX_HI(relax attitude) Switch gps based on vertical accuracy build automation fix Add motor disarm check for autodisarm. Change yaw controller thrust angle error limits
…er alignment and handover logic
96dd024 to
7a31a29
Compare
Greptile SummaryThis PR adds a heading alignment phase for tailsitter aircraft that yaws to a target bearing and holds for 1 second before proceeding with the forward transition, using a 6-second timeout and one-shot logic. It also refactors the attitude-relaxation logic in
Confidence Score: 3/5Not safe to merge without resolving the AP_Param index collision in AC_AttitudeControl; upgrading any previously flown vehicle will silently corrupt the RELX_LO threshold, activating pitch-setpoint suppression during hover. The attitude-relaxation refactor reuses AP_Param group index 21 for the new RELX_LO parameter while the old RELX_TC lived there. On any vehicle that stored a prior RELX_TC value, upgrading loads that value (e.g., 0.4) into RELX_LO, setting the deactivation threshold to 0.4° instead of 30°. Because RELX_EN is now forced to 1 in defaults.parm, the relaxation will be permanently active, suppressing the pitch setpoint during normal hover. Additionally, the tailsitter alignment phase carries several static-variable fragility issues identified in prior review rounds that are still unresolved. libraries/AC_AttitudeControl/AC_AttitudeControl.cpp — parameter index migration and static-variable lifetime; ArduPlane/tailsitter.cpp — prior-round findings on static locals and stale pitch initialisation. Important Files Changed
Flowchart%%{init: {'theme': 'neutral'}}%%
flowchart TD
A[TRANSITION_ANGLE_WAIT_FW entered] --> B{tailsitter.enabled?\nAnd Auto/Guided mode?}
B -- No --> K[Standard transition code]
B -- Yes --> C{target_bearing_latched?}
C -- No --> D[Latch bearing from prev_WP→next_WP\nReset I-terms\nLog Alignment start]
D --> E[Compute heading error and gyro.x yaw rate]
C -- Yes --> E
E --> F{is_aligned?\nerr 20deg AND rate 8deg/s}
F -- Yes --> G[Start / hold alignment_done_ms timer]
F -- No --> H[Reset alignment_done_ms]
G --> I{wait_complete?\n1s aligned}
H --> I
I -- No --> J{timeout_expired?\n6s total}
J -- No --> L[Altitude hold\npos_control update\ninput yaw to target bearing\nmotors_output\nreturn BLOCK]
L --> A
J -- Yes --> M[Log warning\nProceed anyway]
I -- Yes --> N[Log Alignment Complete]
M --> O[Mark alignment_completed_for_this_flight=true\nReset fw_transition_start_ms\nReset I-terms and nav_pitch_cd]
N --> O
O --> K
K --> P{transition_fw_complete?}
P -- Yes --> Q[TRANSITION_DONE]
P -- No --> R[Ramp nav_pitch_cd down\nusing fw_transition_initial_pitch]
R --> A
%%{init: {'theme': 'base', 'themeVariables': {"darkMode": true, "background": "#0d1117", "primaryColor": "#21262d", "primaryTextColor": "#e6edf3", "primaryBorderColor": "#8b949e", "lineColor": "#8b949e", "textColor": "#e6edf3", "edgeLabelBackground": "#161b22", "actorBkg": "#21262d", "actorBorder": "#8b949e", "actorTextColor": "#e6edf3", "actorLineColor": "#8b949e", "signalColor": "#8b949e", "signalTextColor": "#e6edf3", "noteBkgColor": "#373320", "noteBorderColor": "#d4a72c", "noteTextColor": "#f0e6c0", "labelBoxBkgColor": "#21262d", "labelBoxBorderColor": "#8b949e", "labelTextColor": "#e6edf3", "loopTextColor": "#e6edf3", "activationBkgColor": "#30363d", "activationBorderColor": "#8b949e"}}}%%
flowchart TD
A[TRANSITION_ANGLE_WAIT_FW entered] --> B{tailsitter.enabled?\nAnd Auto/Guided mode?}
B -- No --> K[Standard transition code]
B -- Yes --> C{target_bearing_latched?}
C -- No --> D[Latch bearing from prev_WP→next_WP\nReset I-terms\nLog Alignment start]
D --> E[Compute heading error and gyro.x yaw rate]
C -- Yes --> E
E --> F{is_aligned?\nerr 20deg AND rate 8deg/s}
F -- Yes --> G[Start / hold alignment_done_ms timer]
F -- No --> H[Reset alignment_done_ms]
G --> I{wait_complete?\n1s aligned}
H --> I
I -- No --> J{timeout_expired?\n6s total}
J -- No --> L[Altitude hold\npos_control update\ninput yaw to target bearing\nmotors_output\nreturn BLOCK]
L --> A
J -- Yes --> M[Log warning\nProceed anyway]
I -- Yes --> N[Log Alignment Complete]
M --> O[Mark alignment_completed_for_this_flight=true\nReset fw_transition_start_ms\nReset I-terms and nav_pitch_cd]
N --> O
O --> K
K --> P{transition_fw_complete?}
P -- Yes --> Q[TRANSITION_DONE]
P -- No --> R[Ramp nav_pitch_cd down\nusing fw_transition_initial_pitch]
R --> A
Reviews (5): Last reviewed commit: "Change ahrs.Yaw_sensor to ahrs_view yaw ..." | Re-trigger Greptile |
![greptile-apps[bot]](https://avatars.githubusercontent.com/in/867647?s=60&v=4){kind=small}
|
|
||
| case TRANSITION_ANGLE_WAIT_FW: { | ||
|
|
||
| const uint32_t now_ = AP_HAL::millis(); |
There was a problem hiding this comment.
millis() call and confusing now_ shadow variable
now is already declared at line 973 for the same timestamp. Introducing a second now_ from a separate AP_HAL::millis() call is a minor timing inconsistency and creates a confusing naming shadow throughout the alignment block. Use the existing now variable directly.
| const uint32_t now_ = AP_HAL::millis(); | |
| // Use the 'now' timestamp already captured at the top of update() |
Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!
c8ce286 to
5ea48b4
Compare
| @@ -166,15 +166,24 @@ const AP_Param::GroupInfo AC_AttitudeControl::var_info[] = { | |||
| // @Range: 0 90 | |||
| // @Increment: 0.01 | |||
| // @User: Standard | |||
| AP_GROUPINFO("RELX_ANG", 22, AC_AttitudeControl, _max_tilt_relax, 45.0f), | |||
| AP_GROUPINFO("RELX_HI", 22, AC_AttitudeControl, _high_tilt_relax, 45.0f), | |||
|
|
|||
| // @Param: RELX_EN | |||
| // @DisplayName: Position control relaxation enable | |||
| // @Description: Enable/disable flag for postion controller relaxation | |||
| // @Description: Enable/disable flag for position controller relaxation | |||
| // @Values: 0:Disabled, 1:Enabled | |||
| // @User: Advanced | |||
| AP_GROUPINFO("RELX_EN", 23, AC_AttitudeControl, _att_relax_enabled, 0), | |||
|
|
|||
| // @Param: RELX_TC | |||
| // @DisplayName: Relaxation time constant | |||
| // @Description: Time constant for the low pass filter on the relaxation factor | |||
| // @Units: s | |||
| // @Range: 0.01 10 | |||
| // @Increment: 0.01 | |||
| // @User: Advanced | |||
| AP_GROUPINFO("RELX_TC", 24, AC_AttitudeControl, _tc_tilt_relax, 1.0f), | |||
There was a problem hiding this comment.
RELX_TC was at group index 21 in the previous firmware and stored that way in EEPROM. The new code repurposes index 21 for RELX_LO (low hysteresis threshold, 30°). Any vehicle upgraded from the previous firmware that had Q_A_RELX_TC stored in flash will have its value—nominally 0.4 s—loaded directly into RELX_LO, setting the low-activation threshold to 0.4° instead of 30°. Because RELX_EN is now 1 in defaults.parm, the relaxation will be active for virtually every hover attitude, silently suppressing the pitch setpoint. New RELX_TC must use a previously unused index (e.g., 25) or the existing RELX_TC index must be kept at 21 and a different index chosen for RELX_LO.
4 participants
Implements a pre-transition alignment phase for tailsitter aircraft that ensures proper heading alignment before proceeding with the forward transition. The aircraft will automatically yaw to match the target bearing and hold position for 2 seconds before transitioning, preventing premature transitions that could cause instability.
Key Features
##Reference
https://www.notion.so/airbound/Heading-alignment-to-WP-2-before-transition-2f121adf4be9802bb36acd78519f64de?source=copy_link
Summary by CodeRabbit
Release Notes
New Features
Chores