HKG tuning #66
Open
HKG tuning #66
+654
−19
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This reverts commit bf456ba.
Reviewer's Guide by SourceryThis pull request introduces custom longitudinal tuning for HKG vehicles. It includes new classes for managing tuning parameters and applying custom acceleration logic. The changes also modify CAN message creation to incorporate jerk limits and comfort band values, and adjust default longitudinal parameters when HKG tuning is enabled. Sequence diagram for calculating limited acceleration with HKG tuningsequenceDiagram
participant CarController
participant HKGLongitudinalController
participant HKGLongitudinalTuning
participant CarState
participant Actuators
CarController->HKGLongitudinalController: calculate_limited_accel(accel, actuators, CS)
alt HKGtuning is enabled
HKGLongitudinalController->HKGLongitudinalTuning: calculate_limited_accel(accel, actuators, CS)
HKGLongitudinalTuning->HKGLongitudinalTuning: make_jerk(CS, accel, actuators)
HKGLongitudinalTuning->CarState: Access vEgo, cruiseState.speed
HKGLongitudinalTuning->Actuators: Access accel
HKGLongitudinalTuning->HKGLongitudinalTuning: Compute accel_delta, ramp_rate, brake_aggressiveness
HKGLongitudinalTuning->HKGLongitudinalTuning: Apply smoothing
HKGLongitudinalTuning->HKGLongitudinalTuning: Enforce jerk limits
HKGLongitudinalTuning->HKGLongitudinalTuning: Gradual transition from positive accel to negative
HKGLongitudinalTuning->HKGLongitudinalTuning: Update accel
HKGLongitudinalTuning-->CarController: return accel
else HKGtuning is disabled
HKGLongitudinalController-->CarController: return accel
end
Updated class diagram for HKGLongitudinalController and HKGLongitudinalTuningclassDiagram
class HKGLongitudinalController {
-CP
-tuning
+__init__(CP)
+apply_tune(CP)
+calculate_limited_accel(accel, actuators, CS)
+get_jerk()
}
class HKGLongitudinalTuning {
-CP
-mpc
-long_control
-DT_CTRL
-params
-hkg_tuning
-has_radar
-last_accel
-brake_ramp
-accel_last
-using_e2e
-accel_raw
-accel_last_jerk
-jerk
-jerk_count
-jerk_upper_limit
-jerk_lower_limit
-cb_upper
-cb_lower
+__init__(CP)
-_setup_controllers()
-_init_state()
+make_jerk(CS, accel, actuators)
+calculate_limited_accel(accel, actuators, CS)
+apply_tune(CP)
+get_jerk()
}
class JerkOutput {
+jerk_upper_limit
+jerk_lower_limit
+cb_upper
+cb_lower
+__init__(jerk_upper_limit, jerk_lower_limit, cb_upper, cb_lower)
}
HKGLongitudinalController -- HKGLongitudinalTuning : uses
HKGLongitudinalTuning -- JerkOutput : returns
File-Level Changes
Tips and commandsInteracting with Sourcery
Customizing Your ExperienceAccess your dashboard to:
Getting Help
|
![sourcery-ai[bot]](https://avatars.githubusercontent.com/in/48477?s=60&v=4){kind=small}
There was a problem hiding this comment.
Hey @Discountchubbs - I've reviewed your changes - here's some feedback:
Overall Comments:
- Consider adding a more descriptive name than
HKGLongitudinalControllerto better reflect its purpose. - It might be cleaner to encapsulate the logic that determines whether to use the custom acceleration logic within the
HKGLongitudinalTuningclass.
Here's what I looked at during the review
- 🟢 General issues: all looks good
- 🟢 Security: all looks good
- 🟢 Testing: all looks good
- 🟡 Complexity: 1 issue found
- 🟢 Documentation: all looks good
Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.
| self.cb_upper = self.cb_lower = 0.0 | ||
|
|
||
|
|
||
| def make_jerk(self, CS, accel, actuators): |
There was a problem hiding this comment.
issue (complexity): Consider refactoring complex tuning logic into smaller, testable helper methods.
Consider refactoring parts of the tuning logic into small helper methods. For example, the nested conditionals in make_jerk could be separated into a helper that computes the jerk limits and comfort bands. This makes the code easier to read and test. Some suggestions:
- Extract the CANFD/non-CANFD logic into its own method:
def _compute_jerk_limits(self, v_error, current_jerk, accel):
jerk_max = 5.0
if self.CP.flags & HyundaiFlags.CANFD.value:
if v_error < 3.0:
jerk_reduction = interp(v_error, [0.0, 1.0, 3.0], [0.3, 0.5, 1.0])
jerk_max *= jerk_reduction
jerk_upper = min(max(0.5, current_jerk * 2.0), jerk_max)
jerk_lower = min(max(1.0, -current_jerk * 4.0), jerk_max)
cb_upper = cb_lower = 0.0
else:
jerk_upper = min(max(0.5, current_jerk * 2.0), jerk_max)
jerk_lower = min(max(1.0, -current_jerk * 2.0), jerk_max)
error_factor = interp(v_error, [0.0, 0.5, 1.0, 5.0], [0.0, 0.1, 0.5, 1.0])
accel_factor = interp(abs(accel), [0.0, 1.0], [0.2, 0.1])
if accel >= 0:
cb_upper = clip(0.8 * error_factor + accel * accel_factor, 0, 1.0)
cb_lower = clip(0.6 * error_factor + accel * accel_factor, 0, 0.8)
else:
cb_upper = clip(1.0 * error_factor + accel * accel_factor, 0, 1.2)
cb_lower = clip(0.8 * error_factor + accel * accel_factor, 0, 1.2)
return jerk_upper, jerk_lower, cb_upper, cb_lower- Use this helper inside
make_jerkto simplify that method:
def make_jerk(self, CS, accel, actuators):
state = getattr(actuators, "longControlState", LongCtrlState.pid)
if not CS.out.cruiseState.enabled:
self.jerk_upper_limit = 0.0
self.jerk_lower_limit = 0.0
self.cb_upper = self.cb_lower = 0.0
self.accel_last_jerk = 0.0
return 0.0
if state == LongCtrlState.stopping:
self.jerk = 1.75 / 2 - CS.out.aEgo
return self.jerk
else:
current_accel = clip(actuators.accel, CarControllerParams.ACCEL_MIN, CarControllerParams.ACCEL_MAX)
self.jerk = (current_accel - self.accel_last_jerk) / self.DT_CTRL
self.accel_last_jerk = current_accel
v_error = abs(CS.out.vEgo - CS.out.cruiseState.speed)
self.jerk_upper_limit, self.jerk_lower_limit, self.cb_upper, self.cb_lower = \
self._compute_jerk_limits(v_error, self.jerk, accel)
return self.jerk- Consider similar extraction in
calculate_limited_accelfor the brake ramp and smoothing logic.
These changes keep all functionality intact while reducing complexity and improving testability.
18Feb2025
Discountchubbs
force-pushed
the
HKG-Tuning
branch
from
4eaeb32 to
d08a0ff
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
sunnypilot/sunnypilot#636
New Feature
Custom tuning for HKG vehicles, taking into account lower, upper jerk, and comfort bands. When the user selects toggles, these plus interface tuning will be applied. Furthermore, if the user presses the button to enable smoother braking, my custom accel logic will be utilized to greatly filter and smoothen accel/braking transitions for a more chill open pilot experience.
Current issue:
--While CB calculation works great to limit rough transitions from acceleration costs. There is an error on cruiseState.Speed target that is still being ironed out. So, at this time, until that issue is completely fixed, Hyundai has the standard value of 0.0 for CB upper and lower.
Summary by Sourcery
Add custom longitudinal tuning for HKG vehicles.
New Features: