mirror of
https://github.com/sunnypilot/sunnypilot.git
synced 2026-02-19 01:53:57 +08:00
9ccc5cc3ea
* modeld: Retain pre-20hz drive model support * Method not available anymore on OP * some fixes * Revert "Long planner get accel: new function args (#34288)" * Revert "Fix low-speed allow_throttle behavior in long planner (#33894)" * Revert "long planner: allow throttle reflects usage (#33792)" * Revert "Gate acceleration on model gas press predictions (#33643)" * Reapply "Gate acceleration on model gas press predictions (#33643)" This reverts commit76b08e37cb. * Reapply "long planner: allow throttle reflects usage (#33792)" This reverts commitc75244ca4e. * Reapply "Fix low-speed allow_throttle behavior in long planner (#33894)" This reverts commitb2b7d21b7b. * Reapply "Long planner get accel: new function args (#34288)" This reverts commit74dca2fccf. * don't need * retain snpe * wrong * they're symlinks * remove * put back into VCS * add back * don't include built * Refactor model runner retrieval with caching support Added caching for active model runner type via `ModelRunnerTypeCache` to enhance performance and avoid redundant checks. Introduced a `force_check` flag to bypass the cache when necessary. Updated related code to handle cache clearing during onroad transitions. * Update model runner determination logic with caching fix Enhances `get_active_model_runner` to utilize caching more effectively by ensuring type consistency and updating cache only when necessary. Also updates `is_snpe_model` to pass the `started` state to the runner determination function, improving behavior for dynamic checks. * parse inputs via metadata * load model and metadata dynamically * cherry pick from devtekve as base * lateral_control_params & prev_desired_curv: MLSIM V0 to Null Pointer * old desired_curv data: MLSIM V1 to Postal Service * Bringing what was on master back then * Cleaning up * Refactor model pipeline for modularity and dynamic input handling Refactored the model pipeline by introducing helper functions to modularize model loading, metadata extraction, and input preparation. Improved flexibility in handling dynamic input keys and parsing outputs based on model configuration. Removed deprecated or unused code segments for cleaner and more maintainable structure. * Push NDv2 because why not and fix modeld * `Refactor model parsing and clean unused code dependencies` Simplified `parse_outputs` by removing unnecessary `input_keys` parameter, ensuring cleaner logic. Updated `PROCESS_NAME` for standardization and eliminated deprecated `Pathlib` dependencies in model paths. Minor adjustments improve input handling for lateral control parameters. * Refactor model and metadata loading functions. Simplified and clarified `load_model` by renaming it to `get_model_path` and removing redundant variable assignments. Streamlined `load_metadata` by directly returning the loaded metadata without intermediate variables. These changes improve code readability and maintainability. * Refactor modeld process selection based on SNPE support. Introduce conditional logic to determine and start the appropriate modeld process (SNPE or default) based on hardware support. This improves flexibility and ensures correct process management. * Walrus baby * Update longitudinal_planner.py * Improve model download progress handling and translations Refactored model download status handling logic for better clarity and added mechanisms to track status changes efficiently. Updated UI text and translations across multiple languages to reflect consistent and accurate model download states. * Revert "Update longitudinal_planner.py" This reverts commitb44a687e4c. * Fix variable naming for curvature size in modeld.py Renamed the variable `len` to `length` to avoid conflict with the built-in `len()` function, improving code clarity and preventing potential errors. Also removed trailing spaces in commented-out sections for better formatting consistency. * sync with upstream * some are different * should work * sim_pose only exists in in ndv3 and prior * dynamic meta constants * fix * uiview: disable power saving * fix this * ain't coasting for y'all * Static analysis * Refactor initialization of model inputs for clarity. Removed redundant pre-initialization of `driving_style`, `nav_features`, and `nav_instructions` variables. Instead, directly initialize these within their respective conditional blocks for better readability and reduced memory usage. * default to none * enable in next PR * more * Revert "more" This reverts commitf5a4220588. * Revert "enable in next PR" This reverts commit621cc4f18e. * no need to cast bool * nuke * fix desired curvature for pre LAv1 models * mypy * static * fix * new json * Going to a test branch with a different model json list * renamed model json * Update model runner handling in custom.capnp and helpers Refactored model runner logic by introducing a `runner` field in `custom.capnp` and simplifying `get_active_model_runner` logic. Removed deprecated function `get_model_runner_by_filename` and added a temporary filter in `fetcher.py` to enforce `snpe` until full tinygrad support is implemented. * Revert "Update model runner handling in custom.capnp and helpers" This reverts commitf34d872c13. * Revert "renamed model json" This reverts commit15c6ed303b. * Revert "Going to a test branch with a different model json list" This reverts commit4c1408fee5. * Reapply "renamed model json" This reverts commitc6fec6912a. Reapply "Going to a test branch with a different model json list" This reverts commit83e253e9a3. * Add 'runner' property to 'ModelBundle' and update relevant functions The 'ModelBundle' class in 'custom.capnp' has been extended to include a 'runner' property. This required updating 'fetcher.py' to handle the 'runner' property when parsing model bundles. Additionally, the helper function 'get_model_runner_by_filename' has been removed from 'helpers.py' as it is no longer needed because the 'runner' property provides this information. The 'get_active_model_runner' function has also been updated in light of these changes. * Refine bundle selection logic in SoftwarePanelSP Improve logic for determining which model bundle to display by considering download status and failure state. Remove unnecessary function call to enhance clarity and maintainability. * tmp * Add retrieval of active model bundle in manager loop Introduce a call to `get_active_bundle` to fetch the active model bundle and store it in `self.active_bundle`. This ensures the active bundle is always up-to-date during the model management process. * Add "Use Default" option for model selection Introduced a "Use Default" option in the model selection dropdown, allowing users to reset to the default model. Adjusted logic to handle default selection and ensure proper parameter updates. Fixed bundle status handling during the download process in the model manager. * Refactor type hint in active_bundle assignment. Removed an unnecessary type hint in the active_bundle assignment for cleaner and more maintainable code. This change does not affect functionality but improves code readability. * no nested * update json url * split out * format * condense * more --------- Co-authored-by: Jason Wen <haibin.wen3@gmail.com> Co-authored-by: Kumar <36933347+rav4kumar@users.noreply.github.com>
openpilot
openpilot is an operating system for robotics.
Currently, it upgrades the driver assistance system in 275+ supported cars.
Docs · Roadmap · Contribute · Community · Try it on a comma 3X
Quick start: bash <(curl -fsSL openpilot.comma.ai)
Using openpilot in a car
To use openpilot in a car, you need four things:
- Supported Device: a comma 3/3X, available at comma.ai/shop.
- Software: The setup procedure for the comma 3/3X allows users to enter a URL for custom software. Use the URL
openpilot.comma.aito install the release version. - Supported Car: Ensure that you have one of the 275+ supported cars.
- Car Harness: You will also need a car harness to connect your comma 3/3X to your car.
We have detailed instructions for how to install the harness and device in a car. Note that it's possible to run openpilot on other hardware, although it's not plug-and-play.
Branches
| branch | URL | description |
|---|---|---|
release3 |
openpilot.comma.ai | This is openpilot's release branch. |
release3-staging |
openpilot-test.comma.ai | This is the staging branch for releases. Use it to get new releases slightly early. |
nightly |
openpilot-nightly.comma.ai | This is the bleeding edge development branch. Do not expect this to be stable. |
nightly-dev |
installer.comma.ai/commaai/nightly-dev | Same as nightly, but includes experimental development features for some cars. |
To start developing openpilot
openpilot is developed by comma and by users like you. We welcome both pull requests and issues on GitHub.
- Join the community Discord
- Check out the contributing docs
- Check out the openpilot tools
- Read about the development workflow
- Code documentation lives at https://docs.comma.ai
- Information about running openpilot lives on the community wiki
Want to get paid to work on openpilot? comma is hiring and offers lots of bounties for external contributors.
Safety and Testing
- openpilot observes ISO26262 guidelines, see SAFETY.md for more details.
- openpilot has software-in-the-loop tests that run on every commit.
- The code enforcing the safety model lives in panda and is written in C, see code rigor for more details.
- panda has software-in-the-loop safety tests.
- Internally, we have a hardware-in-the-loop Jenkins test suite that builds and unit tests the various processes.
- panda has additional hardware-in-the-loop tests.
- We run the latest openpilot in a testing closet containing 10 comma devices continuously replaying routes.
Licensing
openpilot is released under the MIT license. Some parts of the software are released under other licenses as specified.
Any user of this software shall indemnify and hold harmless Comma.ai, Inc. and its directors, officers, employees, agents, stockholders, affiliates, subcontractors and customers from and against all allegations, claims, actions, suits, demands, damages, liabilities, obligations, losses, settlements, judgments, costs and expenses (including without limitation attorneys’ fees and costs) which arise out of, relate to or result from any use of this software by user.
THIS IS ALPHA QUALITY SOFTWARE FOR RESEARCH PURPOSES ONLY. THIS IS NOT A PRODUCT. YOU ARE RESPONSIBLE FOR COMPLYING WITH LOCAL LAWS AND REGULATIONS. NO WARRANTY EXPRESSED OR IMPLIED.
User Data and comma Account
By default, openpilot uploads the driving data to our servers. You can also access your data through comma connect. We use your data to train better models and improve openpilot for everyone.
openpilot is open source software: the user is free to disable data collection if they wish to do so.
openpilot logs the road-facing cameras, CAN, GPS, IMU, magnetometer, thermal sensors, crashes, and operating system logs. The driver-facing camera is only logged if you explicitly opt-in in settings. The microphone is not recorded.
By using openpilot, you agree to our Privacy Policy. You understand that use of this software or its related services will generate certain types of user data, which may be logged and stored at the sole discretion of comma. By accepting this agreement, you grant an irrevocable, perpetual, worldwide right to comma for the use of this data.