Skip to content

feat: allow conditionally launch diffusion planner#1752

Open
Owen-Liuyuxuan wants to merge 19 commits intoautowarefoundation:mainfrom
tier4:feat/conditional_e2e_launch
Open

feat: allow conditionally launch diffusion planner#1752
Owen-Liuyuxuan wants to merge 19 commits intoautowarefoundation:mainfrom
tier4:feat/conditional_e2e_launch

Conversation

@Owen-Liuyuxuan
Copy link
Copy Markdown
Contributor

@Owen-Liuyuxuan Owen-Liuyuxuan commented Feb 13, 2026

Description

  1. Conditionally launch diffusion planner instead of rule-based planner.
  2. Switch MRM settings for e2e launch.
  3. Add diffusion planner related rviz settings.
  4. New e2e_components placement for future flexible integration and development.

Tested with old model using the current settings.
image

How was this PR tested?

PSim

Notes for reviewers

None.

Effects on system behavior

None.

Signed-off-by: YuxuanLiuTier4Desktop <619684051@qq.com>
Signed-off-by: YuxuanLiuTier4Desktop <619684051@qq.com>
@github-actions github-actions bot added component:system System design and integration. (auto-assigned) component:planning Route planning, decision-making, and navigation. (auto-assigned) component:ui User interface, user experience, graphical user interfaces. (auto-assigned) labels Feb 13, 2026
@github-actions
Copy link
Copy Markdown

github-actions bot commented Feb 13, 2026

Thank you for contributing to the Autoware project!

🚧 If your pull request is in progress, switch it to draft mode.

Please ensure:

Copy link
Copy Markdown
Contributor

@shmpwk shmpwk left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

/**:
ros__parameters:
plugins_path: $(find-pkg-share autoware_tensorrt_plugins)/plugins/libautoware_tensorrt_plugins.so
onnx_model_path: $(env HOME)/autoware_data/diffusion_planner/v3.0/diffusion_planner.onnx
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

memo: This comes from autowarefoundation/autoware#6710

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@shmpwk
Copy link
Copy Markdown
Contributor

shmpwk commented Feb 13, 2026

Signed-off-by: YuxuanLiuTier4Desktop <619684051@qq.com>
@Owen-Liuyuxuan
Copy link
Copy Markdown
Contributor Author

Owen-Liuyuxuan commented Mar 27, 2026

@shmpwk @SakodaShintaro @danielsanchezaran cc: @yukkysaito

  • I updated the launch that almost completely following our projects. With DP->modifier (only stop fix) -> Optimizer -> output
  • Please remember to update autoware_launch after we pushed to autoware_universe

@Owen-Liuyuxuan
Copy link
Copy Markdown
Contributor Author

@Kotakku cc: @SakodaShintaro
Following your updates:

autowarefoundation/autoware_universe#12387
https://github.com/tier4/autoware_launch.x2/pull/2073

I updated parameter files and the rviz

Comment thread autoware_launch/launch/e2e_components/diffusion_planner.launch.xml Outdated
Comment thread autoware_launch/launch/e2e_components/diffusion_planner.launch.xml Outdated
Comment on lines +90 to +96
<!-- planning evaluator -->
<group>
<include file="$(find-pkg-share autoware_planning_evaluator)/launch/planning_evaluator.launch.xml">
<arg name="input/objects" value="/planning/trajectory_generator/neural_network_based_planner/diffusion_planner_node/output/predicted_objects"/>
<arg name="input/reference_trajectory" value="/planning/trajectory_generator/neural_network_based_planner/diffusion_planner_node/output/trajectory"/>
</include>
</group>
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Shouldn’t we evaluate against /planning/trajectory? At this point, I think it would be fine to remove it until the validation is complete.

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@yukkysaito Thank you. This is the implementation from planning evaluation members, I will transfer the question.

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@yukkysaito
The planning evaluator is mainly evaluating /planning/trajectory and this is a different input topic

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sorry,
The input/reference_trajectory means which topic the /planning/trajectory compare to

In the rule-based planner, it was (here)

  <arg name="input/trajectory" default="/planning/trajectory"/>
  <arg name="input/reference_trajectory" default="/planning/scenario_planning/lane_driving/motion_planning/path_optimizer/trajectory"/>

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@yukkysaito
We need this evaluator to calculate normal planning metrics like TTC, PET, DRAC, object distance, and kinematics, etc., for data collection and Evaluator tests.

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@xtk8532704 Thanks for the clarification. In that case, would it make sense to change this to /planning/trajectory instead? 👍

Comment thread autoware_launch/launch/e2e_components/diffusion_planner.launch.xml Outdated
* feat: move dp as parallel to scenario planner

Signed-off-by: YuxuanLiuTier4Desktop <619684051@qq.com>

* style(pre-commit): autofix

* feat: change name to learning_based_planning

Signed-off-by: YuxuanLiuTier4Desktop <619684051@qq.com>

* feat: fix topic connection

Signed-off-by: YuxuanLiuTier4Desktop <619684051@qq.com>

* style(pre-commit): autofix

---------

Signed-off-by: YuxuanLiuTier4Desktop <619684051@qq.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
@@ -0,0 +1,77 @@
<launch>
Copy link
Copy Markdown
Contributor

@yukkysaito yukkysaito Apr 15, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@Owen-Liuyuxuan Is this file necessary? The same file is also being added in this PR.
#1752 (comment)

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@yukkysaito Sorry which files are being repeated?

Image

For planning evaluator, the updates are here

Image

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Copy link
Copy Markdown
Contributor Author

@Owen-Liuyuxuan Owen-Liuyuxuan Apr 15, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@yukkysaito That tier4 PR is merged into this PR branch, which changed from
DP as parallel to planning to DP as parallel to scenario_planning

image

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

component:planning Route planning, decision-making, and navigation. (auto-assigned) component:system System design and integration. (auto-assigned) component:ui User interface, user experience, graphical user interfaces. (auto-assigned) run:build-and-test-differential

Projects

None yet

Development

Successfully merging this pull request may close these issues.

7 participants