Autoware Diffusion Planner

Overview

The Autoware Diffusion Planner is a trajectory generation module for autonomous vehicles, designed to work within the Autoware ecosystem. It leverages the Diffusion Planner model, as described in the paper "Diffusion-Based Planning for Autonomous Driving with Flexible Guidance" by Zheng et al.

This planner generates smooth, feasible, and safe trajectories by considering:

• Dynamic and static obstacles
• Vehicle kinematics
• User-defined constraints
• Lanelet2 map context
• Traffic signals and speed limits

It is implemented as a ROS 2 component node, making it easy to integrate into Autoware-based stacks. The node is aimed at working within the proposed Autoware new planning framework.

---

How to use

Currently, some launch files must be changed to run the planning simulator with autoware_diffusion_planner.

(1) /path/to/src/launcher/autoware_launch

diff --git a/autoware_launch/config/control/trajectory_follower/longitudinal/pid.param.yaml b/autoware_launch/config/control/trajectory_follower/longitudinal/pid.param.yaml
--- a/autoware_launch/config/control/trajectory_follower/longitudinal/pid.param.yaml
+++ b/autoware_launch/config/control/trajectory_follower/longitudinal/pid.param.yaml
@@ -6,7 +6,7 @@
     enable_overshoot_emergency: false
     enable_large_tracking_error_emergency: true
     enable_slope_compensation: true
-    enable_keep_stopped_until_steer_convergence: true
+    enable_keep_stopped_until_steer_convergence: false

     # state transition
     drive_state_stop_dist: 0.5
diff --git a/autoware_launch/config/system/diagnostics/planning.yaml b/autoware_launch/config/system/diagnostics/planning.yaml
--- a/autoware_launch/config/system/diagnostics/planning.yaml
+++ b/autoware_launch/config/system/diagnostics/planning.yaml
@@ -11,19 +11,7 @@ units:
           - { type: link, link: /autoware/planning/trajectory_validation }

   - path: /autoware/planning/trajectory_validation
-    type: and
-    list:
-      - { type: link, link: /autoware/planning/trajectory_validation/finite }
-      - { type: link, link: /autoware/planning/trajectory_validation/interval }
-      - { type: link, link: /autoware/planning/trajectory_validation/curvature }
-      - { type: link, link: /autoware/planning/trajectory_validation/angle }
-      - { type: link, link: /autoware/planning/trajectory_validation/lateral_acceleration }
-      - { type: link, link: /autoware/planning/trajectory_validation/acceleration }
-      - { type: link, link: /autoware/planning/trajectory_validation/deceleration }
-      - { type: link, link: /autoware/planning/trajectory_validation/steering }
-      - { type: link, link: /autoware/planning/trajectory_validation/steering_rate }
-      - { type: link, link: /autoware/planning/trajectory_validation/velocity_deviation }
-      - { type: link, link: /autoware/planning/trajectory_validation/trajectory_shift }
+    type: ok

   - path: /autoware/planning/routing/state
     type: diag

(2) /path/to/autoware/src/universe/autoware_universe

diff --git a/launch/tier4_planning_launch/launch/planning.launch.xml b/launch/tier4_planning_launch/launch/planning.launch.xml
--- a/launch/tier4_planning_launch/launch/planning.launch.xml
+++ b/launch/tier4_planning_launch/launch/planning.launch.xml
@@ -40,12 +40,34 @@
       </include>
     </group>

+    <!-- trajectory generator -->
+    <group>
+      <push-ros-namespace namespace="trajectory_generator"/>
+      <include file="$(find-pkg-share autoware_diffusion_planner)/launch/diffusion_planner.launch.xml">
+        <arg name="input_odometry" value="/localization/kinematic_state"/>
+        <arg name="input_acceleration" value="/localization/acceleration"/>
+        <arg name="input_route" value="/planning/mission_planning/route"/>
+        <arg name="input_traffic_signals" value="/perception/traffic_light_recognition/traffic_signals"/>
+        <arg name="input_tracked_objects" value="/perception/object_recognition/tracking/objects"/>
+        <arg name="input_vector_map" value="/map/vector_map"/>
+        <arg name="input_turn_indicators" value="/vehicle/status/turn_indicators_status"/>
+        <arg name="output_trajectories" value="/planning/generator/diffusion_planner/candidate_trajectories"/>
+        <arg name="output_turn_indicators" value="/planning/turn_indicators_cmd"/>
+      </include>
+      <include file="$(find-pkg-share autoware_trajectory_optimizer)/launch/trajectory_optimizer.launch.xml">
+        <arg name="input_trajectories" value="/planning/generator/diffusion_planner/candidate_trajectories"/>
+        <arg name="output_traj" value="/planning/trajectory"/>
+        <arg name="output_trajectories" value="/planning/generator/trajectory_optimizer/candidate_trajectories"/>
+      </include>
+    </group>
+
     <!-- planning validator -->
     <group>
       <include file="$(find-pkg-share autoware_planning_validator)/launch/planning_validator.launch.xml">
         <arg name="container_type" value="component_container_mt"/>
         <arg name="input_trajectory" value="/planning/scenario_planning/velocity_smoother/trajectory"/>
-        <arg name="output_trajectory" value="/planning/trajectory"/>
+        <arg name="output_trajectory" value="/planning/trajectory/unused"/>
         <arg name="input_objects_topic_name" value="$(var input_objects_topic_name)"/>
         <arg name="input_pointcloud_topic_name" value="$(var input_pointcloud_topic_name)"/>
         <arg name="planning_validator_param_path" value="$(var planning_validator_param_path)"/>
diff --git a/launch/tier4_planning_launch/launch/scenario_planning/lane_driving/behavior_planning/behavior_planning.launch.xml b/launch/tier4_planning_launch/launch/scenario_planning/lane_driving/behavior_planning/behavior_planning.launch.xml
--- a/launch/tier4_planning_launch/launch/scenario_planning/lane_driving/behavior_planning/behavior_planning.launch.xml
+++ b/launch/tier4_planning_launch/launch/scenario_planning/lane_driving/behavior_planning/behavior_planning.launch.xml
@@ -240,7 +240,7 @@
         <remap from="~/input/accel" to="/localization/acceleration"/>
         <remap from="~/input/scenario" to="/planning/scenario_planning/scenario"/>
         <remap from="~/output/path" to="path_with_lane_id"/>
-        <remap from="~/output/turn_indicators_cmd" to="/planning/turn_indicators_cmd"/>
+        <remap from="~/output/turn_indicators_cmd" to="/planning/turn_indicators_cmd/unused"/>
         <remap from="~/output/hazard_lights_cmd" to="/planning/behavior_path_planner/hazard_lights_cmd"/>
         <remap from="~/output/modified_goal" to="/planning/scenario_planning/modified_goal"/>
         <remap from="~/output/stop_reasons" to="/planning/scenario_planning/status/stop_reasons"/>

(3) launch the planning simulator

ros2 launch autoware_launch planning_simulator.launch.xml \
  map_path:=/path/to/your/map \
  vehicle_model:=sample_vehicle \
  sensor_model:=sample_sensor_kit

Note: Make sure the appropriate version weight is set for the path specified in planning/autoware_diffusion_planner/config/diffusion_planner.param.yaml.

$ ls ~/autoware_data/diffusion_planner/v2.0/
diffusion_planner.onnx diffusion_planner.param.json

This can be downloaded from setup-dev-env.sh.

Features

• Diffusion-based trajectory generation for flexible and robust planning

  

• Integration with Lanelet2 maps for lane-level context

  

• Dynamic and static obstacle handling using perception inputs

  

  

• Traffic signal and speed limit awareness

  

• ONNX Runtime inference for fast neural network execution
• ROS 2 publishers for planned trajectories, predicted objects, and debug markers

---

Parameters

Name	Type	Description	Default	Range
plugins_path	string	Path to libautoware_tensorrt_plugins.so file for the diffusion planner	$(find-pkg-share autoware_tensorrt_plugins)/plugins/libautoware_tensorrt_plugins.so	N/A
artifact_dir	string	Path to the Artifact(onnx model, etc.) directory	$(env HOME)/autoware_data/diffusion_planner	N/A
onnx_model_path	string	Path to the ONNX model file for the diffusion planner	$(var artifact_dir)/diffusion_planner.onnx	N/A
args_path	string	Path to model argument/configuration file	$(var artifact_dir)/diffusion_planner.param.json	N/A
planning_frequency_hz	float	Planning frequency in Hz	10.0	>0
ignore_neighbors	boolean	Ignore neighbor agents	False	N/A
ignore_unknown_neighbors	boolean	Ignore neighbor agents with unknown class	True	N/A
predict_neighbor_trajectory	boolean	Predict trajectories for neighbor agents	True	N/A
traffic_light_group_msg_timeout_seconds	float	Timeout for traffic light group messages (seconds)	0.2	>0
build_only	boolean	On build only, do not run the planner	False	N/A
batch_size	integer	Batch size for multi-batch inference	1	≥1
temperature	array	List of temperature for sampling trajectories. Its length must be the same as batch_size. Each value means the coefficient of the standard normal distribution which is used for each initial noise of corresponding batch index. val = N(0,1) * temperature	[0.5]	N/A
velocity_smoothing_window	integer	Window size for velocity smoothing. Set to 1 to disable smoothing.	8	≥1
stopping_threshold	float	Threshold for keeping the stopping state [m/s]	0.0	N/A
debug_params.publish_debug_map	boolean	Publish debug map markers	False	N/A
debug_params.publish_debug_route	boolean	Publish debug route markers	True	N/A

Parameters can be set via YAML (see config/diffusion_planner.param.yaml).

---

Inputs

Topic	Message Type	Description
~/input/odometry	nav_msgs/msg/Odometry	Ego vehicle odometry
~/input/acceleration	geometry_msgs/msg/AccelWithCovarianceStamped	Ego acceleration
~/input/tracked_objects	autoware_perception_msgs/msg/TrackedObjects	Detected dynamic objects
~/input/traffic_signals	autoware_perception_msgs/msg/TrafficLightGroupArray	Traffic light states
~/input/vector_map	autoware_map_msgs/msg/LaneletMapBin	Lanelet2 map
~/input/route	autoware_planning_msgs/msg/LaneletRoute	Route information
~/input/turn_indicators	autoware_vehicle_msgs/msg/TurnIndicatorsReport	Turn indicator information

Outputs

Topic	Message Type	Description
~/output/trajectory	autoware_planning_msgs/msg/Trajectory	Planned trajectory for the ego vehicle
~/output/trajectories	autoware_internal_planning_msgs/msg/CandidateTrajectories	Multiple candidate trajectories
~/output/predicted_objects	autoware_perception_msgs/msg/PredictedObjects	Predicted future states of dynamic objects
~/output/turn_indicators	autoware_vehicle_msgs/msg/TurnIndicatorsCommand	Planned turn indicator command
~/debug/lane_marker	visualization_msgs/msg/MarkerArray	Lane debug markers
~/debug/route_marker	visualization_msgs/msg/MarkerArray	Route debug markers

---

Testing

Unit tests are provided and can be run with:

colcon test --packages-select autoware_diffusion_planner
colcon test-result --all

---

ONNX Model and Versioning

The Diffusion Planner relies on an ONNX model for inference. To ensure compatibility between models and the ROS 2 node implementation, the model versioning scheme follows major and minor numbers: The model version is defined either by the directory name provided to the node or within the diffusion_planner.param.json configuration file.

• Major version Incremented when there are changes in the model inputs/outputs or architecture.

  Models with different major versions are not compatible with the current ROS node.

• Minor version Incremented when only the weight files are updated. As long as the major version matches, the node remains compatible, and the new model can be used directly.

To download the latest model, simply run the provided setup script: How to set up a development environment

Model Version History

Version	Release Date	Notes	ROS Node Compatibility
0.1	2025/07/05	- First public release - Route planning based on TIER IV real data	NG
1.0	2025/09/12	- Route Termination learning - Output turn-signal (indicator) - Lane type integration in HD map for improved accuracy - Added datasets:   - Synthetic Data: 4.0M points   - Real Data: 1.5M points	NG
2.0	2025/11/26	- Increased the number of acceptable lane types ("crosswalk", "pedestrian_lane" and "walkway") for left and right boundaries. - Added Polygon and LineString as acceptable input types. - Increased the maximum length of each history record to 3 seconds. - Added support for turn_indicator as an input (this is just an interface, not used in v2.0 weights). - Increased NUM_SEGMENTS_IN_LANE from 70 to 140.	OK

---

Development & Contribution

• Follow the Autoware coding guidelines.
• Contributions, bug reports, and feature requests are welcome via GitHub issues and pull requests.

---

References

• Diffusion Planner (original repo)
• Diffusion planner (our fork of the previous repo, used to train the model)
• "Diffusion-Based Planning for Autonomous Driving with Flexible Guidance"

---

License

This package is released under the Apache 2.0 License.