feat: allow conditionally launch diffusion planner

autoware_launch/config/planning/neural_net_planner/diffusion_planner.param.yaml

@@ -0,0 +1,29 @@
+/**:
+  ros__parameters:
+    plugins_path: $(find-pkg-share autoware_tensorrt_plugins)/plugins/libautoware_tensorrt_plugins.so
+    onnx_model_path: $(env HOME)/autoware_data/diffusion_planner/v4.0/diffusion_planner.onnx
+    args_path: $(env HOME)/autoware_data/diffusion_planner/v4.0/diffusion_planner.param.json
+    planning_frequency_hz: 10.0
+    ignore_neighbors: false
+    ignore_unknown_neighbors: true
+    traffic_light_group_msg_timeout_seconds: 0.2
+    batch_size: 1
+    temperature: [0.0]
+    velocity_smoothing_window: 8
+    stopping_threshold: 0.3
+    turn_indicator_keep_offset: -1.25
+    turn_indicator_hold_duration: 1.0
+    shift_x: false
+    delay_step: 0
+    line_string_max_step_m: 5.0
+    use_time_interpolation: false
+    planning_factor:
+      enable_stop: true
+      enable_slowdown: false
+      stop_keep_duration_threshold: 1.0  # [s]
+      stop_velocity_threshold: 0.1       # [m/s]
+      slowdown_accel_threshold: -0.3     # [m/s^2]
+    debug_params:
+      publish_debug_route: true
+      publish_debug_map: false
+      publish_debug_linestrings: true

autoware_launch/config/planning/neural_net_planner/trajectory_modifier.param.yaml

@@ -0,0 +1,11 @@
+/**:
+  ros__parameters:
+    use_stop_point_fixer: true
+
+    # Stop Point Fixer Plugin Parameters
+    stop_point_fixer:
+      force_stop_long_stopped_trajectories: true     # Force zero velocity trajectory for trajectories with long stops
+      force_stop_close_stopped_trajectories: true    # Force zero velocity trajectory for trajectories with stops close to ego
+      velocity_threshold_mps: 0.25
+      min_distance_threshold_m: 1.0
+      min_stop_duration_s: 1.0

autoware_launch/config/planning/neural_net_planner/trajectory_optimizer.param.yaml

@@ -0,0 +1,23 @@
+/**:
+  ros__parameters:
+    # Plugin execution order - plugins will run in this order
+    plugin_names:
+      - "autoware::trajectory_optimizer::plugin::TrajectoryPointFixer"
+      - "autoware::trajectory_optimizer::plugin::TrajectoryKinematicFeasibilityEnforcer"
+      - "autoware::trajectory_optimizer::plugin::TrajectoryQPSmoother"
+      - "autoware::trajectory_optimizer::plugin::TrajectoryKinematicFeasibilityEnforcer"
+      - "autoware::trajectory_optimizer::plugin::TrajectoryVelocityOptimizer"
+      - "autoware::trajectory_optimizer::plugin::TrajectoryEBSmootherOptimizer"
+      - "autoware::trajectory_optimizer::plugin::TrajectorySplineSmoother"
+      - "autoware::trajectory_optimizer::plugin::TrajectoryMPTOptimizer"
+      - "autoware::trajectory_optimizer::plugin::TrajectoryExtender"
+
+    # Plugin activation flags - control runtime enable/disable
+    use_akima_spline_interpolation: true
+    use_eb_smoother: false
+    use_qp_smoother: true
+    use_trajectory_point_fixer: true
+    use_velocity_optimizer: true
+    use_trajectory_extender: false
+    use_kinematic_feasibility_enforcer: true
+    use_mpt_optimizer: false

autoware_launch/config/planning/neural_net_planner/trajectory_optimizer_plugins/trajectory_extender.param.yaml

@@ -0,0 +1,9 @@
+/**:
+  ros__parameters:
+    trajectory_extender:
+      # Trajectory matching thresholds
+      nearest_dist_threshold_m: 1.5  # [m] Distance threshold for trajectory matching
+      nearest_yaw_threshold_deg: 60.0  # [deg] Yaw threshold for trajectory matching
+
+      # Backward extension
+      backward_trajectory_extension_m: 5.0  # [m] Length to extend trajectory backward using ego history

autoware_launch/config/planning/neural_net_planner/trajectory_optimizer_plugins/trajectory_kinematic_feasibility_enforcer.param.yaml

@@ -0,0 +1,8 @@
+/**:
+  ros__parameters:
+    # Plugin-specific parameters
+    trajectory_kinematic_feasibility:
+      # Maximum yaw rate [rad/s]. See kinematic_feasibility_enforcer.md for details
+      max_yaw_rate_rad_s: 0.7
+      # Time discretization
+      time_step_s: 0.1  # Fixed time step for yaw rate calculations [s]

autoware_launch/config/planning/neural_net_planner/trajectory_optimizer_plugins/trajectory_mpt_optimizer.param.yaml

@@ -0,0 +1,91 @@
+/**:
+  ros__parameters:
+    trajectory_mpt_optimizer:
+      # Bounds Generation (custom parameters for trajectory_optimizer)
+      corridor_width_m: 3.5
+      enable_adaptive_width: true
+      curvature_width_factor: 0.5
+      velocity_width_factor: 0.3
+      min_clearance_m: 0.5
+      reset_previous_data_each_iteration: true
+      enable_debug_info: false
+      output_delta_arc_length_m: 1.0
+      output_backward_traj_length_m: 5.0
+      ego_nearest_dist_threshold_m: 3.0
+      ego_nearest_yaw_threshold_deg: 45.0
+      acceleration_moving_average_window: 5  # Moving average window size (larger = smoother)
+
+    # MPT Algorithm Parameters (required by autoware_path_optimizer MPTOptimizer)
+    # NOTE: These are solver parameters passed to the underlying MPT optimizer.
+    # Clearance parameters use "road" terminology from the solver, but in this implementation
+    # they represent clearance from the generated corridor bounds (simple perpendicular offsets),
+    # NOT actual road boundaries or drivable area from maps.
+    mpt:
+      option:
+        steer_limit_constraint: false
+        visualize_sampling_num: 1
+        enable_manual_warm_start: false
+        enable_warm_start: false
+        enable_optimization_validation: false
+
+      common:
+        num_points: 100
+        delta_arc_length: 1.0
+
+      clearance:
+        hard_clearance_from_road: 0.0  # Clearance from corridor bounds (not actual road)
+        soft_clearance_from_road: 0.1  # Clearance from corridor bounds (not actual road)
+
+      weight:
+        soft_collision_free_weight: 1.0
+        lat_error_weight: 1.0
+        yaw_error_weight: 0.0
+        yaw_error_rate_weight: 0.0
+        steer_input_weight: 1.0
+        steer_rate_weight: 1.0
+        terminal_lat_error_weight: 100.0
+        terminal_yaw_error_weight: 100.0
+        goal_lat_error_weight: 1000.0
+        goal_yaw_error_weight: 1000.0
+
+      avoidance:
+        max_bound_fixing_time: 1.0
+        max_longitudinal_margin_for_bound_violation: 1.0
+        max_avoidance_cost: 0.5
+        avoidance_cost_margin: 0.0
+        avoidance_cost_band_length: 5.0
+        avoidance_cost_decrease_rate: 0.05
+        min_drivable_width: 0.2
+
+        weight:
+          lat_error_weight: 0.0
+          yaw_error_weight: 10.0
+          steer_input_weight: 100.0
+
+      collision_free_constraints:
+        # NOTE: Collision avoidance is NOT active in this implementation.
+        # These parameters are part of the MPT solver configuration but unused
+        # because we don't provide obstacle data - only geometric corridor bounds.
+        option:
+          l_inf_norm: true
+          soft_constraint: true
+          hard_constraint: false
+
+        vehicle_circles:
+          method: "fitting_uniform_circle"
+
+          bicycle_model:
+            num_for_calculation: 3
+            front_radius_ratio: 1.0
+            rear_radius_ratio: 1.0
+
+          uniform_circle:
+            num: 3
+            radius_ratio: 1.0
+
+          fitting_uniform_circle:
+            num: 3
+
+      validation:
+        max_lat_error: 5.0
+        max_yaw_error: 1.046

autoware_launch/config/planning/neural_net_planner/trajectory_optimizer_plugins/trajectory_point_fixer.param.yaml

@@ -0,0 +1,9 @@
+/**:
+  ros__parameters:
+    trajectory_point_fixer:
+      # Distance thresholds for point processing
+      remove_close_points: true    # Enable removal of close proximity points
+      resample_close_points: true     # Enable resampling of close proximity points
+      min_dist_to_remove_m: 0.01  # [m] Minimum distance to remove close proximity points
+      min_dist_to_resample_m: 0.05   # [m] Minimum distance to merge close proximity points
+      stop_detection_velocity_threshold_mps: 0.3  # [m/s] Velocity threshold for fallback stop detection in slow clusters

autoware_launch/config/planning/neural_net_planner/trajectory_optimizer_plugins/trajectory_qp_smoother.param.yaml

@@ -0,0 +1,34 @@
+/**:
+  ros__parameters:
+    trajectory_qp_smoother:
+      # QP Smoother Parameters
+      # Path-only optimization: smooths geometric path (x,y positions)
+      # Velocity/acceleration are derived from smoothed positions in post-processing
+
+      # Optimization weights
+      weight_smoothness: 10.0                 # Weight for path curvature minimization
+      weight_fidelity: 1.0                    # Baseline fidelity (used when velocity-based disabled)
+
+      # Time discretization
+      time_step_s: 0.1                        # Fixed time step for velocity/acceleration calculations [s]
+
+      # OSQP solver settings (tuned for real-time performance)
+      osqp_eps_abs: 1.0e-4                    # Absolute tolerance for convergence
+      osqp_eps_rel: 1.0e-4                    # Relative tolerance for convergence
+      osqp_max_iter: 100                      # Maximum solver iterations
+      osqp_verbose: false                     # Print solver output to console
+
+      # Orientation preservation
+      preserve_input_trajectory_orientation: false  # Copy orientations from input trajectory to smoothed output
+      max_distance_for_orientation_m: 5.0          # Max distance for nearest neighbor matching when copying orientations [m]
+
+      # Velocity-based fidelity weighting
+      use_velocity_based_fidelity: true      # Enable velocity-dependent fidelity (low weight at slow speeds)
+      velocity_threshold_mps: 0.3             # Speed at which sigmoid transitions (midpoint) [m/s]
+      sigmoid_sharpness: 50.0                 # Sigmoid steepness (higher = sharper transition)
+      min_fidelity_weight: 0.01                # Minimum fidelity at very low speeds
+      max_fidelity_weight: 1.0                # Maximum fidelity at high speeds
+
+      # Point constraints
+      num_constrained_points_start: 3        # Number of points from start to constrain (preserves initial state)
+      num_constrained_points_end: 0          # Number of points from end to constrain

autoware_launch/config/planning/neural_net_planner/trajectory_optimizer_plugins/trajectory_spline_smoother.param.yaml

@@ -0,0 +1,7 @@
+/**:
+  ros__parameters:
+    trajectory_spline_smoother:
+      # Spline interpolation
+      interpolation_resolution_m: 0.2  # [m] Interpolation resolution for Akima spline
+      max_distance_discrepancy_m: 5.0  # [m] Maximum position deviation allowed for orientation copying
+      preserve_input_trajectory_orientation: false  # Flag to indicate if orientation from original trajectory should be copied

autoware_launch/config/planning/neural_net_planner/trajectory_optimizer_plugins/trajectory_velocity_optimizer.param.yaml

@@ -0,0 +1,35 @@
+/**:
+  ros__parameters:
+    trajectory_velocity_optimizer:
+      # Trajectory matching thresholds
+      nearest_dist_threshold_m: 1.5  # [m] Distance threshold for trajectory matching
+      nearest_yaw_threshold_deg: 60.0  # [deg] Yaw threshold for trajectory matching
+
+      # Engage behavior
+      target_pull_out_speed_mps: 1.0  # [m/s] Target speed during pull-out maneuver
+      target_pull_out_acc_mps2: 1.0   # [m/s²] Target acceleration during pull-out
+      set_engage_speed: false         # Set minimum speed during engage
+
+      # Speed limiting (uses external_velocity_limit topic to set max speed)
+      limit_speed: true               # Enable speed limiting
+
+      # Lateral acceleration limiting
+      max_lateral_accel_mps2: 2.0     # [m/s²] Maximum lateral acceleration
+      limit_lateral_acceleration: false  # Enable lateral acceleration limiting, should be launched with velocity smoother
+      min_limited_speed_mps: 3.0        # [m/s] Minimum speed when applying lateral acceleration limit
+
+      # Velocity smoothing
+      smooth_velocities: false        # Enable jerk-filtered velocity smoothing
+
+      # Continuous jerk smoother settings
+      # Note: Kinematic limits (max_accel, min_decel, max_jerk, min_jerk) are loaded
+      # from the common namespace (limit.max_acc, limit.min_acc, limit.max_jerk, limit.min_jerk)
+      # Note: max_velocity uses max_speed_mps from parent namespace
+      continuous_jerk_smoother:
+        # QP optimization weights
+        jerk_weight: 30.0                     # Weight for jerk minimization (smoothness)
+        over_v_weight: 3000.0               # Weight for velocity limit violation (slack)
+        over_a_weight: 30.0                 # Weight for acceleration limit violation (slack)
+        over_j_weight: 10.0                  # Weight for jerk limit violation (slack)
+        velocity_tracking_weight: 1.0        # Weight for tracking reference velocity from input trajectory
+        accel_tracking_weight: 300.0         # Weight for tracking reference acceleration from input trajectory

autoware_launch/config/system/diagnostics/autoware-main-e2e.yaml

@@ -0,0 +1,15 @@
+files:
+  - { path: $(dirname)/autoware-main.yaml }
+
+edits:
+  - { type: remove, path: /autoware/planning/trajectory_validation/finite }
+  - { type: remove, path: /autoware/planning/trajectory_validation/interval }
+  - { type: remove, path: /autoware/planning/trajectory_validation/curvature }
+  - { type: remove, path: /autoware/planning/trajectory_validation/angle }
+  - { type: remove, path: /autoware/planning/trajectory_validation/lateral_acceleration }
+  - { type: remove, path: /autoware/planning/trajectory_validation/acceleration }
+  - { type: remove, path: /autoware/planning/trajectory_validation/deceleration }
+  - { type: remove, path: /autoware/planning/trajectory_validation/steering }
+  - { type: remove, path: /autoware/planning/trajectory_validation/steering_rate }
+  - { type: remove, path: /autoware/planning/trajectory_validation/velocity_deviation }
+  - { type: remove, path: /autoware/planning/trajectory_validation/trajectory_shift }

autoware_launch/launch/autoware.launch.xml

@@ -8,6 +8,7 @@
   <arg name="data_path" default="$(env HOME)/autoware_data" description="packages data and artifacts directory path"/>
 
   <!-- launch module preset -->
+  <arg name="planning_setting" default="rule_based" description="planning setting diffusion_planner / rule_based"/>
   <arg name="planning_module_preset" default="default" description="planning module preset"/>
   <arg name="control_module_preset" default="default" description="control module preset"/>