Configuration and CLI Workflows¶

Scenario YAML files configure the entire pipeline. Each scenario contains five core sections whose parameters are consumed by strongly typed dataclasses in src/common/config.py.

map – points to a PNG (white = driveable, black = occupied), specifies resolution (meters_per_cell), and defines the world origin of the bitmap.
start / goal – three-element poses [x, y, yaw] expressed in metres and radians. For the bundled map_raw.png (80 m × 40 m) the conservative scenario starts at (5, 5) and targets (35, 32) while the high-speed scenario finishes near (72, 32).
planner_config – sampling radius, neighbour count, smoothing options, and safety margins for FMT.
tracker_config – MPC horizon length, timestep, weighting matrices, and cruise-speed targets.
visualization – toggles for planner debug windows, animation outputs, file locations, and vehicle drawing params.

map:
  png_path: maps/map_raw.png
  meters_per_cell: 0.2
  origin: [0.0, 0.0]
start: [5.0, 5.0, 0.0]
goal: [35.0, 32.0, 0.0]
visualization:
  show_planner_debug: true
  animate_planner: true
  animate_simulation: true
  vehicle_draw:
    wheelbase: 2.7
    width: 1.55
    length: 4.5
    front_overhang: 0.9
    rear_overhang: 0.8
    wheel_track: 1.2
    tire_radius: 0.33
    tire_width: 0.22

The meters_per_cell value \(m_{\mathrm{cell}}\) acts as a scale factor when converting between pixel indices \((i, j)\) and metric coordinates \((x, y)\):

\[ x = x_0 + i \cdot m_{\mathrm{cell}}, \qquad y = y_0 + j \cdot m_{\mathrm{cell}}, \]

where \((x_0, y_0)\) denotes the origin specified in the YAML file. This mapping is used symmetrically when rendering simulation results back onto the occupancy grid.

CLI Modes¶

The entry point python -m src.cli <config> <mode> (or the installed console script fmt-cli <config> <mode>) supports four execution modes:

plan_only – run mapping + FMT to inspect the planned corridor and optional debug plot.
track_only – run planning and generate the first MPC command without stepping the simulator.
run_e2e – complete pipeline from planning through simulation and visualisation.
export_map – inflate and export the occupancy grid referenced by the scenario.

Enable or disable plots and animations in the visualization block without modifying code. Output directories default to artifacts/<scenario_name>/ and are created automatically.