How rjtdeo/warp-av works
Warp AV is a CARLA-based autonomous cargo-vehicle stack, with a single Python process running an approximately 10 Hz autonomy loop and serving an operator API/console. The main workflow is operator mission request → sensor acquisition and pose estimation → perception and world state → route/behavior decisions → safety supervision → vehicle control. Major documented branches include scenario execution and a separate reinforcement-learning parking environment. Relationships are limited to the README’s documented workflow and relationships visible in the supplied SOURCE INDEX; several subsystem internals were not sampled.
Main entry points: main.py Autonomy loop and API · camera_lidar_perception.py Perception stack · runner.py Scenario runner · carla_sensor_adapter.py Sensor adapter
Components 24
Autonomy runtime
- Autonomy loop and API
src/warp_av/main.py - Mission management
src/warp_av/mission/mission_manager.py - Operator console
src/warp_av/console
Sensing and state
- Sensor adapter
src/warp_av/adapters/carla_sensor_adapter.py - Pose source
src/warp_av/localization/pose_source.py - Localization
src/warp_av/localization/localization.py - Perception stack
src/warp_av/perception/camera_lidar_perception.py - Detection scheduler
src/warp_av/perception/detection_worker.py - Object tracking
src/warp_av/perception/tracking.py - Occupancy grid
src/warp_av/perception/occupancy.py - World model
src/warp_av/world_model.py
Planning and control
- Route and parking planner
src/warp_av/planning/planner.py - Behavior decisions
src/warp_av/behavior/behavior.py - Safety supervisor
src/warp_av/safety/safety_supervisor.py - Vehicle controller
src/warp_av/control/controller.py - Vehicle interface
src/warp_av/vehicle_interface.py
Scenarios and validation
- Scenario runner
src/warp_av/scenario_engine/runner.py - Scenario catalog
src/warp_av/scenario_engine/catalog.py - Scenario evaluation
src/warp_av/scenario_engine/evaluator.py
Parking learning
- Parking practice environment
rl/parking_env.py - Parking math and scoring
rl/parking_math.py - Parking training
rl/train_parking.py
Other components
- Operator
- CARLA simulator
Connections 23
- Operator → Autonomy loop and API: requests mission
README.md - Autonomy loop and API → Operator console: serves console
README.md - Autonomy loop and API → Mission management: manages mission
README.md - CARLA simulator → Sensor adapter: provides sensor data
README.md - Autonomy loop and API → Sensor adapter: reads sensors
README.md - Autonomy loop and API → Localization: estimates pose
src/warp_av/main.py - Sensor adapter → Pose source: provides capture pose
src/warp_av/adapters/carla_sensor_adapter.py - Localization → Pose source: uses pose source
src/warp_av/localization/localization.py - Autonomy loop and API → Perception stack: runs perception
src/warp_av/main.py - Perception stack → Detection scheduler: schedules detections
src/warp_av/perception/camera_lidar_perception.py - Perception stack → Object tracking: tracks objects
src/warp_av/perception/camera_lidar_perception.py - Perception stack → Occupancy grid: updates occupancy
src/warp_av/perception/camera_lidar_perception.py - Autonomy loop and API → Route and parking planner: plans route
src/warp_av/main.py - Autonomy loop and API → Behavior decisions: decides behavior
src/warp_av/main.py - Autonomy loop and API → Safety supervisor: checks safety
README.md - Autonomy loop and API → Vehicle controller: computes commands
README.md - Vehicle controller → Vehicle interface: sends commands
src/warp_av/control/controller.py - Vehicle interface → CARLA simulator: actuates vehicle
README.md - Operator → Scenario runner: runs scenarios
README.md - Scenario runner → Scenario catalog: loads scenarios
scenarios/run_scenario.py - Scenario runner → Scenario evaluation: evaluates run
- Parking practice environment → Parking math and scoring: uses parking math
rl/parking_env.py - Parking training → Parking practice environment: trains in