Perception Packages¶

The perception module is responsible for state estimation and environment understanding. It processes sensor data to provide the robot's best estimate of its pose, velocity, and surrounding environment.

Overview¶

Perception forms the foundation of the autonomy stack by:

State Estimation: Fusing sensor data (IMU, cameras, GPS) to estimate robot position, orientation, and velocity
Sensor Processing: Converting raw sensor data into usable formats for downstream modules
Environment Understanding: Detecting and tracking objects, obstacles, and features in the environment

Launch¶

Module launch files are located under robot/ros_ws/src/perception/perception_bringup/launch/. The stack entry files include them directly:

ros2 launch perception_bringup stereo_image_proc.launch.xml
ros2 launch perception_bringup topic_keepalive.launch.xml

Key Topics¶

Outputs¶

/{robot_name}/odometry_conversion/odometry (nav_msgs/Odometry) - Best estimate of robot state (position, orientation, velocities). This is the v1 canonical state topic — see Interface Conventions §2; plain /{robot_name}/odometry is the intended v2 name.

Inputs¶

Raw sensor data from sensors layer (cameras, IMU, GPS, depth sensors)

Modules¶

State estimation and related perception packages live under robot/ros_ws/src/perception/. Only external motion capture is documented below today; other approaches (onboard sensor fusion, visual-inertial odometry, etc.) will be added here in future releases.

External pose (motion capture)¶

OptiTrack (asm_optitrack module) — NatNet mocap support (rigid-body poses from a Motive PC, PX4 external-vision fusion bridges), provided by the asm_optitrack module. Add it with airstack module add https://github.com/castacks/asm_optitrack --version <tag>; see AirStack Modules.

Configuration¶

Perception parameters live in each module package's own config/ YAML files, overridden per-stack via launch arguments in the stack entry file (stacks/<name>/launch/*.launch.xml). Common parameters include:

Sensor topics to subscribe to
Fusion algorithm parameters
Output frame IDs
Publishing rates