Skip to content

Latest commit

 

History

History
129 lines (99 loc) · 4.94 KB

architecture.md

File metadata and controls

129 lines (99 loc) · 4.94 KB

Pronto Architecture

The system offers a set of modules that manage different sensors subscribing to specific topics. Available modules:

  1. ins: inertial navigation system (IMU)
  2. legodo: legged odometry
  3. bias_lock: bias updater while the system is standing
  4. scan_matcher: LiDAR scan matcher
  5. qualysis_mt: Qualisys motion tracker

ROS2 Code UML

Prova

Pronto Estimator Parameters

Estimator parameters

  • pose_topic: estimated pose topic name
  • pose_frame_id: pose tf frame name
  • twist_topic: estimated twist topic name
  • publish_pose: if true, publish pose
  • publish_tf: if true, publish tf
  • tf_child_frame_id
  • republish_sensors
  • init_sensors: sensor list used to initialize the filter
  • active_sensors: sensor list used to update the estimation
  • utime_history_span: maximum update temporal span
  • sigma0: initial covariance
  • x0: initial state

Base-module parameters

  • topic: module subscription topic name
  • roll_forward_on_receive: if true, update the estimation when a message is received
  • publish_head_on_message: if true, publish the estimation when a message is received

INS parameters

  • q_gyro: gyro measurement covariance
  • q_accel: accelerometer measurement covariance
  • q_gyro_bias: gyro bias covariance
  • q_accel_bias: accelerometer bias covariance
  • num_to_init: number of messages needed to initialize the INS module
  • gyro/accel_bias_initial: initial bias value
  • gyro/accel_bias_recalc_at_start: if true, recalculate bias during initialization
  • gyro/accel_bias_update_online: if true, update bias online
  • frame: IMU tf frame name
  • base_frame: base tf frame name

Legodo parameters

  • legodo_mode: covariance mode (STATIC_SIGMA, VAR_SIGMA, IMPACT_SIGMA, WEIGHTED_AVG, ALPHA_FILTER)
  • stance_mode: contact-detection mode (THRESHOLD, HYSTERESIS, REGRESSION)
  • stance_threshold
  • stance_hysteresis_low
  • stance_hysteresis_high
  • stance_hysteresis_delay_low
  • stance_hysteresis_delay_high
  • stance_alpha
  • stance_regression_beta_size
  • stance_regression_beta
  • r_vx: initial covariance in x direction
  • r_vy: initial covariance in y direction
  • r_vz: initial covariance in z direction
  • sim: if true, use sensor_msgs; otherwise use pi3hat_msgs

Bias-lock parameters

  • torque_threshold: minimum knee torque to detect ground contact
  • velocity_threshold: maximum velocity to consider the robot standing
  • secondary_topic: joint-state topic name
  • sim: if true, use sensor_msgs; otherwise use pi3hat_msgs

Scan-matcher parameters

  • mode: correction mode (position, yaw, position_yaw)
  • r_yaw: yaw covariance
  • r_pxy: position covariance

Wheel-odometry parameters

  • mode: correction mode (linear_velocity, angular_velocity, both)
  • r_linear: linear-velocity covariance
  • r_chi: angular covariance

Qualysis-MT parameters

  • robot_name: Qualisys rigid-body name
  • r_xyz: position covariance
  • r_chi: orientation covariance
  • mode: correction mode (position, yaw, position_yaw, orientation, position_orientation)

Classes and Methods

  1. ROS_FrontEnd (Pronto ROS)

    • Constructor: reads topic names for pose, velocity and tf from ROS parameters; initialises filter state and covariance.
    • initializeState: reads initial pose and velocity from parameters and sets the initial state.
    • initializeCovariance: reads the initial covariance matrix and sets it.
    • initializeFilter: initializes the filter once all required sensor modules are ready.
    • areModulesInitialized: checks whether all sensors that need initialisation are ready.
    • addSensingModule: creates data structures (maps and topics) for a sensor.
    • InitCallback: callback to initialize a sensing module; destroys the init topic when done.
    • callback: subscriber callback that updates the estimation from incoming data.

  2. InsHandlerRos (Pronto ROS) — manages the InsModule for the frontend.

    • Constructor: builds the InsModule and creates its topic.
    • ProcessMessage: receives an IMU message and returns the state and covariance update.
    • processMessageInit: processes an IMU message and returns whether initialization succeeded.

  3. SensingModule (Pronto_Core) — virtual base class for a generic sensor.

    • ProcessMessage (virtual): processes sensor data and returns the update.
    • processMessageInit (virtual): handles initialization messages.

  4. DualSensingModule (Pronto_Core) — virtual class for a module with two sensor inputs.

    • ProcessMessage (virtual)
    • processMessageInit (virtual)

  5. InsModule (Pronto_Core)

    • Constructor: builds the inertial sensing module from the IMU configuration and IMU-to-body transformation.
    • ProcessMessage
    • processMessageInit