Discs

Contents

Discs#

A sensor that updates a sensing environment state with the tabular equivalent to the symbolic geometric state computed by Geometric with limited range. It updates fields “discs” with a fixed-size table containing the nearest N neighbors and obstacles in relative coordinates (providing center, radius, velocity, … depending on the sensor configuration).

It is mainly used to train Machine-learning behaviors to be compared with the model-based navigation behaviors using a symbolic geometric representation of neighbors and obstacles.

Example#

The video has been recorded using one of the navground_learning tutorials with the following configuration

steps: 1800
time_step: 0.1
record_poses: true
scenario:
  type: Cross
  agent_margin: 0.1
  side: 4
  target_margin: 0.1
  tolerance: 0.5
  groups:
    -
      type: thymio
      number: 20
      radius: 0.1
      control_period: 0.1
      speed_tolerance: 0.02
      color: red
      kinematics:
        type: 2WDiff
        wheel_axis: 0.094
        max_speed: 0.12
      behavior:
        type: Policy
        policy_path: ...
        max_acceleration: 1
        max_angular_acceleration: 1
        use_acceleration_action: true
        include_target_distance: true
        include_velocity: true
        include_angular_speed: true 
        flat: true
        deterministic: true
      state_estimation:
        type: Discs
        number: 5
        range: 5.0
        max_speed: 0.12
        max_radius: 0.0
Contents