-
Notifications
You must be signed in to change notification settings - Fork 32
Parameters
The following planner parameters correspond to the Estimation and Propagation pipelines' operation, as implemented in the rovio_bsp package (see rovio_bsp/cfg/rovio.info):
-
cam_FoVx: Horizontal Field-of-View opening of the camera sensor [deg], primarily used to evaluate Line-of-Sight visibility of locally tracked landmarks (against possible octomap occlusions) during propagation. -
cam_FoVy: Vertical Field-of-View opening of the camera sensor [deg], primarily used to evaluate Line-of-Sight visibility of locally tracked landmarks (against possible octomap occlusions) during propagation. -
cam_FoVz: Max range of the camera sensor [m], primarily used to evaluate Line-of-Sight visibility of locally tracked landmarks (against far-away viewpoints) during propagation. -
map_featureVisibilityThreshold: Occlusion checking threshold when performing octomap ray-casting. -
bsp_Ts: Forward-simulation step sampling time used during propagation [s].
(The above refer only to the additional parameters compared to the original rovio pipeline.)
The following planner parameters correspond to the volumetric_mapping process, and are loaded by the bsp_planner node in the form of ros parameters (see provided bsp_planner/launch/bsp_planner.launch):
-
tf_frame: The mapping frame. -
treat_unknown_as_occupied: Used for collision checking methods. -
change_detection_enabled: Track map changes. -
resolution: The mapping resolution. -
visualize_min_z: MarkerArray visualization min z-coordinate. -
visualize_max_z: MarkerArray visualization max z-coordinate. -
sensor_max_range: Max range to consider valid measurement insertion. -
probability_hit: Octomap hit probability. -
probability_miss: Octomap miss probability. -
threshold_min: Clamping threshold for min cell probability. -
threshold_max: Clamping threshold for max cell probability. -
threshold_occupancy: Probability threshold to consider cell as occupied. -
map_publish_frequency: Throttled frequency to publish the map.
The following planner parameters correspond to the autonomous exploration behavior, and are loaded by the bsp_planner node from a yaml configuration file (see provided bsp_planner/cfg/bsp_settings.yaml):
-
system/v_max: Max translational speed [m/s] (double). -
system/dyaw_max: Max yaw rate [rad/s] (double). -
system/camera/pitch: Mounting pitch of the camera depth sensor [deg] (double). -
system/camera/horizontal: Horizontal Field-of-View opening of the camera depth sensor [deg] (double). -
system/camera/vertical: Vertical Field-of-View opening of the camera depth sensor [deg] (double). -
system/bbx/x: Safe bounding box around vehicle for collision avoidance, x dimension [m] (double). -
system/bbx/y: Safe bounding box around vehicle for collision avoidance, y dimension [m] (double). -
system/bbx/z: Safe bounding box around vehicle for collision avoidance, z dimension [m] (double). -
system/bbx/x_offset: Offset for the safe bounding box in the x dimension [m] (double). -
system/bbx/y_offset: Offset for the safe bounding box in the y dimension [m] (double). -
system/bbx/z_offset: Offset for the safe bounding box in the z dimension [m] (double). -
system/bbx/overshoot: Collision-checking extension, added to the length of the sampled path segment [m] (double).
-
nbvp/gain/unmapped: Gain for visible Unmapped voxels, used as volumetric exploration gain (double). -
nbvp/gain/free: Gain for visible Free voxels, used in combination with the probabilistic reobservation gain (double). -
nbvp/gain/occupied: Gain for visible Occupied voxels, used in combination with the probabilistic reobservation gain (double). -
nbvp/gain/probabilistic: Gain for probabilistic reobservation of already mapped (Free or Occupied) voxels (double). -
nbvp/gain/zero: Zero-gain value, 0.0 (double). -
nbvp/gain/range: Max camera depth sensor range - max distance of voxels to be considered for gain computation [m] (double). -
nbvp/gain/degressive_coeff: Coefficient lambda for penalizing cumulative traveled distance when calculating gain at each node (double). -
nbvp/gain/degressive_switchoffLoops: Number of planning loops for which the degressive penalization remains active (int). -
nbvp/tree/extension_range: Max extension range for new RRT branches [m] (double). -
nbvp/tree/exact_root: Start the newly sampled tree from the previous end point -true-, or from the current pose -false- (bool). -
nbvp/tree/initial_iterations: Initial number of iterations for RRT creation (int). -
nbvp/tree/cutoff_iterations: Cutoff number of iterations for exploration sampling, if no information gain higher than the zero-gain value has been found, sampling stops there (int). -
nbvp/dt: Path planning time step [s] (double).
-
bsp/enable: Enable/disable uncertainty-aware nested replanning layer (bool). -
bsp/replanning_distance_min: Min edge distance of 1-st layer next-best-viewpoint to consider for replanning (double). -
bsp/replanning_distance_ratio: Max allowed travelled distance for path replanning, given as a ratio with respect to original 1-st layer edge (double). -
bsp/replanning_extension_ratio: Max extension range for new RRT path replanning branches, given as a ratio with respect to original 1-st layer edge (double). -
bsp/replanning_initial_iterations: Initial number of iterations for RRT path replanning (int). -
bsp/replanning_cutoff_iterations: Cutoff number of iterations for RRT path replanning, if no better optimality gain than the straight-transition along 1-st layer edge has has been found, sampling stops there (int). -
bsp/replanning_gain_range: Max camera image sensor range - max distance of locally tracked landmarks to be considered visible when sampling the new path - at least 3 landmarks have to be visible from each new viewpoint (double).
-
init/softStart: Flag to clear -mark as Free- an extended bounding box around vehicle until first successful planning iteration - used when the camera depth sensor Field-of-View is limited and the view cone does not clear enough space for collision-free planning to proceed given the system bounding box and overshoot safety thresholds (bool). -
init/softStartMinZ: When soft start is active, allows clearing voxels only above this height level [m] (double).
-
bbx/minX: Minimum x-coordinate value of the scenario, bounds the RRT planning [m] (double). -
bbx/minY: Minimum y-coordinate value of the scenario, bounds the RRT planning [m] (double). -
bbx/minZ: Minimum z-coordinate value of the scenario, bounds the RRT planning [m] (double). -
bbx/maxX: Maximum x-coordinate value of the scenario, bounds the RRT planning [m] (double). -
bbx/maxY: Maximum y-coordinate value of the scenario, bounds the RRT planning [m] (double). -
bbx/maxZ: Maximum z-coordinate value of the scenario, bounds the RRT planning [m] (double). -
bbx/explorationExtensionX: Extension for exploration along the x-coordinate, [minX-explorationExtensionX , maxX+explorationExtensionX] bounds the gain computation [m] (double). -
bbx/explorationExtensionY: Extension for exploration along the y-coordinate, [minY-explorationExtensionY , maxY+explorationExtensionY] bounds the gain computation [m] (double). -
bbx/explorationMinZ: Minimum z-coordinate for exploration, bounds the gain computation [m] (double). -
bbx/explorationMaxZ: Maximum z-coordinate for exploration, bounds the gain computation [m] (double).
-
markers/planMarker_lifetime: Visualization lifetime of MarkerArray message corresponding to the full RRT of the Next-Best-View 1st planning layer [s] (double). -
markers/bestPlanMarker_lifetime: Visualization lifetime of MarkerArray message corresponding to the best RRT branch of the Next-Best-View 1st planning layer [s] (double). -
markers/replanMarker_lifetime: Visualization lifetime of MarkerArray message corresponding to the full RRT of the Belief-Space-Propagation 2nd planning layer [s] (double). -
markers/bestReplanMarker_lifetime: Visualization lifetime of MarkerArray message corresponding to the best RRT branch of the Belief-Space-Propagation 2nd planning layer [s] (double).
-
pcl_throttle: Minimum time between subsequent pointcloud insertions [s] (double).
Contact details: Christos Papachristos, Shehryar Khattak, Kostas Alexis, Autonomous Robots Lab