Skip to content
Andreas Bircher edited this page May 11, 2015 · 22 revisions

To chose the used model (rotorcraft or fixed-wing) make the corresponding definition in the file “plan.hpp” that can be found in the package koptplanner in the folder koptplanner/include/koptplanner/

For the rotorcraft model:

#define USE_ROTORCRAFT_MODEL

And for the fixed-wing model

#define USE_FIXEDWING_MODEL

Below, available parameters are listed. For both systems fixed-wing and rotorcraft, default values are given if applicable. The parameters have to be loaded before the call to the service. A sample parameter file is given as “koptplanner/koptParam.yaml”. After changing the model type, rebuild the package with

catkin_make

invoked at the catkin workspace.

Sensor Parameters

The sensor is modeled as a camera with a rectangular field of view. The parameters of that model are the angular opening of the field of view in horizontal and vertical direction, as well as the pitch angle of the centre of the field of view.

alt text

Parameter Rotorcraft Fixed-wing Optional Description
camera/horizontal - - No Horizontal angle of field of view
camera/vertical - - No Vertical angle of field of view
camera/pitch - - No Sensor front down pitch

Rotorcraft Parameters

Dynamical constraints on the motion of the rotorcraft are imposed as a maximum translational speed, as well as a turnrate limit.

alt text

Parameter Rotorcraft Fixed-wing Optional Description
rotorcraft/maxSpeed 0.25 NA No Translational speed in m/s
rotorcraft/maxAngularSpeed 0.5 NA No Maximal yaw rate in rad/s

Fixed-wing Parameters

The fixed-wing model moves with a constant flight speed, either straight or in a left or right turn with a constant radius. It climbes or sinks with up to a maximum rate, adding loitering circles to the end of the path to overcome steeper connections.

alt text

Parameter Rotorcraft Fixed-wing Optional Description
fixedwing/speed NA 9.0 No Flightspeed of UAV
fixedwing/Rmin NA 60.0 No Minimal turn radius
fixedwing/maxClimbSinkRate NA 0.1 No Maximal climb and sink rate

Algorithm Parameters

This part contains parameters for the algorithm. Namely, vp_tol gives a tolerance on the Euclidean distance deviation of the viewpoint to the one sampled in the previous iteration. If the deviation does not exceed vp_tol, the old viewpoint is maintained along with its RRT* tree. The size of the bounding box of said RRT* trees is defined as rrt_scope in all directions, centred at the root. Every time an attempted connection fails, rrt_it iterations are performed, before another attempt is made. rrt_it_init gives the number of apriori iterations. The space around obstacles is searched individually on all sides for good viewpoints. This recursive search allows a maximum of max_obstacles_depth obstacles encountered. If lazy_obstacle_check, also lazy connection (whose distance is bigger than rrt_scope) are checked for obstacles. Collision checks of connections are performed using a discretization with step size discretization_step. Additional parameters allow tuning of the viewpoint sampling.

Parameter Rotorcraft Fixed-wing Optional Description
algorithm/vp_tol 0.005 0.05 Yes Tolerance to reuse RRT* tree (default value problem dependent)
algorithm/rrt_scope 25.0 500.0 Yes Size of RRT*-space (default value problem dependent)
algorithm/rrt_it 50 100 Yes Iterations of RRT* planner if no connection could be established
algorithm/rrt_it_init 0 0 Yes Initial RRT* planner iterations
algorithm/ max_obstacle_depth 3 3 Yes Depth of recursive obstacle avoidance search (large depths slow the algorithm down)
algorithm/ discretization_step 0.1 10.0 Yes Collision check intervall in m (default value problem dependent)
algorithm/ angular_discretization_step 0.2 0.2 Yes Discretization for exhaustive grid search of orientation in rad
algorithm/const_A - 1000.0 Yes Additional weight factor for height differences. Scaling with A’ = A/maxClimbSinkRate
algorithm/const_B - 3.0 Yes Weight factor for the squared distance to neighbour viewpoints
algorithm/const_C - 1e12 Yes Weight factor for the slack variable constraining the minimal distance to neighbours
algorithm/const_D 1.0 3000 Yes Weight factor for the squared distance to the viewpoint in the previous iteration
algorithm/ lazy_obstacle_check false false Yes

Scenario Parameters

To avoid collision, a minimal distance to obstacles is enforced when chosing viewpoints. To account for the dynamics of fixed-wing systems, the minimal radius is added in horizontal direction to the chosen security distance.

Parameter Rotorcraft Fixed-wing Optional Description
scenario/security_distance 2.0 Rmin Yes Minimal distance for viewpoints to an obstacle
Clone this wiki locally