This is a repository of the data presented in the research paper *Model predictive altitude and velocity control in ergodic potential field directed multi-UAV search* by Luka Lanča, Karlo Jakac and Stefan Ivić.
The supplementary data for all presented test cases in the manuscript includes the following:
- FEM domain meshes in gmsh2 format (.msh file)
- Initial target probablity (m0) and terrain elevation (elevation) fields (.vtk file)
- Initial state of the UAVs (.txt file)
- Search animation (.mp4 file)
- Final state after search completion (.npz file)
The case solutions computed using the HEDAC method with Model Predictive Control (MPC) are located in the "HEDAC with MPC" folder. The solution files and animations for the methods used for comparison can be found in the "HEDAC" and "Lawnmower with MPC" folders. Files that are shared among all methods are stored in the root case folder.
Example for loading .npz files:
import numpy
npz = numpy.load("plastic_world_final_state.npz")
print(npz.files)
AXYZ = npz["AXYZ"]
print(AXYZ)
The .npz file contains:
- it - iteration when state was saved
- AXYZ - Agents coordinates matrix - [Agent index, Timestep, (0,1,2) -> (x,y,z) coords]
- AD - Heading angle matrix - [Agent index, Timestep]
- AV - Agent velocities matrix - [Agent index, Timestep, (0,1) -> (vs,vz)]
- ARPO - Agent control matrix - [Agent index, Timestep, (0,1,2) -> (rho, phi, omega)]
- AEM - Agent escape maneuver [Agent index, Timestep, (0,1,2) -> (rho, phi, omega)]
- PPOPT - Initial optimization point saved from last step (not used)
- c - coverage (at every mesh point)
- gfu - gradient function of the potential field *u*
- execution time - processes computation times for every timestep
- eta - survey accomplishment for all timesteps
