CAMS is a software framework for cloud exploration using fleets of UAVs. The system drives a fleet of UAVs inside clouds acoording to user's plans and collects sensor data. Afterwards the information is processed by dedicated atmosphere analysis algorithms to produce 4D cloud maps that are simultaneously stored in a database for future use and shown to operator through a graphical user interface. This graphical user interface also provides operators the means to manage the fleet of UAVs .
The goal of this system is to provide an easy to use tool considering the following four design features:
- Monitor data acquisition:
- Interactive predicted 4D cloud maps (3D + time).
- Various data profiles, including raw data display and vertical atmospheric profiles.
- Generate flight plans:
- Easy management of the fleet of UAVs to plan cloud exploration missions. Generate complex flight plans based on adaptive cloud exploration trajectory patterns.
- Safe operation of the fleet with a dedicated tool for UAV remote pilots to validate generated flight plans.
- Monitoring of the UAV fleet status:
- Display of the UAV position as an overlay of the cloud map.
- Status updates for each UAV plan, including current task, flight time, battery voltage and many others.
- Simulation :
- Using the simulation capabilities of the Paparazzi framework and MesoNH atmospheric simulation data, the system can run with a combination of real and simulated components.
Modes of operation¶
CAMS designed to work in several modes of operation fulfilling different user use cases:
- Pure simulation
- Field deployment
- Mission replay
The software is made of several key sub-systems:
- Graphical user interface: Web-based user interface for the operation of the fleet of UAVs and display of sensor data and cloud maps.
- Paparazzi interface: The paparazzi autopilot is a libre software and hardware platform for autonomous unmanned aerial vehicles encompassing the flight controller and the ground station. CAMS provides a software interface that takes care of the communication with UAVs and the ground station to receive sensor data and control the fleet.
- Full dense maps of clouds using sparse data generated by the UAVs, with the help of Gaussian Process Regression (GPR).
- Higher level functions to estimate some cloud parameters from a dense map (segmentation, area, border, etc...).
- Single map interface, behind which can be a GPR predicted map, or a section of MesoNH data to be used as ground-truth in simulation.
- MesoNH interface: Provides functions to read MesoNH atmospheric simulation databases so they can be used to simulate realistic cloud environments for UAV flight and sensor data acquisition. Wind, pressure, liquid water content, and many other variables can be exploited as emulated sensor data.
- Data server: Aggregates all data coming down from flying or simulated UAVs (or any other source) into a single server instance. Any module which needs data, like the mapping module or the GUI module must require data from the data server.