Several weeks have passed since the last article about technical advances.
It’s time to reveal them.

Let's begin by a demonstration video of the project available on the left.
The definition of a mission, which includes the creation of a patrol area and the position of the bases are shown on the map.

The UAV takes off on its own, follows a predefined path, and then lands automatically on its base.
Due to a lack of funding, we only have one UAV, but the principle stays the same for a swarm.

The control interface has been revised to be as user-friendly as possible: just a few clicks are needed to define the mission and launch the swarm.
In order to ease the area's selection, an OpenStreetMap has also been implemented.
You will notice that areas of interest* and exclusion* can be defined in order to narrow the patrol area.

Once the mission is launched, the patrol agent forward a HD video stream which can be seen directly through the control interface. The ground operator can visualize it through that way. This stream is forwarded through a secure internet network thanks to a long range WiFi antenna.

However, this video stream is not only relayed, it is also analysed in order to determine relevant elements in function of the mission.
Currently, the patrol agents are equipped with an intelligence allowing them a simple detection of human faces and pedestrians.
Detection and tracking of intruders seen from the sky are in development.

For its part, the Artificial Intelligence has been improved by including a survival algorithm.
This algorithm allows to the patrol agents to never be in a delicate situation. Their trajectory is calculated to never cross each other.
Concerning their energetic management, each of them computes in advance the energetic resource needed to move from a point to another and to land on their base before being out of energy.

Furthermore, in order to enlarge our applicative field and not to limit ourselves to an UAVs swarm, we decided to use the arduino project.
That allowed us to perform our first concluding test with a terrestrial vehicle: the vehicle moves to the given positions without any human interaction.
We also finished our API allowing controlling our patrol agents (take off, landing and automatic navigation).
Finally, interactions between UAVs and terrestrial vehicle are currently in integration.

*Areas of interest : area inspected more frequently by patrol agents
*Areas of exclusion : area not visited by patrol agents