US 12,393,202 B2
Method and device for automatically guiding an autonomous aircraft
Sébastien Descude, Toulouse (FR); Christian Sannino, Toulouse (FR); and Hugo Boronat, Toulouse (FR)
Assigned to THALES, Courbevoie (FR)
Appl. No. 18/275,489
Filed by THALES, Courbevoie (FR)
PCT Filed Feb. 2, 2022, PCT No. PCT/EP2022/052395
§ 371(c)(1), (2) Date Aug. 2, 2023,
PCT Pub. No. WO2022/167441, PCT Pub. Date Aug. 11, 2022.
Claims priority of application No. 2101143 (FR), filed on Feb. 5, 2021.
Prior Publication US 2024/0118706 A1, Apr. 11, 2024
Int. Cl. G05D 1/229 (2024.01); B64U 10/25 (2023.01); B64U 101/31 (2023.01); G05D 1/46 (2024.01); G05D 1/689 (2024.01); G05D 101/15 (2024.01); G05D 105/85 (2024.01); G05D 109/22 (2024.01); G05D 109/25 (2024.01)
CPC G05D 1/2295 (2024.01) [G05D 1/46 (2024.01); G05D 1/689 (2024.01); B64U 10/25 (2023.01); B64U 2101/31 (2023.01); B64U 2201/10 (2023.01); G05D 2101/15 (2024.01); G05D 2105/85 (2024.01); G05D 2109/22 (2024.01); G05D 2109/254 (2024.01)] 11 Claims
OG exemplary drawing
 
1. A method for guiding an autonomous aircraft, the aircraft comprising an automatic pilot, a plurality of sensors and an imaging unit, the aircraft being configured to fly over a geographic zone comprising overflight prohibited zones and a safe corridor that does not pass through said overflight prohibited zones, said imaging unit being able to photograph one or more objects of interest located in said geographic zone, wherein the guidance method comprises a phase of real flight of the autonomous aircraft between a position of departure and a predefined position of arrival that are associated with said safe corridor, by using a given guidance law, the guidance method comprising the following steps, implemented at a given instant of said real flight phase:
determining a current state of said autonomous aircraft as a function of a plurality of input parameters comprising input parameters supplied by said plurality of sensors;
determining an optimum action to be executed by using a neural network receiving said current state;
determining a plurality of control instructions compatible with said guidance law based on said optimum action to be executed;
transmitting to said automatic pilot said plurality of control instructions, which provides a new state of said autonomous aircraft;
and in that the method comprises a preliminary learning phase consisting in optimizing said neutral network based on a plurality of fictitious flights of said aircraft, each of said fictitious flights comprising a succession of fictitious steps,
and wherein said neural network represents a value function or a strategy associating with each current state of said autonomous aircraft at least one action, one of said at least one action being said optimum action to be executed.