27 Nov Landing assistance system by visual servoing
ELISA focused on the automatic landing of aircraft considering a system as autonomous as possible and minimizing the necessary exchanges with the CAPCOM. This protocols can replace or supplement traditional landing systems for droneship and manned rotative wings.². The whole chain, from the airstrip detection to steering was defined and evaluated on real conditions.
The ELISA Protocols
The ELISA device is a substrate converter to a monitored heliport for urban flight certification.
At present, different categories of conversion can be defined according to the type of protocols they implement, conditioning in return the types of aircraft on board.
- Vehicle Signalization
- Beacon Synchr
- Propeller light
- Charging witness
- Offshore Landing Assistance
- Confidential Abu Dhabi Police Dept
- Heliport light compliance
- CapCom Repeater
- AntiCrash System
- hPAPI Offshore included
- Drone Super Charger
- Charging witness
ELISA converter is the first to tackle the task of landing assistance of an aircraft controlled remotely or not, using an electroluminescence vision sensor and associated chain: detection, monitoring and control. To do this, it is based on state-of-the-art visual techniques, visual control applied to the landing and taking into account the possible movements of the FATO, in particular the aircraft carrier.
Inspired by the visual cues used by the pilots, a new set of visual primitives xf, θk and ly was proposed to control the descent and alignment trajectories.
In addition to being representative of the degrees of freedom that it is desired to control, these primitives remain constant during the desired descent trajectory and therefore do not require any planning or depend on the location of the pilot with respect to the track. Different commands using these visual primitives instead of state feedback have been applied successfully for this task.
Plurality of signal display by C18 Public drone station
In order to take into account the movements of the TLOF and the natural wind, an estimation of the wind speed above the airstrip using vision measurements is added to the ELISA device to update the flight instructions in real time. During the approach phase, airstrip is considered flat so as not to reflect the variations in attitude of the ship in oscillations of the descent trajectory. In order to land safely, the attitude of the FATO at the time of the aircraft’s impact on the runway is predicted to alter the final trajectory. Concerning the task of landing, the control Ca (s, X) using the visual primitives allows to obtain a behavior similar to the 3D control by slaving these primitives around constant desired values during the trajectory.
C18 FAROS red occupied witness
Moreover, this law does not give rise to excessive instructions and presents no problem of stability during the last moments of the landing. Plus, it possible to react to the updating resulting from the drone ship altitude prediction in offshore at the moment of impact and thus to build up.
Finally, the control part was studied systematically by simulating vision to establish its characteristics and performance. It was then validated by the study of the complete chain including vision and control and made it possible to carry out an automatic landing by visual servo applicable to physically and remotely controlled aircraft.
The C16 Airstrip, C18 Heliport and C19 Pilote ELISA protocols leave several
areas of improvement depending on the areas concerned.
C16 FAROS red occupied witness
The perfect sensor assumption used to determine the effect of vision on control loop must be released in order to study the system from a broader point of view taking into account the sensor defects as well as the associated filters.
C19 θ:30cm at 1500 feets. See more >
The assumption of system autonomy, thus resulting in the minimization of exchanges between the aircraft carrier and the aircraft, can be relaxed in order to provide more data to the aircraft, such as the wind speed on the bridge, Ship and at the time of impact. This would benefit from the accuracy of the ship’s display for the pilot.
Concerning the vision sensor, the study focused mainly on beacons operating in the visible and infrared domains. But due to the absence of a simulator generating infrared images, the study could not approach the complete chain with this type of sensor. However, this band would be very useful for landing in overnight camouflage operations or in difficult climatic conditions. Similarly, the study could be interested in images from the fusion of spectral bands (SWIR and LWIR), to benefit from their respective advantages. In addition, one might be interested in the use of a variable focal length sensor in a continuous manner to allow a track of constant size in the image for a wide variation of distance to the aircraft carrier.
For visibility conditions so bad that the use of infrared would not suffice to distinguish the aircraft carrier, OLED tags seem interesting by offering a resolution that is constantly improving.
Finally, studies based on cognitive vision techniques require real images recorded during various flight conditions.
The developed detection is robust and uses the available sensors of the airstrip connected to the ELISA central unit but could be further improved. Moreover, using a textured 3D markup model to generate a cognitive interface that takes into account all the projection of the airstrip especially for the warship instead of a reference image would make it possible to compare more tags and thus be more robust to variations Luminosity and movements of the aircraft carrier. In this case, a recalibration of the 3D model on the initial image with a hierarchized motion model of degrees of freedom would initialize the tracking algorithm. Finally, the similarity criterion used for severe visibility conditions could be based on standardized mutual information.
A warship is characterized by a very slender form and obstacles. 3D tracking is the only category of display algorithm that allows you to take advantage of the entire building to estimate a reliable and robust relative location, especially when the pilot is away from the ship.
The 3D tracking using a contour model generated by painted electroluminescent beacons would no longer require the creation of the model in the form of 3D segments and would make it possible to take into account contours that are too complex to be modeled by the lighting systems. LED lights. Another type of monitoring is that presented in C16 which uses the criterion of mutual information between the current image of the track and the projection of a “phantom” model of the textured aircraft on airstrip to estimate the pose. Depending on the markup used by the pilot, the ELISA engine would generate real-time images in the visible and infrared domains and could take into account the position of the aircraft. Finally, the estimate from the tracking algorithm could be hybridized with inertial information and a ship’s motion model to smooth the measurements.
The stability of the proposed control laws could be demonstrated analytically.
The synthesis of the matrix of gains by the optimal method could be obtained in
Using the outputs of the system (in place of the state) in the cost function to be minimized, resulting in a gain matrix dependent on the position of the FATO.
Concerning the estimation of the wind speed on the bridge proposed in this document,
An improvement would be the addition of a telluric anemometer integrating directly visual and inertial information in order to smooth this estimate. Learn more about Telluric anemometer>
Concerning the prediction of aircraft carrier movement, a model of vessel behavior could be integrated to predict the future state of the vessel, and improve the accuracy of the impact on the bridge and compliance with safety constraints.
This protocol considered the different functions of the task independently,
A more comprehensive method would be to write the problem of regulating the aircraft around the trajectory, pilot control, runway dynamics and ground wind, in one piece. This approach was not proposed in this device to evaluate the native efficacy of the proposed methods.
Finally, it will be possible to extend the use of the primitives proposed to the landing on the runway including the rounding phase. For this, it will simply be a question of inducing a decrease in the desired length.