DE19602607C1 - Autonomous docking of passenger aircraft e.g. with passenger entry ramp, deck of aircraft-carrier - Google Patents

Abstract

The method employs a range-imaging camera (1) to generate raw data for evaluation by an integrated microprocessor (2). A programmable read-only memory (3) holds software for recognition and position measurement of the aircraft's nosecone. An external interface (4) allows communication at a low data rate with an external computer. The field of view of the camera is fixed and the nose position is located by Hough transform allowing a predetermined imprecision in at least one co-ordinate direction to compensate for statistical inaccuracies in the measurements.

Description

The invention relates to a method for docking an aircraft according to the Preamble of claim 1.

An aircraft docking system does the job when a pass arrives to replace pilot aircraft required at the gate. Without a solder The pilot is unable to set his stop point with sufficient accuracy drive.

So far, aircraft docking systems based on so-called parallel are known process (passive) work. Along the Einrollitlinie are one-dimensional Rangefinder arranged.

DE 40 09 668 A1 describes a method for docking an aircraft known in which the current position of the aircraft by evaluating a Video image is determined. Instead of a video camera, an Ultra can also be used sound detector line or a radar antenna can be used as a sensor.

DE 43 01 637 A1 describes a further method for docking egg  nes aircraft, in which a distance image camera cyclically removes Images of the docking aircraft are recorded. From the The distance data recorded is compared with an aircraft type The fuselage tip template determines the current position of the fuselage tip telt.

The object of the invention is a method for docking an aircraft indicate that even with a noisy signal and regardless of the con the current position values of the dock reports the aircraft bugs in real time.  

These tasks are accomplished with the patentan method spell 1 solved. Advantageous training is the subject of further claims.

In the method according to the invention, the pilot is during the one the respective position of his aircraft is automatically displayed during the parking process, e.g. B. on a widely visible, attached to the gate display. Such one Docking system requires an optical device in the "seeing" component Form of a distance camera, which is permanently installed at the gate and which detects the parking plane from the front.

The process works automatically. It only receives at the beginning of each one docking process as input to determine the aircraft type, its bow position is. Output are the current position coordinates in two dimensions.

The process works autonomously. In particular, neither is a mark to attach to the aircraft nor must these or the tarmac in ir in some way.

In addition, the invention has the following advantages:
The procedure works equally during the day and at night.

The method has a range of at least 100 m.

Imaging sensor - the bow position is in both horizontal dimensions determined.  

Object detection - the method delivers good ge even when the aircraft is moving accuracy values.

High accuracy in position determination. The procedure has in the after range a positioning accuracy of 0.1 m (variance).

The method has an image rate of at least 3 Hz, so that the moving aircraft bow can be measured.

The method outputs the measured position values in real time.

The procedure only responds to the registered aircraft.

A device for performing the method is housed in a single housing. Through integration The sensor and processor have compact dimensions.

The device can be programmed to operate differently positions are adjusted.

The following is an embodiment of the system according to the invention explained in detail with the help of figures. Show it:

Fig. 1 shows the overall configuration of the system according to the invention

Fig. 2 is a flowchart of the docking system state

Fig. 3 sketch to explain the template-based unsharp Hough transformation.

In the embodiment shown in FIG. 1, the docking system according to the invention consists of a distance camera 1 for raw data generation, an integrated, commercial microprocessor 2 for raw data evaluation, a PROM 3 with software for aircraft bow detection and position measurement, and an external interface 4 with a low data rate for communication with an external computer (output of the position results).

All elements 1 , 2 , 3 , 4 can be integrated in one housing, so that there is a compact overall system.

The distance image camera (EBK) 1 generates distance images of the following quality:

• - Field of view: 30 ° × 30 ° (horizontal × vertical)
• - Pixel resolution: 0.5 ° × 0.5 ° (horizontal × vertical)
• - Range: 100 m
• - Distance resolution: 0.06 m
• - Measured value calibration: available
• - Measurement direction calibration (with mechanical scan): available
• - Refresh rate: 4 Hz

For the implementation of the SFBC, e.g. B. DE 39 42 770 C2 and DE 43 20 485 A1 can be used as a basis.

The microprocessor 2 is a commercial RISC processor, which is implemented with RAM, PROM and two interfaces on a circuit board. The processor board is integrated as an additional plug-in card in the electronics module of the EBK and must meet the following performance requirements:

• - Processor performance: 100 MIPS
• - RAM memory: 8 MB
• - PROM memory: 2 MB
• - 1st interface (internal to the sensor front end): 40 kB / second
• - 2nd interface (external): 100 baud

The software 3 consists of a commercial, real-time capable, multiprocessing operating system and three application modules A, B, C, which are operated as separate processes. The application modules are a module A for reading in the raw data from the SFBC via the internal interface; a module B for communication with an external computer via the external interface 4 ; an image processing module C for bug detection and position measurement.

The image processing module C consists of a sequence control, which is between system conditions

• (1) Standby
• (2) initialization
• (3) Docking

toggles.  

The default state is the so-called standby. The Sy is waiting in this mode stem to an external input that causes a docking process and the Type of aircraft to be docked. With this information the Sy occurs stem into initialization mode and then into docking mode.

The docking process runs cyclically in the picture cycle. The following processing steps are processed one after the other ( Fig. 2).

• (1) Transfer of the raw data from the data buffer (image-wise).
• (2) decoding and transformation of raw data
  The sensor measured values are converted into distance measured values (in the unit of meter), checked for validity, calibrated and assigned to an image line and column; the associated measuring directions are determined dynamically with high precision and used for the subsequent transformation of the measured values into their vector representation in the earth-fixed coordinate system.
• (3) Template-based, fuzzy Hough transformation
  The task of this module is the computationally efficient, robust localization of an aircraft bow in the vector data (see Appendix 1). This requires a 3D template, which is taken from a database belonging to the software during the initialization phase. The conventional, "sharp" Hough transformation is replaced by an "unsharp" to compensate for statistical inaccuracies in the measured values. By evaluating the Hough transform, you can first determine whether the bug you are looking for is in the image; secondly, its optimal position is determined with sub-pixel accuracy.
  The Hough transformation described can only be used with vector data as generated by the SFBC. The software described therefore only works in conjunction with an SFBC.
• (4) Analysis and output of results
  A current result of the bug detection is compared with the results from previous images. Only when a bug has been detected several times in successive pictures is an external result output initiated.

The subject of claim 1 includes those for the solution of the techni critical combination of special hardware with problems special software. The software described is only in connection with the described hardware and only to solve the present technical Problem usable.

Template-based, fuzzy Hough transformation to the Buger aircraft identification

The following is the flow of the template-based fuzzy hough transformation as performed in the present invention be wrote:

Given an earth-proof, right-handed, Cartesian coordinate system with N measuring points {(x _i , y _i , z _i ) | i = 1. . . N}. The x-axis is oriented antiparallel to the longitudinal axis of the incoming aircraft, the z-axis points upwards. The x, y plane runs parallel to the runway.

There is also a template function

x ′ = σ (y ′, z); (y ′, z) ε D, (1)

which describe the aircraft bow in three dimensions and locally ( FIG. 3). If the bow tip is at x₀, y₀, all points (x, y, z) that lie on the bow surface satisfy the equation

x-x₀ = σ (y-y₀, z); (y-y₀, z) ε D (2)

It is assumed here that the bow template in the Height is fixed and that the front longitudinal axis runs parallel to the x-axis. Later, these assumptions can be made - at the price of a larger rake effort - to be canceled.

The task is to find a location (x₀, y₀) on the runway for which as many measurement points (x _{i (k)} , y _{i (k)} , z _{i (k)} ) apply at the same time:

(y _{i (k)} -y₀, z _{i (k)} ) ε D
k = 1. . .MN
x _{i (k)} 3-x₀ = σ (y _{i (k)} -y₀, z _{i (k)} ) (3)

For the sake of simplicity, it is initially assumed that the measuring points are error-free.

The Hough transform h (x, y) is generated to solve this task. It receives the initial value

h (x, y): ≡ 0 (4)

in the entire x, y plane. For each measuring point (x _i , y _i , z _i ) let a curve segment x = f _i (y) be defined as follows for all y with (y _i -y, z _i ) ε D:

f _i (y): = x _i - σ (y _i - y, z _i ); (y _i - y, z _i ) ε D (5)

On the graph Γ _i : = {(f _i (y), y) | (y _i -y, z _i ) ε D} are all potential bug positions (x₀, y₀) for which (x _i , y _i , z _i ) would be a point on the bow. You can see this immediately by inserting (5) into (3). The facts are shown for the level z ≡ z _i in FIG. 3. To calculate the Hough transform, the value of the function h (x, y) at all locations (x, y) ε Γ _i is increased by one for all measuring points (x _i , y _i , z _i ):

h (x, y): = h (x, y) + 1 for all (x, y) ε Γ _i (6)

At the end of this loop, h (x, y) has the following meaning: Put the bow tip at the location (x, y), there would be h (x, y) measuring points on the bow casing. To evaluate h, the maximum (x₀, y₀) of the Houghtransfor determined:

h (x₀, y₀) h (x, y) for all x, y. (7)

K <0 denotes the minimum number of points that fall on a bug must be recognized so that it can be recognized. If

h (x₀, y₀) K, (8)

(x₀, y₀) is the desired position of the bug; otherwise no bug was detected.

With the unsharp Hough transformation, the inaccuracy of the measured values (x _i , y _i , z _i ) is taken into account. A measurement inaccuracy of at most ± Δx _max in the x direction is permitted. For | Δx | Δx _max Γ _i (Δx) denote the parallel shifted graph Γ _i :

Γ _i (Δx): = {(f _i (y) + Δx, y) | (y _i -y, z _i ) ε D} (9)

Let w (Δx) denote a weight function with values between 0 and 1:

0 w (Δx) 1 (10)

which assigns a weight to a measuring point with error Δx. Usually is

w (0) = 1;
w (± Δx _max ) = 0 (11)

With these agreements, the one caused by the ith measuring point changes Finding (6) the Hough transformed h (x, y) as follows:

for all -Δx _max Δx Δx _max
and set all (x, y) ε Γ _i (Δx)

h (x, y): = h (x, y) + w (Δx) (6 ′)

In Fig. 3 is other than Γ _i (0) still Γ _i (-Δx _max) and Γ _i (Ax _max) located.

If the template height is also to be determined from the measuring points, is to generate a Hough transform in three dimensions. Be given again a template function

x ′ = σ (y ′, z ′); (y ′, z ′) ε D (12)

If the bow tip is in place (x₀, y₀, z₀), all local ones are sufficient Bug points (x, y, z) of the equation

x-x₀ = σ (y-y₀, z-z₀); (y-y₀, z-z₀) ε D (13)

The Hough transformed h (x, y, z) is initialized according to (4).

h (x, y, z): ≡ 0 in the entire x, y, z space (14)

The graph Γ _i (Δx) defines the patch

Γ _i (Δx): = (x _i -σ (y _i -y, z₁-z) + Δx, y, z) | (y _i -y, z _i -z) ε D (15)

and the update rule (6 ′) is:

for all -Δx _max ΔxΔx _max
and set all (x, y, z) ε Γ _i (Δx)
h (x, y, z): = h (x, y, z) + w (Δx) (16)

After generating the Hough transform, the maximum (x₀, y₀, z₀) of h (x, y, z) as a bow position in three dimensions in question.

For didactic reasons, the present presentation is based on a conti Nuclear Hough transform h. For a numerical implementation h and the elevation are discretized in a conventional manner.

Claims (6)

1. Autonomous method for docking an aircraft, where at

• - cyclically taking distance images of the docking aircraft with a distance image camera,
• - With a microprocessor, the distance image data are transformed into a vector representation in a fixed coordinate system and the detection and localization of the current position of the aircraft bow on the basis of a three-dimensional template, which corresponds to the bow of the docking aircraft type, is carried out from the vector data ,

characterized in that the field of view of the distance image camera is fixed during the course of the method, and the detection and localization of the current position of the aircraft bow is carried out by performing a Hough transformation, a predetermined measurement accuracy being permitted in at least one coordinate direction when the Hough transformation is carried out, to compensate for statistical inaccuracies in the measured values. 2. The method according to claim 1, characterized in that a Output of the current bow position only takes place after the bug is out several successive distance image recordings recognized has been. 3. The method according to any one of the preceding claims, characterized ge indicates that the frame rate is at least 3 Hz.   4. The method according to any one of the preceding claims, characterized ge indicates that the output of the current position values of the Bugs occurred in real time. 5. The method according to any one of the preceding claims, characterized ge indicates that the distance camera has a field of view of 30 ° × 30 ° in the horizontal and vertical directions.