DE3923458C2 - Guiding system for guiding an unmanned, steerable vehicle along a predetermined path - Google Patents

Guiding system for guiding an unmanned, steerable vehicle along a predetermined path

Info

Publication number
DE3923458C2
DE3923458C2 DE19893923458 DE3923458A DE3923458C2 DE 3923458 C2 DE3923458 C2 DE 3923458C2 DE 19893923458 DE19893923458 DE 19893923458 DE 3923458 A DE3923458 A DE 3923458A DE 3923458 C2 DE3923458 C2 DE 3923458C2
Authority
DE
Germany
Prior art keywords
vehicle
angle
course
signal
wheel
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
DE19893923458
Other languages
German (de)
Other versions
DE3923458A1 (en
Inventor
Uwe Krogmann
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Bodenseewerk Geratetechnik GmbH
Original Assignee
Bodenseewerk Geratetechnik GmbH
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Bodenseewerk Geratetechnik GmbH filed Critical Bodenseewerk Geratetechnik GmbH
Priority to DE19893923458 priority Critical patent/DE3923458C2/en
Publication of DE3923458A1 publication Critical patent/DE3923458A1/en
Application granted granted Critical
Publication of DE3923458C2 publication Critical patent/DE3923458C2/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05DSYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00Control of position, course or altitude of land, water, air, or space vehicles, e.g. automatic pilot
    • G05D1/02Control of position or course in two dimensions
    • G05D1/021Control of position or course in two dimensions specially adapted to land vehicles
    • G05D1/0268Control of position or course in two dimensions specially adapted to land vehicles using internal positioning means
    • G05D1/0272Control of position or course in two dimensions specially adapted to land vehicles using internal positioning means comprising means for registering the travel distance, e.g. revolutions of wheels

Description

Technical field

The invention relates to a guidance system for guiding a unmanned, steerable vehicle along a predetermined Web according to the preamble of claim 1.

Underlying state of the art

Unmanned vehicles are used in modern large-scale production for transporting assembly parts between different Points of a manufacturing plant. These vehicles are guided along a given path. Here the guidance is usually carried out by laying in the ground Induction loops. The installation of such Management system therefore requires the relocation of Induction loops in the floor. It is very time-consuming. Also such guidance systems are very inflexible: the Movements of the vehicles are limited to the railways, along which the induction loops are laid. If the tracks along which the vehicles are guided, should be changed, then new induction loops must be can be laid in the ground with great effort.  

Navigation devices are known which use the method the dead reckoning work (DE 25 45 025 B2). Here is the navigation of manned vehicles in terrain. It's not about getting a vehicle automatically to run along a predetermined path. In the Coupling navigation is made from the possibly changeable Course and the also variable speed of the vehicle by dismantling the components Speed and integration the position of the vehicle determined in two coordinates. The course angle will measured here by a gyroscope.

From DE 26 59 094 B2 it is also known to a navigation device for navigating manned Vehicles in the field when reaching a geodetic Point, i.e. a striking site with well-known Coordinates, by comparing the displayed and the actual position not just the position indicator too correct but also for the slip on the Attached speed or distance measurement Correction factor so that the position continues with a corrected correction factor is determined.

US 4,847,769 shows a vehicle according to a dead reckoning is performed, the Dead reckoning by stationary, sensed by the vehicle Brands is supported. In US 4,847,767 course and Speed with the help of the turning angle of one wheel and the rotation of this wheel. In doing so, a "Continue coupling" is carried out over relatively short time intervals. The Brands are fixed, arranged on the side of the vehicle Reverse reflectors, which are aimed using a laser. In this respect, the arrangement according to US 4,847,769 also coincides U.S. 4,817,000, in which the navigation at all only based on such targeted lateral Reverse reflectors are done.

The signal processing of US 4,847,769 receives as Input variable the turning angle and the angle of rotation of the Wheel. A "coupling computer" forms from it Course angle increment and a path increment. The Course angle increment is integrated and provides one Estimate of the heading angle. The path increment is applied a transformation circuit, to which the estimated value the course angle is activated. Surrender to Outputs estimates for the position coordinates. Out of it an estimated value for the Direction to the reversing reflector calculated. This is with the observed direction compared. The difference is the Input variable for a Kalman filter, through which the System parameters can be varied integrating.

There will be each reversing reflector in the different "Bearing points" targeted several times. It's like targeting Lighthouses in coastal shipping: It gets through Dead reckoning obtained an estimate for the position. By aiming at lighthouses, always the one same lighthouses, only at different angles appear, the true position is determined. This can not only to correct the position but also - in the Shipping - the drift or a course error.

The transfer of this principle to vehicles in one Factory hall becomes quite complex. In addition, the side Bearing z. B. be affected by obstacles.

EP 0 193 985 A1 provides a grid of passive markings from which the position of the vehicle results. It is no navigation system is provided, the first one Path controlled by dead reckoning.

Disclosure of the invention

The invention has for its object a guidance system of the type mentioned (US 4,847,769) so that it can be installed and modified with little effort can.

According to the invention this object is achieved by the Characteristic of claim 1 specified measures.  

The path along which the vehicle is to be guided is no longer due to guiding means such as Induction loops set but in memory entered. The position of the vehicle is indicated by a autonomous navigation system determined. A controller and steering means ensure that the vehicle from its through the Navigation system determined position on the in memory given path is returned if it is from this Path deviates. All parts of the management system are autonomous included in the vehicle. It is not necessary, Guide means to be laid along the entire track in the ground. If routes are to be changed, they need to be changed saved routes only to the additional possible Lanes added, d. H. to be entered into memory.

In contrast to the state of the art solutions Marks only placed in discreet places on the floor. These marks directly define the path, like a polygon, along which the vehicle is to move. Any of these Brands are only scanned once when the vehicle is moved along this path. The side storage of the vehicle of each of these marks provides a correction to both the Position and possibly the parameters of the dead reckoning.

From the "lighthouse" function of the reversing reflectors after the US 4,847,769 differs the guidance system according to the Invention in that

  • - the arrangement of the marks on the floor itself the way determine the length along which the vehicle is to move, and
  • - Each of these marks provides a single fixed point, which is scanned only once each time the vehicle is run.
    The subject of the application differs from the usual known arrangements with induction loops or the like in the ground (for example US Pat. No. 4,716,530) in that
  • - Navigates the vehicle with an autonomous dead reckoning and only in certain points the dead reckoning by brands is supported.

Such brands are also much easier to affix than about induction loops. You don't need in the ground to be relocated and can be easily changed.

The guidance system can work without a gyro. This will the guidance system is very simple and inexpensive. Through the Correction of the parameters can be achieved that the vehicle a path determined by the marks on the floor "learns" increasingly better.

It differs from the arrangement according to EP 0 193 985 A1 the guidance system according to the invention through the use an autonomous dead reckoning. It is not a network of brands from which position and path is determined and that is not much less complex than a system of Induction loops. Individual brands, targeted along the arranged moving way, serve to support the Dead reckoning system.

Embodiments of the invention are the subject of the dependent claims.

Some embodiments of the invention are below with reference to the accompanying drawings explained.

Brief description of the drawings

Fig. 1 is a schematic plan view of a vehicle that can be guided along a predetermined path and also shows the names of the coordinates and angles.

FIG. 2 is a block diagram of a guidance system for a vehicle according to FIG. 1.

FIG. 3 shows, as a block diagram, a navigation computer in a guidance system according to FIG. 2, which works without a gyro.

FIG. 4 shows a block diagram similar to FIG. 3 of a navigation computer in a guidance system according to FIG. 2, in which the course angle is supplied by a course gyroscope.

FIG. 5 shows a block diagram similar to FIG. 2 of a navigation computer in a guidance system which works without a gyroscope and in which the vehicle is guided along a straight path.

FIG. 6 shows a schematic plan view of a vehicle in a coordinate system and illustrates a first type of detection of navigation errors.

FIG. 7 shows a schematic top view of a vehicle in a coordinate system and illustrates a second way of determining navigation errors.

Fig. 8 is a flow chart showing the process of error compensation.

FIG. 9 is a block diagram and shows the navigation computer with a first type of controller and the adjusting means for steering and speed.

Fig. 10 shows a model control loop for determining the parameters of the controller.

FIG. 11 is a block diagram similar to FIG. 9 and shows a guidance system for any lanes with the navigation computer, an adaptive controller and the adjusting means for steering and speed.

FIG. 12 is a block diagram of an adaptation unit in the adaptive controller of FIG. 11.

Fig. 13 shows an example of a Anordnuung waypoints and path sections in a field of use.

FIG. 14 shows a logical link network derived from the waypoints of FIG. 13.

FIG. 15 shows a "decision tree" derived from the link network of FIG. 14 for obtaining a special route from a location to a destination.

Fig. 16 is a schematic representation of an auto-associative memory to determine the optimum distances between the current position and a destination.

FIG. 17 is a schematic illustration for explaining the function of the associative memory using an example.

Fig. 18 is a diagram illustrating the teaching process of the auto-associative memory.

Preferred embodiments of the invention

Fig. 1 shows schematically a vehicle 10 in a coordinate system with the orthogonal coordinate axes x and y. The vehicle 10 has a first pair of wheels 12 and 14 which are mounted on the vehicle so as to be freely rotatable coaxially to one another at a relatively large distance from one another. A third wheel 16 and a fourth wheel 18 form a second pair of coaxial wheels, which are mounted on the vehicle at a short distance from one another in the vicinity of the longitudinal center plane 20 of the vehicle 10 . For steering purposes, the two wheels 16 and 18 can be pivoted about a vertical steering axis 22 lying in the longitudinal center plane 20 of the vehicle 10 . The third and fourth wheels 16 and 18 are driven. As a result, they run in a horizontal direction perpendicular to their axis. This is shown in Fig. 1 by the speed vector v (t). The angle between the longitudinal center plane 20 and the speed vector v (t) in the horizontal plane is referred to as the "turning angle" α (t). The angle between the median longitudinal plane 20 and the direction of the x-axis in a horizontal plane is the heading angle ψ (t).

Figure 2 is a block diagram of the guidance system. Block 10 represents the vehicle. Actuators for the drive, that is to say for setting the speed v (t) and for the steering angle α (t), are provided on the vehicle. This is represented by block 24 . The actuators 24 act on the kinematics of the vehicle 10 represented by block 26 . Sensors represented by block 28 provide either the speed v (t) and the turning angle α (t) or, alternatively, the speed v (t) and the heading angle ψ (t) and the turning angle α (t). The signals supplied by the sensors 28 are sent to a navigation computer 30 . From these signals, the navigation computer determines the actual coordinates x (t) and y (t) of the vehicle 10 in the coordinate system and the course angle ψ (t). A path which the vehicle 10 is to track is stored in a memory 32 . The memory 32 accordingly provides time-dependent command variables for the coordinates and course angles supplied by the navigator computer 30 . The control variables from the memory 32 and the actual coordinates and course angles from the navigation computer 30 are fed to a controller 34 . The controller generates control signals from the deviations of the actual coordinates and course angles from the command variables. These control signals are applied to the actuators 24 .

Fig. 3 shows an embodiment of the navigation computer 30th In the embodiment of the navigation computer 30 according to FIG. 3, the position and heading angle of the vehicle are determined without the aid of inertial sensors (heading gyroscope) from the speed v (t) and the turning angle α (t), which are measured by suitable sensors 36 and 38, respectively. The encoder 36 sits on the wheels 16 and 18 and delivers the speed of the vehicle 10 . The transmitter 38 sits on the chassis of the vehicle 10 and on the steering frame of the wheels 16 and 18 and supplies the steering angle α (t) of the wheels 16 and 18 .

The navigation computer 30 forms a course angle change signal (t) from the quantities v (t) and α (t) according to the relationship

This is represented by block 40 in FIG. 3. "A" is a vehicle parameter that indicates the distance between the navigation reference point on vehicle 10 and the point of application of the speed vector ( FIG. 1). The course angle change signal (t) is integrated over time. This is represented by block 42 . The initial value of the heading angle ψ (0) is entered at an input 44 . The course angle ψ (t) is thus obtained at an output 46 .

The signals v (t) and α (t) of the two transmitters 36 and 38 and the course angle ψ (t) obtained in the manner described above are also used in the navigation computer 30 to assign the time derivatives (t) and (t) to the coordinates form, so practically the components of the speed of the vehicle on its path in the coordinate system.

The time derivatives of the coordinates are after the relationship

won. This is represented by block 48 in FIG. 3. At block 48 , the signals from the two transmitters 36 and 38 and the course angle ψ (t) obtained by the integration are “switched on”. The time derivatives of the coordinates obtained in this way are integrated again. This is represented by block 50 in FIG. 3. The initial values y (0) and x (0) are entered at an input 52 . A position vector with the coordinates y (t) and x (t) is then obtained at an output 56 .

Another embodiment of the navigation computer is shown in FIG. 4. In the navigation computer of FIG. 4, a course gyroscope or a turning gyroscope 58 is used as a further transmitter. The turning gyro 58 immediately delivers the heading angle change signal (t). This course angle change signal (t) is integrated over time. This is represented by block 60 . The initial course der (0) is again input at an input 62 . The course angle ψ (t) is then obtained again at the output 46 .

The remaining part of the navigation computer of FIG. 4 is constructed in the same way as the corresponding parts of the navigation computer of FIG. 3. The corresponding parts in FIG. 4 are given the same reference numerals as in FIG. 3.

FIG. 5 shows an embodiment of the navigation computer 30 , which is designed for driving the vehicle 10 straight ahead along a straight line. The coordinate system is then placed with the x-axis in the direction of the straight path. This simplifies the calculation because the angles α (t) and ψ (t) can be assumed to be small. The sine can then be replaced by the angle and the cosine can be assumed to be "1".

The signals α (t) and v (t) of the two transmitters 38 and 36 then become a course angle change signal (t) according to the simplified relationship

educated. This is represented by block 64 in FIG. 5. The course angle change signal (t) thus obtained is integrated over time. This is represented by block 66 . The initial course der (0) is again entered at an input 68 . This provides the course angle ψ (t) at an output 70 . Here, this course angle means the deviation of the direction of movement of the vehicle 10 from the direction of the predetermined straight path along which the vehicle is to be guided.

The time derivatives of the coordinates y (t) and x (t) are again determined in the navigation computer 30 from the signals α (t) and v (t) from the two transmitters 38 and 36 , respectively. That happens here according to the relationship

This is represented by block 72 . Similar to block 48 in FIGS. 3 and 4, in addition to the signals from the transmitters 36 and 38 , the heading angle ψ (t) from the output 70 is “switched on” to the block 72 . The time derivatives of the coordinates y (t) and x (t) obtained in this way are integrated over time. This is represented in FIG. 5 by block 74 , which corresponds to block 50 in FIGS. 3 and 4. The initial coordinates are entered at an input 76 . The coordinates of the vehicle 10 are then obtained at an output 78 .

Autonomous navigation of the type described is fraught with errors. Such errors can result from errors in the transmitters 36 and 38 , errors in the assumed diameter of the wheels 16 and 18, and slip. Due to the integration, such errors lead to increasingly larger position and course deviations. The position of the vehicle 10 can therefore deviate somewhat from the desired position after having traveled a path. It is therefore important to support the position from time to time, that is to say to determine the position independently of the position calculator and to correct the position displayed by the position calculator in order to correct the error that is then determined. The signals of the transmitters 36 and 38 can also be corrected accordingly with regard to zero point and scale factor errors, so that the position computer 30 continues to work with corrected parameters.

For this purpose, marks 80 are attached in certain waypoints of the predetermined path according to FIG. 6. The vehicle 10 is stopped when it has reached such a waypoint 80 according to the coordinates appearing at the exit 78 of the navigation computer 30 . In fact, the vehicle 10 has then reached a position which deviates from the position of the mark 80 by the coordinate differences Δy and Δx. A sensor 82 is attached to the vehicle 10 and responds optically or inductively or in some other way to the mark 80 .

In FIG. 6, it is assumed that the predetermined path is a straight line the x-axis is along the coordinate system. The mark 80 is located in a waypoint WP2 on the x-axis. A previous waypoint WP1, at which the vehicle 10 started to travel or the last position support took place, lies a known distance D along the x-axis.

In FIG. 6, the two deviations Ax and Ay are measured by the sensor 82. The position supplied by the position computer 30 is corrected by these deviations. In addition, the course angle and the scale factor of the encoder 36 for the speed are corrected, so that the position computer 30 continues to work with corrected input signals.

Another method for error compensation is shown in FIG . A route 84 is marked there by a mark 84 which runs perpendicular to the path, ie the x-axis, in waypoint WP2. The vehicle 10 is stopped when it reaches this mark 84 anywhere. The position error Δx then becomes from the relationship

Δx = x - D (5)

certainly. Here x is the coordinate of the position, which is supplied by the position computer 30 at the output 78 . However, the "correct" coordinate is D. The difference therefore provides the display error. The position error Δy is measured by the sensor 82 . The displayed position and the errors of the transmitters 36 and 38 can also be corrected with these two values.

In FIG. 8, the flow of the error correction is shown in a flow chart.

According to block 86 , the passage of vehicle 10 is sensed by mark 84 at waypoint WP2. The vehicle 10 is stopped. Δx is calculated from equation (5). Δy is measured. This is represented by block 88 . The next step is the calculation of estimates for the course error and for the scale factor error of the encoder 36 . This step is represented by block 90 . The course error and the scale factor error result here in a simple manner from the position errors according to the relationships specified in block 92 .

A check is then carried out to determine whether the values obtained are plausible. This test is represented by rhombus 94 . If the values are not plausible (N), then only according to block 96 the distance D in the memory becomes after the relationship

D new = D old + Δx (6)

corrected. However, if the values are plausible (Y), the scale factor and course angle are also corrected by the determined scale factor and course angle errors. This is represented by block 98 .

Fig. 9 shows the guiding system with a first embodiment of the regulator. The navigation computer 30 corresponds to the embodiment of FIG. 5. The controller 34 here contains a summing point 100 on which the heading angle ψ (t) with a factor K and the position deviation Δy from the straight path (x-axis) with a factor K y are activated. The factors are represented by blocks 102 and 104 . The sum signal formed therefrom as a control signal corresponds to a reference variable α target for the steering angle α. This control signal is connected to a steering actuator 106 . This steering actuator 106 has a transfer function which is denoted by K α (s). The steering actuator 106 delivers the steering angle α (t). This steering angle is picked up by the encoder 38 . The signal from the transmitter 38 is connected to the navigation computer 30 . In addition, the signal from the encoder 38 serves as a position feedback. For this purpose, the signal from the transmitter 38 is fed back via a feedback loop 108 to the input of the steering servomotor 106 . At the input of the steering servomotor 106 , the difference between the command variable α soll and the signal α (t) of the transmitter 38 is formed in a summing point 110 .

The speed v (t) is kept constant in this embodiment. A speed servomotor 112 has a transfer function K v (s). The transmitter 36 outputs a signal corresponding to the speed v (t) to the navigation computer 30 . The signal from the encoder 36 is fed back via a feedback loop 114 to the input of the speed servomotor 112 . At the input of the speed servomotor 112 , the difference between a constant setpoint signal from a setpoint input 118 and the signal from the transmitter 36 is formed in a summing point 116 .

The most favorable feedback gains K y and K ψ for the steering can be determined for the entire occurring speed range using the model control loop shown in FIG. 10. In the model control loop of Fig. 10 it is assumed that the steering angle α and the heading angle ψ are small and that the velocity v is constant. The quantities ψ, α and y are represented as a function of s, i.e. by their Laplace transforms. It is further assumed that the steering actuator 106 has the transfer function "1".

At a summing point 100 , the feedback signals of the two feedback loops 118 and 120 are superimposed on a reference angle of α ref . This results (with the transfer function "1" of the steering servomotor 106 ) in a steering angle α (s) at the input of the navigation computer 30 . In accordance with block 122 , the steering angle is multiplied by -V / A. This corresponds to block 64 in FIG. 9. According to block 124, the product is multiplied by 1 / s. This corresponds to the integration according to block 66 of FIG. 9. The Laplace transform ψ (s) of the course angle results.

At a summing point 126 , the steering angle α (s) and the heading angle ψ (s) are added. The sum is multiplied by v according to block 128 . This corresponds to the first line of the matrix equation in block 72 :

(t) = v (t) α (t) + ψ (t) v (t)

or

(t) = (α (t) + ψ (t)) v (t).

The result is multiplied by 1 / s according to block 130 . This corresponds to the integration according to block 74 of FIG. 9. As a result, the Laplace transform Y (s) of the deviation y (t) of the vehicle 10 from the straight path running along the x-axis is obtained.

The feedback loop 120 contains the proportionality factor K y , represented by block 132 , which corresponds to block 102 of FIG. 9. Instead of a pure P-return z. B. PID feedback can be provided. Then the feedback branch contains additional elements dependent on s.

The feedback loop 118 contains the proportionality factor K ψ , represented by block 134 . The course angle must be returned with a different sign, depending on the direction of the constant speed v. This is represented in the model by block 136 , which provides the sign of the speed v.

Using the described model, the factors K y and K ψ can be calculated for the control loop of FIG. 9.

FIG. 11 shows a control circuit similar to FIG. 9 for guiding the vehicle 10 along any non-rectilinear path. The navigation computer 30 corresponds to FIG. 3. Corresponding parts are provided with the same reference numerals as there. The simplification for a linear path according to FIGS. 5 and 9 is omitted.

The navigation computer 30 supplies the heading angle ψ (t), which can no longer be assumed to be small, and the position of the vehicle 10 with the coordinates y (t) and x (t). The course angle ψ (t) supplied by the navigation computer 30 is compared in a summing point 140 with a target value ψ (t) *. A control deviation signal Δ ψ (t) is thereby obtained. The coordinates y (t) and x (t) of the position of the vehicle 10 provided by the navigation computer 30 are compared in a summing point 142 with target values y * (t) and x * (t). Control deviation signals Δy (t) and Δx (t) are obtained from this. The control deviation signals Δψ (t), Δy (t) and Δx (t) are linearly combined to form control signals Δα (t) for the steering angle and Δv (t) for the speed. The linear combination takes place with variable coefficients, which form a positioning matrix -S (t). The positioning matrix is represented by block 143 . The actuation signals Δα (t) and Δv (t) obtained with the actuation matrix -S (t) control actuators 144 and 146, similar to the actuators 106 and 112 of FIG. 9.

The control signal Δα (t) is superimposed on an initial signal α (0) by an input 150 in a summing point 148 . This results in a first command variable α target (t) for the steering angle α. This reference variable is compared in a summing point 152 with a feedback signal α (t), which is supplied by an encoder 154 (corresponding to encoder 38 from FIG. 3) via a feedback loop 156 . The signal α (t) from the transmitter 154 is again sent to the navigation computer 30 .

Similarly, the control signal Δv (t) is superimposed on an initial signal v (0) by an input 160 in a summing point 158 . This results in a second reference variable v des (t) for the speed v. This reference variable is compared in a summing point 162 with a feedback signal v (t), which is supplied by an encoder 164 (corresponding to encoder 36 from FIG. 3) via a feedback loop 166 . The signal v (t) from the transmitter 164 is also sent to the navigation computer 30 .

The adjustment matrix -S (t), represented by block 143 , is influenced by an adaptation unit 168 . The adaptation unit 168 receives the heading angle ψ (t) from the output 70 of the navigation computer 30 , the coordinates y (t) and x (t) from the output 78 of the navigation computer 30 , the turning angle α (t) from the transmitter 154 and the speed v (t ) from encoder 164 .

The adaptation unit 168 influences the elements of the positioning matrix in such a way that there is always an acceptable or favorable behavior of the control loop, in each case adapted to the track and the driving state of the vehicle 10 . The adaptation unit 168 is shown schematically in FIG. 2. It contains a model control loop 170 , which simulates the control loop described with reference to FIG. 11 in the vicinity of the reference path or target path. It can be shown that such a control loop has the structure shown in FIG. 12: A manipulated variable vector Δu (t) is multiplied by a variable matrix C (t). This matrix is represented by block 172 in FIG . A feedback signal (also in the form of a vector) is "connected" to the vector obtained in a "summing point" 174 . The difference vector obtained is integrated. This is represented by block 176 . An initial value of the vector Δx (0) describing the state of the controlled system is entered at an input 178 . The integration provides a vector Δx (t) . The vector Δx (t) is fed back in two ways: The vector Δx (t) is multiplied by a matrix B (t). The matrix B (t) is represented by a block 178 . The vector thus obtained is fed back to the summing point 174 in an "inner" feedback loop 180 as said feedback signal. Furthermore, the vector Ax (t) in an "external" feedback loop 182 with the actuating matrix -S (t), here represented by block 184, multiplied and delivers the manipulated variable vector .DELTA.u (t).

The actual position y (t), x (t) and the heading angle ψ (t) as well as the actual manipulated variables, the steering angle α (t) and the speed v (t) are fed to this model control loop. The model control loop describes the problem or domain knowledge. The model control loop 170 provides the two matrices C (t) and B (t).

The matrices C (t) and B (t) describe the controlled system. These matrices are fed to a knowledge processing system 186 . This knowledge processing system 186 contains rules and criteria for the design of the time-variant model control loop. These rules and criteria represent the problem solving knowledge. A positioning matrix S (t) is determined according to these rules and criteria, which leads to a solution that is favorable in terms of control technology or at least acceptable. The knowledge-processing system practically corresponds to the control expert, who, for a specific control system described by the matrices C (t) and B (t), designs the returns according to known rules of control technology in such a way that acceptable control behavior results. The rules and criteria need not necessarily be in the form of a mathematical quality criterion. It can be rule-based, symbolic algorithms or - in general - a combination of arithmetic and symbolic representations.

The guidance system, as described so far, is able to guide the vehicle along a straight path between two waypoints or on any curved, predetermined path. In a further development of this guidance system, a vehicle 10 of the type described is now to be able to find the way from a current location (a waypoint) to a destination (another waypoint) within a specific area of application. It is therefore not only a question of guiding the vehicle along certain lanes, but of selecting the predetermined lanes, usually a larger number of predetermined lane sections, so that a specific destination is reached. This is shown in Fig. 13 using a simplified example.

In Fig. 13 waypoints and spans between these waypoints are specified for a field of use. The waypoints in FIG. 13 are "1", "2", "3", "4", "5", "6", "7", "8",. . . Inscribed "N". The waypoints are possible locations and destinations. The track sections can be straight or curved, e.g. B. circular arc.

A driving order to the autonomous vehicle 10 now consists in getting from a location to a destination. Here, the vehicle 10 itself must find the most favorable sequence of waypoints to be traveled with reference to predetermined, permissible path sections which lead from the current location to the destination. For this purpose, suitable track sections must be called up in succession in a certain sequence from the track sections stored in the memory 34 ( FIG. 2). This can be done in the following ways:

A logical linking network can be derived from the geometric model of the area of use with the track sections and waypoints shown in FIG. 13, as shown in FIG. 14. The nodes in this network are the waypoints marked accordingly. The connections between the nodes are provided with a weighting which preferably corresponds to the distance of the associated path section between the waypoints represented by the nodes. The weight of the connection between the nodes or waypoints "i" and "k" is designated W i k .

To obtain the special route from a location to a destination, a "decision tree" is generated from this network for each target, which shows the alternative solutions. Such a decision tree is shown in FIG. 15 for a journey from waypoint "5" to waypoint "1". There are several alternative routes. The vehicle can drive from waypoint 5 "to the left" via waypoints "4", "3", "2" to waypoint "1" ( Fig. 13). The vehicle can also "around" ( Fig. 13) on waypoints "7", "8". . . "N" to waypoint "1". Finally, the vehicle 10 can drive directly to waypoint "1" via waypoints "6" and "2"". The track sections used in the various alternatives are now provided with the weights W i k mentioned. These weights can represent, for example, the lengths of the relevant track sections, and it can serve as a selection criterion that the sum of the weights should be as small as possible, that is, the total length of the track traveled as short as possible. Then the vehicle travels the shortest route to its destination. However, other criteria can also be used for the selection of the train used from the location ("5") to the destination ("1").

In the present case, the path is selected as the shortest route via waypoints "6" and "2". The time-dependent coordinates of the relevant track sections are fed to navigation computer 30 and steering computer 34 .

In the case of a large number of waypoints with a highly branched network of track sections, the steps for determining the cheap track can take a lot of computing time and can take a long time. An auto-associative memory is therefore provided in a further embodiment of the management system described. Such an auto-associative memory is shown schematically in FIG. 16 and labeled 188 . An auto-associative memory is a signal processing element that stores copies of different sets of input signals x (p) , with p = 1, 2, 3,. . . k. The auto-associative memory gives a certain sentence

x (r) = (. ξ₁ (r), ξ₂ (r),.. ξ n (r))

input signals stored in a learning phase, where rεp, to an output as soon as an input signal set

x = (ξ₁, ξ₂,... ξ n )

is present, of which a specified subset of the values ξ i matches a corresponding subset ξ i (r) from x (r) . In FIG. 16, x denotes the sets of input signals and y denotes the sets of output signals of the associative memory 188 . Then it applies

y = (η₁, η₂,... η m ) = (ξ₁ (r) , ξ₂ (r) ,... ξ n (r) ),

if

If this associative memory now the found tracks for the respective locations and Driving destinations are communicated, the associative receives Stores each a sequence of routes that drive through need to be from the location to the destination to get. Over time, this memory learns all possible connections of the area of application.

In Fig. 17 the function of the associative memory is explained using the example discussed above. In the "learning phase" the associative memory 188 already contains the possible sequences of waypoints and path sections, namely "5", "6", "2", "1" and "5", "4", "3", " 2 "," 1 "saved. If waypoints "5" and "1" are now applied to the input of the associative memory, then associative memory 188 links ("associates") these elements (waypoints) with stored complete input vectors, namely the sequences specified above. These stored input vectors are then output, as indicated in FIG. 17. These episodes are then immediately available as possible lanes. The sequences of waypoints thus obtained are then used, as explained in connection with the decision tree of FIG. 15, to select the cheapest path. The creation of the lanes by an auto-associative memory works very quickly. After a certain learning phase, the auto-associative memory 188 immediately delivers the cheapest train from an entered location to a likewise entered destination, without having to use the logical connection network of FIG. 14 or the decision tree of FIG. 15. The learning phase can also be carried out off-line using a sample and simulated trips.

The processes described are implemented as an intelligent signal processing unit. This is shown in Fig. 18. The signal processing unit 190 receives a geometric model of the area of use, as shown in FIG. 13. This is indicated by block 192 in FIG. 18. A logical link network according to FIG. 14 results from the geometric model of the area of use . This is indicated in FIG. 18 by block 194 . The signal processing unit 190 is input via input 196, the location of the vehicle 10 (eg "5"). A decision tree according to FIG. 15 is then formed in the manner already described, and the cheapest, z. B. shortest path selected. This is represented by block 198 . The algorithms to be processed are predominantly symbolic in nature (rule-based, search techniques), but sometimes also arithmetic.

The cheapest train is output at an exit 200 . At the same time, the determined path is stored in the auto-associative memory 188 ( FIGS. 16 and 17).

The location and destination of the vehicle 10 are each simultaneously input to the auto-associative memory 188 via input 202 . If the path between the two entered waypoints has already been stored in this memory from an earlier process, the auto-associative memory 188 outputs the path directly at an exit 204 .

In a further development of this process, the first three stages 192 , 194 and 198 of signal processing are only part of a simulation setup. They train the auto-associative memory as a "teacher" during a learning phase and let it "learn" all possible paths. This happens automatically by running through a large number of destinations and the generation of the lanes in the simulation, controlled by random generators, in a relatively short time. Then only the auto-associative memory 188 is actually implemented in the vehicle 10 .

Claims (10)

1. Guide system for guiding an unmanned, steerable vehicle along a predetermined path, comprising
  • (a) a memory ( 32 ) provided in the vehicle ( 10 ) in which the predetermined path is stored,
  • (b) an autonomous navigation system ( 30, 36, 38 ) which is arranged in the vehicle ( 10 ) and which contains an angle transmitter ( 38 ) for determining the turning angle (α (t)) of at least one vehicle wheel ( 16, 18 ) and signal processing means , to which a steering angle signal from the angle sensor ( 38 ) and a speed signal (v (t)) from a running speed sensor ( 36 ) are applied, for determining the heading angle (ψ (t)) and which of the position (x (t), y (t)) of the vehicle using the dead reckoning method,
  • (c) a controller ( 34 ), which responds to the deviations of the position (x (t), y (t)) provided by the navigation system ( 30, 36, 38 ) from the stored path and an actuating signal to correct this deviation supplies,
  • (d) actuators ( 24 ) which are actuated by the actuating signal of the controller ( 34 ) for guiding the vehicle ( 10 ) on the predetermined path,
  • (e) means for supporting the navigation system with the help of fixed marks ( 80, 84 ) and
  • (f) means ( 82 ) for scanning these marks ( 80, 84 ),
    characterized in that
  • (f) the stationary marks ( 80, 84 ) are attached to the floor at discrete waypoints of the predetermined path,
  • (g) the vehicle ( 10 ) has means for determining the linear offset of the vehicle from the marks ( 80, 84 ).
2. Guide system according to claim 1, characterized in that the navigation system further contains means ( 98 ) for correcting the navigation parameters in accordance with the storage determined by the brands ( 80, 84 ).
3. guidance system according to claim 1, characterized in that the signal processing means
  • (a) means ( 40 ) for generating a signal representing the course angle change rate ((t)) from the turning angle (α (t)) and the speed (v (t)) and
  • (b) contain means ( 42 ) for integrating the signal representing the course angle change rate over time to generate a course angle signal (ψ (t)).
4. guidance system according to claim 3, characterized in that
  • (a) the vehicle ( 10 ) has a first pair of coaxial, free-running wheels ( 12, 14 ) arranged laterally at a greater distance from one another,
  • (b) the vehicle ( 10 ) further has at least one third wheel ( 16 ) which is arranged near the longitudinal central axis ( 20 ) of the vehicle ( 10 ) at a distance (A) from the first pair of wheels ( 12, 14 ),
  • (c) the third wheel ( 16 ) is steerable by the steering means ( 24 ) and driven by drive means,
  • (d) on the third wheel ( 16 ) the running speed sensor ( 36 ) and the angle encoder ( 38 ) are provided and
  • (e) the signal processing means the signal reflecting the course angle change rate according to the relationship where A is the distance between the third wheel ( 16 ) and the pair of free-running wheels ( 12, 14 ), v (t) is the running speed measured on the third wheel and α (t) is the turning angle of the third wheel ( 16 ) is.
5. Guiding system according to claim 4, characterized in that the third wheel ( 16 ) forms a second pair of coaxial wheels with a fourth wheel ( 18 ) arranged closely next to it and connected thereto, which is about a in the longitudinal center plane ( 20 ) of the vehicle ( 10 ) lying axis ( 22 ) is pivotable for steering purposes.
6. Guide system according to claim 4, characterized in that
  • (a) Means for forming position and course deviation signals are provided as differences between the position or course signals calculated by the navigation system from the steering angle and speed and the setpoint values of position or course stored by the memory ( 32 ),
  • (b) the position and course deviation signals are connected to the controller ( 34 ) and the controller ( 34 ) control signals for the actuators ( 24 ) and the drive means as a linear combination of the position deviation signals (Δy (t), Δx (t)) and forms the course deviation signals (Δψ (t)) in accordance with an actuating matrix (-S (t)) which are connected to steering or speed actuators ( 144, 146 ),
  • (c) the elements of the positioning matrix (-S (t)) can be changed by an adaptation unit ( 168 ) and
  • (d) on the adaptation unit ( 168 ) the course and position signals (ψ (t), y (t), x (t)) as well as the steering angle (α (t)) and the running speed (v (t)) of the vehicle ( 10 ) are connected so that the positioning matrix (-S (t)) is changed in accordance with these signals.
7. guidance system according to claim 6, characterized in that
  • (a) the adaptation unit ( 168 ) contains a time-variant model control loop ( 170 ) which simulates the dynamic behavior of the management system, this model representing the problem or domain knowledge, and
  • (b) the adaptation unit ( 168 ) also contains a knowledge-processing system ( 186 ), in which rules and criteria for the design of the time-variant model control circuit ( 170 ) are stored, according to which an acceptable or favorable design of the time-variant actuating matrix (- S (t)) is determined.
8. guidance system according to one of claims 1 to 7, characterized by
  • (a) an intelligent signal processing unit ( 190 ),
    • - into which a geometric model of an area of application ( FIG. 13) with waypoints ("1", "2",... "N") and web sections connecting these can be entered,
    • - Which, according to this geometric model, generates a logical link network ( FIG. 14), which contains the possible connections between the waypoints and weights (W i k ) for the intermediate track sections (block 194 )
    • - Which, after entering a location and a destination from the linking network ( FIG. 14), generates a “decision tree” ( FIG. 15), which determines the alternatively possible paths between the location and destination in the application area according to the geometric model ( FIG. 13) and
    • - Which selects an optimal path from the possible paths determined in this way on the basis of the weights and connects it to the controller ( 34 ) as a predetermined path (block 198 ).
9. guidance system according to claim 8, characterized in that
  • (a) the signal processing unit ( 190 ) contains an auto-associative memory ( 188 ), in which the tracks determined for the entered locations and destinations can be stored, and
  • (b) the locations and destinations entered during use can also be connected to the auto-associative memory ( 188 ), so that the latter immediately outputs the associated path if it already exists in the memory ( 188 ) from an earlier journey of the vehicle ( 10 ) is stored here.
10. guidance system according to one of claims 1 to 9, characterized in that the memory ( 32 ) is an auto-associative memory in which the optimal tracks for entered locations and destinations can be stored, the auto-associative memory in a learning process Simulated locations and destinations supplied by random generators were loaded with the paths using a geometric model ( FIG. 13), a logical link network derived therefrom ( FIG. 14) and a decision tree ( FIG. 15).
DE19893923458 1989-07-15 1989-07-15 Guiding system for guiding an unmanned, steerable vehicle along a predetermined path Expired - Fee Related DE3923458C2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
DE19893923458 DE3923458C2 (en) 1989-07-15 1989-07-15 Guiding system for guiding an unmanned, steerable vehicle along a predetermined path

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
DE19893923458 DE3923458C2 (en) 1989-07-15 1989-07-15 Guiding system for guiding an unmanned, steerable vehicle along a predetermined path

Publications (2)

Publication Number Publication Date
DE3923458A1 DE3923458A1 (en) 1991-01-24
DE3923458C2 true DE3923458C2 (en) 1995-03-16

Family

ID=6385126

Family Applications (1)

Application Number Title Priority Date Filing Date
DE19893923458 Expired - Fee Related DE3923458C2 (en) 1989-07-15 1989-07-15 Guiding system for guiding an unmanned, steerable vehicle along a predetermined path

Country Status (1)

Country Link
DE (1) DE3923458C2 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6301532B1 (en) 1998-12-17 2001-10-09 Daimlerchrysler Ag Method for correction of a signal of at least one sensor on the basis of which a path curve can be determined, on which the vehicle is moving

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
FR2936324B1 (en) 2008-09-19 2010-10-15 Balyo Systems Sarl Navigation method and system implementing such a method
US10403139B2 (en) 2017-09-20 2019-09-03 Ford Global Technologies, Llc Local navigation system for vehicle navigation

Family Cites Families (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE2545025B2 (en) * 1975-10-08 1980-09-11 Bodenseewerk Geraetetechnik Gmbh, 7770 Ueberlingen
DE2659094C3 (en) * 1976-12-27 1989-09-21 Bodenseewerk Geraetetechnik Gmbh, 7770 Ueberlingen, De
US4144571A (en) * 1977-03-15 1979-03-13 E-Systems, Inc. Vehicle guidance system
GB8313338D0 (en) * 1983-05-14 1983-06-22 Gen Electric Co Plc Vehicle control
JPH0434773B2 (en) * 1983-11-24 1992-06-09 Toyoda Chuo Kenkyusho Kk
US4716530A (en) * 1984-05-21 1987-12-29 Kabushiki Kaisha Meidensha System for automatically controlling movement of unmanned vehicle and method therefor
DE3427020C2 (en) * 1984-07-21 1990-05-03 Messerschmitt-Boelkow-Blohm Gmbh, 8012 Ottobrunn, De
GB8501012D0 (en) * 1985-01-16 1985-02-20 Gen Electric Co Plc Automated vehicle drift correction
NL8500529A (en) * 1985-02-25 1986-09-16 Ind Contractors Holland Bv System for determining the position of a vehicle not bonded to a fixed track.
DE3538908A1 (en) * 1985-11-02 1987-05-21 Holzapfel Wolfgang Prof Dr Ing Autonomous on-board locating system for determining the position and protecting against collision of robot and industrial trucks
US4817000A (en) * 1986-03-10 1989-03-28 Si Handling Systems, Inc. Automatic guided vehicle system
US4809178A (en) * 1986-05-22 1989-02-28 Kabushiki Kaisha Toyota Chuo Kenkyusho Obstacle data processing system for unmanned vehicle
DE3802337C1 (en) * 1988-01-27 1989-07-13 Messerschmitt-Boelkow-Blohm Gmbh, 8012 Ottobrunn, De

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6301532B1 (en) 1998-12-17 2001-10-09 Daimlerchrysler Ag Method for correction of a signal of at least one sensor on the basis of which a path curve can be determined, on which the vehicle is moving

Also Published As

Publication number Publication date
DE3923458A1 (en) 1991-01-24

Similar Documents

Publication Publication Date Title
Hentschel et al. Autonomous robot navigation based on openstreetmap geodata
Milanés et al. Autonomous vehicle based in cooperative GPS and inertial systems
Guo et al. Design of automatic steering controller for trajectory tracking of unmanned vehicles using genetic algorithms
EP2818954A2 (en) Driverless transport vehicle, method for planning a virtual track and method for operating a driverless transport vehicle
US5545960A (en) Autonomous mobile machine, and system and method for controlling a mobile machine
Zelinsky et al. Planning paths of complete coverage of an unstructured environment by a mobile robot
US8260483B2 (en) Guidance, navigation, and control system for a vehicle
Lee et al. Fuzzy motion planning of mobile robots in unknown environments
CN101326425B (en) Speed control method for vehicle approaching and traveling on a curve
Bevly et al. Cascaded Kalman filters for accurate estimation of multiple biases, dead-reckoning navigation, and full state feedback control of ground vehicles
CN101793528B (en) System and method of lane path estimation using sensor fusion
JP6712088B2 (en) Method, system and non-transitory computer readable storage medium for controlling a vehicle
Durrant-Whyte et al. Localization of autonomous guided vehicles
FI115668B (en) Initialization of position and direction of mining vehicles
CA1269740A (en) Automated vehicle drift correction
Sasiadek et al. Sensor data fusion using Kalman filter
WO2017104775A1 (en) Method for controlling vehicle and control system of vehicle
CN100568144C (en) Mobile robot's multirow is for merging automatic navigation method under a kind of circumstances not known
Garimort et al. Humanoid navigation with dynamic footstep plans
US20120046819A1 (en) System for integrating dynamically observed and static information for route planning in a graph based planner
Madhavan et al. Distributed cooperative outdoor multirobot localization and mapping
US20120179322A1 (en) System and method for autonomous navigation of a tracked or skid-steer vehicle
JP2609846B2 (en) Operation control device and operation method for unmanned self-propelled vehicle
Lee et al. Navigation of automated guided vehicles using magnet spot guidance method
EP0788044A1 (en) An automated guided vehicle navigation steering control system

Legal Events

Date Code Title Description
OM8 Search report available as to paragraph 43 lit. 1 sentence 1 patent law
8110 Request for examination paragraph 44
D2 Grant after examination
8364 No opposition during term of opposition
8339 Ceased/non-payment of the annual fee