DE102018214828A1 - Method, device and computer program for predicting a location of a plurality of objects - Google Patents

Method, device and computer program for predicting a location of a plurality of objects

Info

Publication number
DE102018214828A1
DE102018214828A1 DE102018214828.7A DE102018214828A DE102018214828A1 DE 102018214828 A1 DE102018214828 A1 DE 102018214828A1 DE 102018214828 A DE102018214828 A DE 102018214828A DE 102018214828 A1 DE102018214828 A1 DE 102018214828A1
Authority
DE
Germany
Prior art keywords
objects
fictitious
speed
depending
determined
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.)
Pending
Application number
DE102018214828.7A
Other languages
German (de)
Inventor
Markus Spies
Vishnu Suganth Prabhakaran
Johannes Maximilian Doellinger
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.)
Robert Bosch GmbH
Original Assignee
Robert Bosch 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 Robert Bosch GmbH filed Critical Robert Bosch GmbH
Priority to DE102018214828.7A priority Critical patent/DE102018214828A1/en
Publication of DE102018214828A1 publication Critical patent/DE102018214828A1/en
Pending legal-status Critical Current

Links

Images

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
    • G05D1/0212Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory
    • G05D1/0221Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory involving a learning process
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01CMEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00Navigation; Navigational instruments not provided for in preceding groups G01C1/00-G01C19/00
    • G01C21/26Navigation; Navigational instruments not provided for in preceding groups G01C1/00-G01C19/00 specially adapted for navigation in a road network
    • G01C21/28Navigation; Navigational instruments not provided for in preceding groups G01C1/00-G01C19/00 specially adapted for navigation in a road network with correlation of data from several navigational instruments
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01CMEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00Navigation; Navigational instruments not provided for in preceding groups G01C1/00-G01C19/00
    • G01C21/26Navigation; Navigational instruments not provided for in preceding groups G01C1/00-G01C19/00 specially adapted for navigation in a road network
    • G01C21/34Route searching; Route guidance
    • G01C21/36Input/output arrangements for on-board computers
    • G01C21/3602Input other than that of destination using image analysis, e.g. detection of road signs, lanes, buildings, real preceding vehicles using a camera
    • 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
    • G05D1/0274Control of position or course in two dimensions specially adapted to land vehicles using internal positioning means using mapping information stored in a memory device
    • GPHYSICS
    • G08SIGNALLING
    • G08GTRAFFIC CONTROL SYSTEMS
    • G08G1/00Traffic control systems for road vehicles
    • G08G1/065Traffic control systems for road vehicles by counting the vehicles in a section of the road or in a parking area, i.e. comparing incoming count with outgoing count
    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05DSYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D2201/00Application
    • G05D2201/02Control of position of land vehicles
    • G05D2201/0217Anthropomorphic or bipedal robot

Abstract

The invention relates to a method for predicting a location of a plurality of objects. A map is provided that includes an action space of the objects and is divided into a plurality of cells. A first variable is assigned to each cell, which characterizes a probability whether at least one of the objects is located in the respective cell. A plurality of fictitious objects is assigned a weight depending on that first size, that of the cell within which the respective fictitious object is placed. A machine learning system determines a plurality of different discrete speeds depending on the card provided. The fictional objects are repositioned depending on one of the discrete speeds. The first size of the cells is determined depending on the weights of those fictitious objects that are placed within the respective cell. The invention further relates to a computer program and a device for executing the method and a machine-readable memory element on which the computer program is stored.

Description

  • Technical field
  • The invention relates to a method for predicting a location of a plurality of objects by means of a machine learning system. The invention also relates to a device and a computer program, each of which is set up to carry out the method.
  • State of the art
  • The unpublished DE 10 2017 217412.9 discloses methods of operating a robot control system that includes a machine learning system. Depending on a map which represents an action space of the robot, the machine learning system determines a course of movement of at least one object in the action space of the robot.
  • The unpublished DE 10 2018 201570.8 discloses a multiple target object tracking method. First, a speed distribution is determined using a speed transition distribution. Thereupon, a transition probability information is calculated, which indicates for objects a probability of reaching the possible neighboring raster cells. Then an occupancy probability is determined for each raster cell on the basis of the transition probability information.
  • Disclosure of the invention
  • In a first aspect, a, in particular computer-implemented, method for predicting a respective location, in particular a course of movement, of a plurality of objects is presented. First, a map is provided that includes an action space of the objects and is divided into a plurality of cells. A first variable is assigned to each cell, which characterizes, in particular a probability, whether at least one of the objects is located within the respective cell. A plurality of fictitious objects are placed on the map and each fictitious object is assigned a weight depending on the first size, that of the cell within which the respective fictitious object is placed. A machine learning system then determines a plurality of different discrete speeds, in particular for each cell, depending on the map provided, in particular a geometry and / or outline and / or shape of the action space. A new position of the fictitious objects is then determined depending on a selected discrete speed of the majority of the different discrete speeds and the fictitious objects are placed on the map based on the new position. The first size of the cells is determined depending on the weights of those fictitious objects that are placed within the respective cell.
  • The method has the advantage that it can be carried out with an arbitrary and unknown number of objects and their locations can be predicted at the same time, since the respective objects do not have to be assigned to a data point, such as a cell or a fictitious object. Furthermore, the method has the advantage that knowledge about the interaction of the objects with the environment that has been learned through the machine learning system can be used for prediction. For example, the machine learning system can be used to derive a movement of the objects from the shape of the action space, which is then used to determine the new positions of the fictitious objects. Predicting the location is understood to mean that the location is predicted at a subsequent point in time.
  • It is proposed that, in particular randomly and repeatedly, the fictitious objects are selected depending on the assigned weights and the selected fictitious objects are placed within the respective cell within which the selected fictitious objects are placed.
  • The advantage of this is that by selecting the fictitious objects depending on their weights, the fictitious objects are distributed on the map in such a way that it is expected that the objects are arranged. This allows a resource-efficient estimate of the current arrangement of the objects. Furthermore, this procedure has the advantage that several modalities of a non-deterministic dynamic can be mapped. Therefore, there is no need to make a decision using the fictitious objects, e.g. At a crossroads, which path is chosen, but the fictitious objects are divided according to probabilities and random initialization along the possible paths. The selection is preferably realized by pulling with replacement according to an urn model, the probabilities that the fictitious object is dragged depending on the weights of the fictitious objects.
  • It is further proposed that the machine learning system determine a plurality of probabilities for the different discrete speeds, in particular for each discrete speed of the fictitious objects from a previous point in time.
  • It is also proposed that the first sizes of the cells depend on a product the weights of the fictitious objects that are placed within the respective cell are determined.
  • It is advantageous here that the first variables can be determined in a computationally and time-efficiently manner after the fictitious objects have been placed in their new positions, the non-deterministic behavior of real objects being modeled as well.
  • It is further proposed that a measurement, in particular a detection of the objects, be carried out, a plurality of the first sizes of the cells being determined, in particular adapted, as a function of the measurement.
  • The advantage is that information from the environment flows in through the measurement, which means that the first values can also be adjusted depending on the measurement. The measurement can have a corrective effect on the method and thus increase the accuracy of the method. Since the method can be initialized randomly, the real arrangement of the objects can be determined more reliably on the basis of the measurements.
  • It is also proposed that the machine learning system be a folding neural network (Convolutional Neural Network). The machine learning system determines a plurality of the different discrete speeds, in particular a speed distribution, for each cell.
  • A speed distribution is understood to mean that the different discrete speeds are each assigned a probability which indicates how likely it is that an object within this cell has the respective discrete speed. The advantage here is that a precise determination of the selected speeds for the fictitious objects can be carried out and the robustness of the method can be increased.
  • It is also presented that a control variable for controlling a robot is determined as a function of the first sizes of the cells. Alternatively, the control variable can be used to control an actuator of a technical system. The technical system can be, for example, an at least partially autonomous machine, an at least partially autonomous vehicle or a flying object, such as a drone.
  • In a further aspect, a computer program is proposed. The computer program is set up to carry out one of the aforementioned methods. The computer program comprises instructions which cause a computer to carry out one of these methods with all of its steps when the computer program runs on the computer. A machine-readable memory module is also proposed, on which the computer program is stored. Furthermore, a device is proposed which is set up to carry out one of the methods.
  • Embodiments of the above-mentioned aspects are shown in the accompanying drawings and explained in more detail in the following description. Show:
  • Figure list
    • 1 a schematic representation of a robot;
    • 2A a schematic representation of a map of the room in which the robot and two objects are located;
    • 2 B a schematic representation of the map of the room, the map being divided into cells;
    • 3 a schematic representation of an embodiment of the method for predicting a location of objects;
  • 1 shows a schematic representation of an at least partially autonomous robot, which in a first exemplary embodiment is provided by an at least partially autonomous vehicle ( 10 ) given is. In a further exemplary embodiment, the at least partially autonomous robot can be a service, assembly or stationary production robot, alternatively an autonomous flying object, such as a drone.
  • The at least partially autonomous vehicle ( 10 ) a registration unit ( 11 ) include. The registration unit ( 11 ) can be, for example, a camera that captures an environment of the vehicle ( 10 ) detected, or be a radar, which distances to objects in the vicinity of the vehicle ( 10 ) detected. In a further embodiment, the detection unit ( 11 ) be a depth sensor. The registration unit ( 11 ) can with a prediction module ( 12 ) be connected. The prediction module ( 12 ) determined as a function of a card provided and preferably as a function of an input variable provided, for example provided by the acquisition unit ( 11 ), an output variable, compare to this 3 , The output variable characterizes, for example, the location at which objects are located in the vicinity of the vehicle ( 10 ) at a subsequent time. The output variable can then be sent to a control unit ( 13 ) to get redirected.
  • The control unit ( 13 ) controls depending on the output of the prediction module ( 12 ) an actuator, preferably controls the actuator such that the vehicle ( 10 ) performs a collision-free maneuver. In the first exemplary embodiment, the actuator can be an engine or a braking system of the vehicle ( 10 ) his. Alternatively, depending on the output size of the prediction module ( 12 ) a trajectory of the vehicle ( 10 ) or the objects can be determined.
  • Furthermore, the vehicle ( 10 ), especially the semi-autonomous robot, a computing unit ( 14 ) and a machine-readable storage element ( 15 ). On the storage element ( 15 ) A computer program can be stored which comprises commands which are executed on the computing unit when the commands are executed ( 14 ) cause the computing unit ( 14 ) the method for predicting the whereabouts of objects such as in 3 shown, executes. It is also conceivable that a download product or an artificially generated signal, which can each comprise the computer program, after being received by a receiver of the vehicle ( 10 ) the computing unit ( 14 ) cause the procedure to be carried out.
  • In an alternative embodiment, the prediction module ( 12 ) with the registration unit ( 11 ) as connected in the first embodiment, can be used for building control. User behavior is recorded by means of a sensor, for example a camera or a motion detector, and the control unit controls, for example, an automated door depending on the output of the prediction module ( 12 ). The prediction module ( 12 ) can then be set up, based on the recorded user behavior, to predict where the user will be moving in order to specifically control the door, for example. Alternatively, the prediction module ( 12 ) are used in a production hall to monitor that no one gets too close to a production machine and switch it off if necessary.
  • In a further exemplary embodiment, the prediction module ( 12 ) are used for an assistance system, for example a parking assistance system, which depends on the prediction module ( 12 ), which predicts the movement of the objects in the area, performs a parking maneuver. It is conceivable that instead of the registration unit ( 11 ) In this exemplary embodiment, stationary sensors are arranged in the vicinity of the assistance system.
  • The prediction module ( 12 ) can comprise a deep neural network, in particular a convolutional neural network. For example, the deep neural network can be the learned deep neural network from one of the documents mentioned at the beginning.
  • 2A shows a schematic representation of a map ( 20th ), which in this embodiment represents a floor plan of a room. There are two objects on the map ( P1 . P2 ), which can move, for example, in the direction of the door shown at the top right. Furthermore, a robot ( R ) which shows the at least partially autonomous vehicle ( 10 ) to 1 can be. The prediction module ( 12 ) can, for example, depending on the card ( 20th ) the whereabouts of the objects ( P1 . P2 ) predict at a subsequent point in time.
  • 2 B shows a schematic representation of the map ( 20th ) that are in a plurality of cells ( 21 ) is divided. As in 2 B shown is the cells ( 21 ) one probability of occurrence ( 22 ) assigned. The probability of occurrence ( 22 ) characterizes how likely it is that at least one of the two objects ( P1 . P2 ) is inside the respective cell. Since the object ( P2 ) located in the spatial vicinity of the narrowing in the middle of the room is that cell, a probability of occurrence with the value 0.9 assigned. For example, due to a clumsily chosen division of the cells, the neighboring cell underneath may have a probability of occurrence with the value 0.67 having.
  • It should be noted that the choice of the cell size, in particular the cell shape (for example round or triangular) can be chosen as desired. It should also be noted that the card ( 20th ) alternatively or additionally, can represent geographical conditions, such as, for example, road layouts or building arrangements.
  • 3 shows a schematic representation of the method ( 30th ) to predict the whereabouts of the objects ( P1 . P2 ), especially on the map ( 20th ).
  • The procedure ( 30th ) starts with step 31 , In step 31 will a card ( 20th ), which maps, for example, an environment of the robot. The map ( 20th ) is then in cells ( 21 ) disassembled. Each cell is initially given an estimated or random probability of occurrence ( 22 ) assigned.
  • The probability of occurrence ( 22 ) can characterize whether one of the objects is within the respective cell or how likely it is that one of the objects is within the respective cell, compare this 2 B , Furthermore, in step 31 a predefinable number of fictitious objects are initialized. The predefinable number of fictitious objects is preferably approximately 1,000 to 1,000,000. Furthermore, each fictitious object can initially have a position on the Map and a speed can be assigned. The position and the speed can be assigned randomly, alternatively the speed can be initialized with 1 m / s.
  • After step 31 step follows 32 , In step 32 the fictional objects are placed on the map, particularly depending on their assigned position.
  • In the next step 33 the fictitious objects are each weighted. The respective weight is dependent on the probability of occurrence ( 22 ) that was assigned to the respective cell within which the respective fictitious object is located. For example, the weight of one of the fictitious objects can be the value of the probability of occurrence of the respective cell within which the fictitious object is located.
  • After completing the step 33 , will step 34 carried out. Depending on the weights of the fictitious objects, a predeterminable number of fictitious objects is drawn here by dragging and dropping, the number of fictitious objects appropriately equaling the number of initialized fictional objects. For example, if 1000 fictional objects are placed on the map and each of these fictional objects has a weight, 1000 fictional objects are randomly drawn from these 1000 fictitious objects with replacement, which may result in individual fictitious objects being drawn several times, especially if they are have a great weight. Then the drawn fictional objects are placed on the map. The placement is based on the position assigned to the fictitious objects. The drawn fictitious objects preferably take over the speed of the respective fictitious objects.
  • Then step 35 be performed. The weights of the fictitious objects are adjusted depending on the number of fictitious objects that are within the same cell. For example, each weight of the fictional objects that are all within the same cell can be divided by the number of fictional objects within that cell.
  • In the next step 36 the speed (v) assigned to the fictitious objects can be standardized. For example, the assigned speed can be a vector that is normalized depending on an amount of the vector. ν ˜ = 1 ν ν
    Figure DE102018214828A1_0001
  • This is followed by step 37 , in step 37 the card is taught to a machine learning system ( 20th ) is provided as an input variable. The learned machine learning system then determines, depending on the map, in particular the outline or geometry of the action spaces of the objects and / or the robot ( 10 ), a velocity distribution for each cell. The speed distribution can be, for example, a matrix, the dimensions of which each correspond to a direction in which the objects on the map move and a speed can be assigned to the columns and / or the rows. The entries in the matrix can indicate the probability with which the fictitious object has the respective direction at the respective speed. Then, depending on the determined speed distribution and the respectively assigned speed of the fictitious objects, each fictitious object is assigned an updated speed from the determined speed distribution, preferably drawn randomly (see equation 4 below).
  • It should be noted that the machine learning system can have been trained for this as in the documents mentioned at the beginning. It should also be noted that the machine learning system also determines the speed distribution as a function of the speed of the, in particular fictional, objects.
  • The assignment of the updated speed can be carried out, for example, using the following equations, the updated speed being drawn randomly according to equation 4: Δ ν ˜ = ν ˜ - ν i . N N
    Figure DE102018214828A1_0002
    u i = e x p ( - Δ ν ˜ l σ )
    Figure DE102018214828A1_0003
    P ( ν i . N N ) = u i i u i
    Figure DE102018214828A1_0004
    where the index i indicates the determined speeds of the machine learning system and σ preferably has a value of 0.3.
  • In step 38 a change in the speed, in particular an acceleration, is then determined as a function of the determined speed distribution. See for example DE 102018201570.8 , In this document, the acceleration is given as a function of a difference between at least two speed values from the speed distribution updated speed determined. Alternatively, an acceleration distribution can be derived from the speed distribution by using the differences to determine speeds from the respective direction of a probability distribution, which characterizes the probability with which the respective acceleration occurs. The change in speed can be drawn randomly from this acceleration distribution, for example.
  • Alternatively, the machine learning system can also be learned in such a way that it depends on the card ( 20th ) an acceleration distribution is determined, from which an acceleration is determined at a given updated speed.
  • After step 38 completed step can 39 consequences. Here the change in speed depends on the normalization of the speed with 1 v .
    Figure DE102018214828A1_0005
    like in step 36 shown, scaled.
  • In step 40 becomes dependent on the updated speed ( ν a k t ˜ )
    Figure DE102018214828A1_0006
    and the change in speed α determines a new speed v 'of the individual fictitious objects. This can be done using equation 5, for example: ν ' = ν a k t ˜ + a
    Figure DE102018214828A1_0007
  • Then in step 40 depending on the new speed (v ') of the fictitious objects, a new position (r') of the fictitious objects is determined. This can be done using equation 6, for example: r ' = r + ν '
    Figure DE102018214828A1_0008
    where r is the current position of the fictitious objects.
  • Then step 41 the occurrence probabilities of the cells depend on and the weights of those fictitious objects, which are located within the respective cells, updated. Example with the equation: P ( O c ) = 1 - p ( 1 - P ( O c p ) )
    Figure DE102018214828A1_0009
    in which P (O cp ) is the weight of the p-th fictional object in cell c. It should be noted that these weights in step 35 have been assigned to the fictional objects. P (O c ) is the probability of occurrence ( 22 ) the objects ( P1 . P2 ) of the cth cell.
  • In a subsequent step, an environment measurement can optionally be carried out, in which, for example, the positions of the objects are recorded. Depending on this environmental measurement, the probability of occurrence of the cells can be adjusted. This can be done using the following equation, for example: P ( O c ) = P ( Z c | O c ) P ( O c )
    Figure DE102018214828A1_0010
    with the determined occurrence probabilities from the environment measurement: P ( Z c = e.g. c | O c = O c ) = { 1 - w e.g. . i f e.g. c = O c w e.g. . e l s e
    Figure DE102018214828A1_0011
    in which w Z characterized a noise of the measurement and o c corresponds to a occupied cell with one of the objects ( P1 . P2 ).
  • This ends the process ( 30th ). The method ( 30th ) cyclically with step 33 started again. This can be done by dragging the fictional objects in step 34 their determined positions and above all their determined speeds are maintained. It should be noted that the sequence of steps can be carried out in any order imaginable.
  • Depending on the probability of occurrence of the cells, the position of the objects ( P1 . P2 ) can be determined at a subsequent point in time. For example, the probability of occurrence of individual cells is relatively high compared to their neighboring cells. It can be deduced from this that an object will be within these cells at the following point in time.
  • QUOTES INCLUDE IN THE DESCRIPTION
  • This list of documents listed by the applicant has been generated automatically and is only included for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.
  • Patent literature cited
    • DE 102017217412 [0002]
    • DE 102018201570 [0003, 0039]

Claims (17)

  1. Method (30) for predicting a location, in particular a course of movement, of a plurality of objects (P1, P2), a map (20) comprising at least one action space of the objects (P1, P2) and being divided into a plurality of cells ( 21) is provided, each cell (21) being assigned a first variable (22) which, in particular a probability (P (O c )), characterizes whether at least one of the objects (P1, P2) is within of the respective cell, a plurality of fictitious objects being placed on the map (20) and the fictitious objects each having a weight depending on the first size (22) which is assigned to the cell within which the respective fictitious object is placed A machine learning system determines a plurality of different discrete speeds depending on the map (20), a new position (r ') of the fictitious objects depending on a selected discrete The speed of the majority of the different discrete speeds is determined and the fictitious objects are placed on the map (20) on the basis of the new position (r '), the first size (22) of the cells depending on the weights of those fictitious objects within of the respective cell are determined.
  2. Method according to one of the preceding claims, wherein the location of the objects (P1, P2) is predicted as a function of the updated first sizes (22) of the cells.
  3. Method according to one of the preceding claims, wherein, in particular randomly and repeatedly, the fictitious objects are selected depending on the assigned weights and the selected fictitious objects are placed within the respective cell within which the selected fictitious objects were placed.
  4. Method according to one of the preceding claims, wherein each fictitious object is assigned a position (r) and a speed (v), a new speed (v ') is determined for each fictitious object depending on the assigned speed of the respective fictitious object and the selected discrete speeds, the new position (r ') of the fictitious objects being determined as a function of the new speed (v').
  5. Method according to one of the preceding claims, the first sizes of the cells being determined as a function of a product of the weights (P (O cp )) of the fictional objects which are placed within the respective cell.
  6. Method according to one of the preceding claims, wherein a measurement, in particular a detection of the objects, is carried out, a plurality of the first sizes (22) of the cells (21) being determined, in particular adapted, as a function of the measurement.
  7. Method according to one of the preceding claims, wherein the speed of the fictitious objects is a vector and the speeds depending on an amount ( 1 v )
    Figure DE102018214828A1_0012
    of the respective vector are normalized.
  8. Method according to one of the preceding claims, wherein a change in the speed, in particular an acceleration (a), is determined as a function of the plurality of different discrete speeds and the selected discrete speeds of the respective fictitious object for each fictitious object, the new position in each case (r ') of the fictitious objects is also determined depending on the change in speed.
  9. Procedure according to Claim 7 and 8th , the determined change in speed depending on the amount ( 1 v )
    Figure DE102018214828A1_0013
    of the speed vector of the respective fictitious object is scaled.
  10. Method according to one of the preceding claims, wherein the weights of those selected fictitious objects that are placed within the same cell are reduced depending on a number of these objects.
  11. A method according to any preceding claim, wherein the machine learning system is a convolutional neural network, the machine learning system each determining a plurality of different discrete speeds for each cell and each discrete speed depending on the previous discrete speed , assigns a probability of occurrence.
  12. Procedure according to one of the Claims 4 to 11 , differences (Δṽ) between the assigned speed and the determined discrete speeds are determined, the selected discrete speed of the fictitious objects depending on the difference between one of the determined discrete speeds being selected at random.
  13. Procedure according to one of the Claims 4 to 12 , wherein a speed of the objects is detected, the assigned speeds of the fictitious objects being determined as a function of the detected speed of the objects.
  14. Method according to one of the preceding claims, wherein a control variable for controlling a robot is determined as a function of the first sizes of the cells.
  15. Computer program which comprises instructions which, when executed by a computer, cause the computer to carry out the method according to one of the preceding claims.
  16. Machine-readable memory element on which the computer program according to Claim 15 is deposited.
  17. Device that is set up, the method according to one of the Claims 1 to 14 to execute.
DE102018214828.7A 2018-08-31 2018-08-31 Method, device and computer program for predicting a location of a plurality of objects Pending DE102018214828A1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
DE102018214828.7A DE102018214828A1 (en) 2018-08-31 2018-08-31 Method, device and computer program for predicting a location of a plurality of objects

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
DE102018214828.7A DE102018214828A1 (en) 2018-08-31 2018-08-31 Method, device and computer program for predicting a location of a plurality of objects
CN201910813290.4A CN110928292A (en) 2018-08-31 2019-08-30 Method, device and computer program for predicting a location of stay of a plurality of objects

Publications (1)

Publication Number Publication Date
DE102018214828A1 true DE102018214828A1 (en) 2020-03-05

Family

ID=69527354

Family Applications (1)

Application Number Title Priority Date Filing Date
DE102018214828.7A Pending DE102018214828A1 (en) 2018-08-31 2018-08-31 Method, device and computer program for predicting a location of a plurality of objects

Country Status (2)

Country Link
CN (1) CN110928292A (en)
DE (1) DE102018214828A1 (en)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102013203239A1 (en) * 2013-02-27 2014-08-28 Bayerische Motoren Werke Aktiengesellschaft grid-based prediction of the position of an object
DE102013223803A1 (en) * 2013-11-21 2015-05-21 Robert Bosch Gmbh Method and device for segmenting an occupancy grid for an environment model of a driver assistance system for a vehicle

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102013203239A1 (en) * 2013-02-27 2014-08-28 Bayerische Motoren Werke Aktiengesellschaft grid-based prediction of the position of an object
DE102013223803A1 (en) * 2013-11-21 2015-05-21 Robert Bosch Gmbh Method and device for segmenting an occupancy grid for an environment model of a driver assistance system for a vehicle

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
GINDELE, Tobias [et al.]: Bayesian occupancy grid filter for dynamic environments using pri-or map knowledge. In: 2009 IEEE Intelligent Vehicles Symposium, Xi'an, 2009, S. 669-676. - ISSN 1931-0587 *

Also Published As

Publication number Publication date
CN110928292A (en) 2020-03-27

Similar Documents

Publication Publication Date Title
US9934688B2 (en) Vehicle trajectory determination
DE102016211182A1 (en) A method, apparatus and system for performing automated driving of a vehicle along a trajectory provided from a map
Kahn et al. Uncertainty-aware reinforcement learning for collision avoidance
DE112017000787T5 (en) Method for controlling the movement of a vehicle and vehicle control system
US9858496B2 (en) Object detection and classification in images
Richter et al. Safe visual navigation via deep learning and novelty detection
Kim et al. Probabilistic vehicle trajectory prediction over occupancy grid map via recurrent neural network
Thrun Finding landmarks for mobile robot navigation
Tovar et al. Planning exploration strategies for simultaneous localization and mapping
Jin et al. Environmental boundary tracking and estimation using multiple autonomous vehicles
Scherer et al. River mapping from a flying robot: state estimation, river detection, and obstacle mapping
JP6586532B2 (en) Deep machine learning method and apparatus for robot gripping
Hülnhagen et al. Maneuver recognition using probabilistic finite-state machines and fuzzy logic
Joseph et al. A Bayesian nonparametric approach to modeling motion patterns
EP2167990B1 (en) Host-vehicle risk acquisition device and method
EP1901152B1 (en) Method, medium, and system estimating pose of mobile robots
US8706298B2 (en) Temporal tracking robot control system
US7765062B2 (en) Method and system for autonomous tracking of a mobile target by an unmanned aerial vehicle
US8554370B2 (en) Machine learning approach for predicting humanoid robot fall
Bautin et al. Inevitable collision states: a probabilistic perspective
CN105892471A (en) Automatic automobile driving method and device
O'Callaghan et al. Contextual occupancy maps using Gaussian processes
EP3106836B1 (en) A unit and method for adjusting a road boundary
Chatterjee et al. A neuro-fuzzy assisted extended Kalman filter-based approach for simultaneous localization and mapping (SLAM) problems
Hoermann et al. Dynamic occupancy grid prediction for urban autonomous driving: A deep learning approach with fully automatic labeling

Legal Events

Date Code Title Description
R163 Identified publications notified
R012 Request for examination validly filed