DE102015209101A1 - Method for generating a map for an autonomous vehicle - Google Patents

Abstract

A method of creating a map (30) for an autonomous vehicle (50), comprising the steps of: - detecting ground data by means of a ground radar device (10) disposed in the vehicle (50) during a defined movement of the vehicle (50) across the map (30) terrain to be imaged (40); - Simultaneous determination of environmental data by means arranged in the vehicle (50) surrounding radar device (20); and - creating the map (30) from the ground data and the surrounding data, wherein in a first trip location coordinates are recorded in a track framing the terrain in question.

Description

The invention relates to a method for producing a map for an autonomous vehicle. The invention further relates to a device for creating a map for an autonomous vehicle.

State of the art

Vehicles, mobile robots and highly automated or autonomous vehicles can determine their position by means of GPS (Global Positioning System) or by means of optical methods. For this purpose, a use of laser scanners of the environment is known.

Frank Demmerle: "A Biconical Multibeam Antenna for Space Division Multiple Access" in IEEE Transactions on Antennas and Propagation, Vol. 6, June 1998 discloses a biconical antenna.

Disclosure of the invention

An object of the invention is to provide an improved method for creating a map for an autonomous vehicle.

• – Ermitteln von Bodendaten mittels einer im Fahrzeug angeordneten Bodenradareinrichtung während eines definierten Bewegens des Fahrzeugs über ein in die Karte abzubildendes Gelände;
• – Gleichzeitiges Ermitteln von Umgebungsdaten mittels einer im Fahrzeug angeordneten Umgebungsradareinrichtung; und
• – Erstellen der Karte aus den Bodendaten und den Umgebungsdaten, wobei in einer Erstfahrt Ortskoordinaten in einer das betreffende Gelände einrahmenden Spur aufgenommen werden.

The object is achieved according to a first aspect with a method for producing a map for an autonomous vehicle, comprising the steps:

• Determining ground data by means of a ground radar device arranged in the vehicle during a defined movement of the vehicle over a terrain to be imaged into the map;
• - Simultaneous determination of environmental data by means arranged in the vehicle surrounding Radareinrichtung; and
• - Creating the map from the ground data and the surrounding data, wherein in a first trip location coordinates are recorded in a track framing the area in question.

• – das Fahrzeug mit:
• – einer Bodenradareinrichtung; und
• – einer Umgebungsradareinrichtung, wobei Daten der Bodenradareinrichtung und Daten der Umgebungsradareinrichtung zur Erstellung der Karte verwendbar sind, wobei in einer Erstfahrt des Fahrzeugs Ortskoordinaten eines in die Karte abzubildenden Geländes in einer das Gelände einrahmenden Spur aufnehmbar sind.

According to a second aspect, the object is achieved with a device for producing a map for an autonomous vehicle, comprising:

• - the vehicle with:
• A ground radar device; and
• An ambient radar device, wherein data of the ground radar device and data of the ambient radar device can be used to produce the map, wherein in a first run of the vehicle, location coordinates of a terrain to be imaged in the map can be recorded in a track framing the terrain.

In this way, data from two independent radar systems for creating a map for an autonomous vehicle are linked together. The vehicle registers in the first drive revolutions of its wheels or tracks and generates from it a location coordinate system. By linking a high accuracy and timeliness of the map material can be achieved, especially in the ground arranged, usually immovable, stationary objects contribute to the aforementioned accuracy. Advantageously, by means of the invention, a use of complex optical information is not required, which requires a considerably higher expenditure in data acquisition and evaluation.

The map created here is an ordered set of data that connects soil and environment properties with the location coordinates, and thus the location at which the soil and environment properties were measured. The location coordinates are introduced for the map making area in which the vehicle is preferably to be located. So a map here is an ordered feature catalog with soil and surrounding data, where each point of the map is linked to soil and environmental data measured at that point. Technically, these are stored in electronic form in a computer system.

As is common in radar systems, the data is the direction, amplitude, phase and transit time of the registered reflected electromagnetic waves emitted by the autonomous vehicle at its measurement position. In addition to the data, the path traveled from an assumed starting position to the measuring position, the assumed distance from the starting position and measuring position and the associated angle to the north direction, and the x and y coordinates of the measuring position are also recorded. The direction of the vehicle (from compass data the vehicle relative to north) is also stored.

The data of the ground radar include, as usual, the direction, amplitude, phase and transit time of the registered reflected electromagnetic waves emitted by the autonomous vehicle at its measuring position.

Advantageous developments of the method and the device are the subject of dependent claims.

An advantageous development of the method provides that the vehicle performs at least one learning run in an initialization phase. In this way, data from stationary objects in the ground can be entered into the map by means of the learning run, whereby a "primary echo map" which is as error-free as possible is provided.

A further advantageous development of the method provides that the vehicle is oriented by means of objects arranged in the ground. In this way, largely stationary, stationary objects are used for orientation of the vehicle in the field. In combination with the acquired environmental data, an improved orientation of the vehicle in the field can be provided in this way.

A further advantageous development of the method provides that at least one reference point is arranged in the card, wherein the reference point and the objects are slightly spaced from each other in relation to the overall dimension of the terrain. In this way, a kind of reference point plane is provided, whereby, based on the reference points from the vehicle, a search character drive with a defined track, e.g. a so-called "spiral travel" can be undertaken to determine whether or not the vehicle is near the reference point.

A further advantageous development of the method provides that triangulation is performed about a reference point by means of a defined closed movement pattern of the vehicle. In this way, an easy to be performed improvement of the map material can be achieved by means of a known trigonometric principle.

A further advantageous development of the method provides that at least one test run is carried out, in which it is checked whether the reference point is found by the vehicle. By the test drive can be advantageously verified whether the maps are still up to date.

An advantageous development of the method provides that during the test run, the soil data and the environmental data are compared with data of the map. This advantageously provides an efficient verification measure for the map data.

A further advantageous embodiment of the method provides that several test drives are performed at defined times. In this way, for example, it can be verified at regular intervals whether the map material is still up-to-date.

An advantageous development of the method provides that the test drives are carried out upon detection of changed environmental data. In this way, the timeliness of the map material can be further increased.

An advantageous development of the device provides that it further comprises a communication device for detecting additional data, for example weather data, wherein the additional data for the processing of the card can be used. In this way, further data can be used for optimizing the map material.

The vehicle runs from an order system from an appropriate lane, whose location is determined by a learning journey. During the learning journey, the vehicle picks up location and the reflection data of the radar waves with your positioning system and checks whether the quality of the data allows a meaningful evaluation in terms of data volume and signal-to-noise ratio. For example, if there is too little data, a second learning trip is taken in a different direction.

The invention will be described below with further features and advantages with reference to several figures in detail. The figures are mainly executed qualitatively and not necessarily true to scale. Same or functionally identical elements have the same reference numerals. The method is applicable to any kind of autonomous vehicles, such as robotic mowers, service robots for various purposes, etc.

In the figures shows:

1 an embodiment of an apparatus for creating a map for an autonomous vehicle;

2 a schematic diagram illustrating an operation of the method;

3 a plan view of a terrain for which a map for an autonomous vehicle is generated; and

4 a schematic representation of an embodiment of the method.

Embodiments of the invention

Conventional positioning systems have limited resolution accuracy. For example, GPS systems are often shadowed by objects, such as buildings or trees, so that it is often used only limited.

1 shows in two views an embodiment of a device 100 to create a map 30 a terrain 40 for an autonomous vehicle 50 , The device 100 includes the vehicle 50 on which a ground radar device 10 , preferably a broadband UWB radar system (Engl. Ultra Wide Band) and an environmental radar device 20 are arranged. The ambient radar device 20 preferably comprises a focusing biconical antenna, by means of which a 360 ° panoramic view with simultaneous direction information can be realized.

By means of ground radar device 10 can the vehicle 50 eg geochemical and structural information or data of the site 40 and to record soil topological information of its infrastructure, such as metal sleepers, cables, pipes and site-specific information of the absorption and reflection behavior of soil structures. In a learning or initialization run, the data of such objects become properties of assigned positions on a travel route of the vehicle 50 from the radar facilities 10 . 20 into the card 30 entered.

The map 30 is preferably within the vehicle 50 arranged, but may alternatively also be external to the vehicle 50 be arranged, in which case the radar devices 10 . 20 the collected data for the purpose of a subsequent readout can also save.

By means of the ambient radar device 20 At the same time, structure and distance information of walls and ceilings or other surrounding structures (not shown), such as trees or houses, can be detected and in the mentioned learning or initialization run as a property of the position on the travel path in the map 30 be deposited. Particularly suitable for this purpose are radar systems, as are also used in the field of autonomous driving.

It is done this way by means of the card 30 formed a terrain and movement model, where the "terrain model" represents the conditions in the ground and the "movement model" conditions above the ground. An adjustment of the two models mentioned is made possible in this way, whereby an improved position determination for the vehicle 50 is possible.

In particular, in the formation of the card 30 significant, not readily variable properties or objects deposited within the soil, such as in the form of stones, pipes, water-bearing veins in the ground, etc. exist and special places on the busy terrain 40 mark.

In the map 30 Furthermore, reference or calibration points can be created. As a result, for example, by means of a rotation of the vehicle 50 at one of said points about its axis an existing distinctive soil structure (for example, the direction of a cable) three-dimensional not only in depth, but also a surface orientation are detected. The said rotational movement can be detected by means of the ambient radar device 20 be combined in alignment at suitable locations, so that in addition to the vehicle 50 modeled rotation angle, a correction of the actual rotation to the modeled rotation angle is made possible. In this way, the modeled rotation angle of the vehicle 50 based on echo data of the ground radar device 10 Getting corrected.

For the device 100 is provided, in a learning phase to a calibration of the card 30 at excellent positions the vehicle 50 defined (eg triangular or quadrilateral) closed trajectories to drive a defined point. In this way, a unique soil profile can be recorded, which is compatible with the environmental data of the surrounding radar device 20 can be linked.

The said learning trip can be designed as an autonomous drive, but can alternatively also be guided by hand, for which purpose the vehicle 50 manually through the terrain to be explored 40 to be led.

In a subsequent autonomous regular operation of the autonomous vehicle 50 can also be in a large terrain 40 a position and direction calibration are performed by means of a more accurate local map 30 can be created around the mentioned reference points.

Likewise, in the device 100 the creation of "calibrated lanes" of the vehicle 50 possible by the ambient radar device 20 is used to generate at a certain distance to existing objects by means of a measurement of Radar echoes a distance function along a traveled distance. On the track, the ground information using the ground radar device 10 taken and together with the by the radar Radareinrichtung 10 determined echo information in the map 30 deposited.

In this way, the boundary of a defined area can be set with increased accuracy, even if partial information of the surroundings changes, as may be the case with passing vehicles, moving people, passing animals, etc. By means of the device 100 can in this way an accurate boundary information of the terrain 40 be determined so that the vehicle 50 If possible, do not cross over the determined limit.

Similarly, the vehicle 50 by defining calibration paths to the Reference points in the map 30 create a triangular-shaped, unique network without any time consuming new ground or environmental data in the map 30 must be deposited. This is especially in case of temporary changes in the environment of the vehicle 50 advantageous because otherwise always a complete learning or Initialisierungsfahrt would have to be covered.

The radar technology used as environmental sensor technology has the advantage over other, for example optical, technologies that it is almost weather-independent. By means of ground radar device 10 and the card 30 can be carried out in this way, even with temporarily variable conditions in the area a secure location, which can be classified even in permanently variable conditions of the environment better in a terrain model. This can be both the orientation and a classification of the terrain of the site 40 serve.

2 shows a principle of the operation of the method. Visible is a terrain 40 For example, in the form of a lawn, a golf course, a football field, etc. By means of Bodenradareinrichtung 10 at the vehicle 50 (not shown) scraps extending into the ground 11a ... 11n generated, eliminating the entire terrain 40 in a sense "deep-cut". In the case of detecting stationary objects 12 in the ground, corresponding echo data is added to the map 30 entered. The echo data of the objects 12 form in this way a kind of network of "ground reference points", which are for the ground radar device 10 are provided for guidance.

3 shows a top view of the terrain 40 on which the procedure for creating the map 30 for an autonomous vehicle 50 is carried out. The site 40 includes above-ground terrain structures 41 , such as trees, shrubs, house walls, etc. It also recognizes the fixed, in the ground of the area 40 arranged objects 12 ,

Also recognizable are several so-called "learning journey reference points" 60 which are created when said learning or initialization run is performed. Subsequently, the learning driving reference points 60 be identified and approached by data received from the environmental radar device 20 be determined.

It is suggested that the learning ride reference points 60 the objects 12 have defined as short distances and therefore are well suited to perform a triangulation process, with an orientation based on the fixed objects 12 is carried out. The closed driving patterns shown between the learning travel reference points 60 or between the objects 12 are merely exemplary and do not necessarily have to be triangles or squares.

It is conceivable for the device 100 Optionally, a communication device (not shown) that can interrogate and transmit additional data (eg weather data) wirelessly or by wire from the Internet, whereby the data of the card can be retrieved using this additional data 30 can be changed. This can be useful, for example, for heavy precipitation, if it is assumed that the objects 12 to a certain extent sinking in the softened ground, moving or changing their echo through a larger amount of water. Similarly, a ride may be initiated which indirectly measures the water content of the soil along a route through the garden, in the form of altered echoes or increasing reflection losses in the soil through the dielectric losses of the water, with water absorbing electromagnetic waves more than dry soil and the dielectric Properties of the soil changed. In this way, by means of the weather data readjustment of the card 30 be performed. Likewise, information about the soil dryness and possibly a necessary irrigation measure is obtained.

The readjustment of the card 30 may preferably be initiated and performed at defined time intervals to thereby synchronize the data from the environmental radar device 20 with the data of the ground radar device 10 to reach. This way is a permanent re-mapping of the map 30 supports, thereby reducing the map material of the card 30 always up to date. An initiation of such readjustment can be done by the vehicle 50 yourself or from an external user.

4 shows a basic sequence of an embodiment of the method for creating the map for an autonomous vehicle.

In a first step 200 is a determination of ground data by means of one in the vehicle 50 arranged Bodenradareinrichtung 10 while moving the vehicle 50 about one in the card 30 terrain to be imaged 40 carried out.

In a second step 210 is a simultaneous determination of environmental data by means of one in the vehicle 50 arranged ambient radar device 20 carried out.

In a third step 220 will eventually create a map 30 performed from the soil data and the environmental data, being in one First trip location coordinates are recorded in a track framing the terrain to be imaged.

In summary, the present invention proposes a method of constructing an autonomous vehicle map and an autonomous vehicle map device, which advantageously allow data from a ground radar to be functionally linked to ambient radar data, and in this way to provide a high quality electronic map for an autonomous vehicle. Attachment of landmarks is possible, but is not required in many terrain topologies. The information structures of the soil and the environment can also be used in partial changes in the soil and / or environment to update the terrain and movement model of the map.

As a result, the autonomous vehicle can determine its position in the field without the use of ambient optical information with an accuracy of about ten cm.

Although the invention has been disclosed by means of specific embodiments, it is by no means limited thereto. The person skilled in the art will thus initially also implement embodiments which are not disclosed or only partially disclosed, without departing from the essence of the invention.

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely 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.

Cited non-patent literature

• Frank Demmerle: "A Biconical Multibeam Antenna for Space Division Multiple Access" in IEEE Transactions on Antennas and Propagation, Vol. 6, June 1998 [0003]

Claims (11)

Method for creating a card ( 30 ) for an autonomous vehicle ( 50 ), comprising the steps: - determining ground data by means of an in-vehicle ( 50 ) arranged ground radar device ( 10 ) during a defined movement of the vehicle ( 50 ) via a in the map ( 30 ) to be imaged terrain ( 40 ); Simultaneous determination of environmental data by means of an in-vehicle ( 50 ) arranged ambient radar device ( 20 ); and - creating the map ( 30 ) from the ground data and the surrounding data, wherein in a first trip location coordinates are recorded in a track framing the terrain in question. Method according to claim 1, wherein the vehicle ( 50 ) performs at least one learning run in an initialization phase. Method according to claim 1 or 2, wherein the vehicle ( 50 ) by means of objects arranged in the ground ( 12 ) oriented. Method according to claim 3, wherein in the map ( 30 ) at least one reference point ( 60 ), the reference point ( 60 ) and the objects ( 12 ) in relation to the overall size of the site ( 40 ) are slightly spaced from each other. Method according to claim 4, wherein by means of a defined closed movement pattern of the vehicle ( 50 ) a triangulation around a reference point ( 60 ) is carried out. Method according to claim 5, wherein at least one test run is carried out in which it is checked whether the reference point ( 60 ) from the vehicle ( 50 ) Is found. Method according to claim 5 or 6, wherein in the test run the ground data and the surrounding data are combined with data of the map ( 30 ). The method of claim 6 or 7, wherein a plurality of test drives are performed at defined times. Method according to one of claims 6 to 8, wherein the test drives are performed upon detection of changed environmental data. Contraption ( 100 ) to create a map ( 30 ) for an autonomous vehicle ( 50 ), comprising: - the vehicle ( 50 ) with: - a ground radar device ( 10 ); and - an ambient radar device ( 20 ), whereby data of the ground radar device ( 10 ) and data of the environmental radar device ( 20 ) for creating the map ( 30 ) are used, wherein in a first trip of the vehicle ( 50 ) Location coordinates of one in the map ( 30 ) to be imaged terrain in a terrain framing track are receivable. Apparatus according to claim 10, further comprising communication means for acquiring additional data, for example weather data, the additional data for processing the card ( 30 ) are usable.

Images