DE102021001932A1 - Method for training the handling of emergency vehicles by an autonomously driving motor vehicle - Google Patents

Abstract

The invention relates to a method for training the handling of emergency vehicles (1) by an autonomously driving motor vehicle (2), in which an emergency drive of such an emergency vehicle (1) is first simulated with an original vehicle (3) on a test site (4) and from this a skeleton (5) and a movement pattern (6) can be detected. The skeleton (5) recorded in this way is then covered with a virtual outer shell (9) of an emergency vehicle (1) and its movement pattern (6) is displayed in at least one detection device (7) of the autonomously driving motor vehicle (2), so that this is via a computer device ( 8) can be analyzed and appropriate driving maneuvers can be initiated.

Description

The present invention relates to a method for training the handling of emergency vehicles by an autonomously driving motor vehicle.

From the DE 10 2017 126 790 A1 a method for recognizing and reacting to emergency vehicles on a roadway is known. In the known method, sensor data are first called up from a large number of sensors on the vehicle, whereupon a detection takes place if the emergency vehicle is approaching the vehicle on a roadway. The detection takes place on the basis of the retrieved sensor data. An avoidance strategy is then determined with which the vehicle can avoid the emergency vehicle, this being done on the basis of additional sensor data. Vehicle components are automatically controlled in order to induce the vehicle to implement the evasive strategy.

In modern road traffic, autonomously driving vehicles are increasingly being used, which detect and react to their surroundings via a large number of sensors. Situations in which emergency vehicles, such as police vehicles, fire engines or rescue vehicles, approach the autonomously driving motor vehicle are of particular importance. These are particularly important because the autonomously driving motor vehicle must react correctly to the approaching emergency vehicle and must, for example, grant them the right of way. In order to be able to practice the safest possible handling of such emergency vehicles, a large number of encounters with such emergency vehicles should be recorded and appropriate driving maneuvers learned or carried out. However, this can make it necessary to drive a large number of kilometers driven, for example hundreds of thousands of kilometers, in order to obtain and implement enough random encounters with emergency vehicles. However, driving so many kilometers represents a not inconsiderable effort with not inconsiderable costs.

The present invention is therefore concerned with the problem of specifying a method for training how to deal with emergency vehicles using an autonomously driving motor vehicle, which in particular overcomes the disadvantages known from the prior art.

According to the invention, this problem is solved by the subject matter of independent claim 1. Advantageous embodiments are the subject of the dependent claims.

The present invention is based on the general idea of circumventing a normally only random encounter between an autonomously driving motor vehicle and an emergency vehicle in normal traffic by simulating the drive of such an emergency vehicle on a cordoned-off test site, extracting the movement patterns that have been used and using a neural network of the autonomously driving motor vehicle is supplied so that it can learn and practice handling such emergency vehicles more easily. In the method according to the invention for training the handling of emergency vehicles by an autonomously driving motor vehicle, an emergency drive of such an emergency vehicle is first simulated on a test site with an original vehicle, that is, any normal motor vehicle. A skeleton and a movement pattern of the starting vehicle are recorded and the recorded skeleton and the recorded movement pattern are then covered with a virtual outer shell of an emergency vehicle, for example a police vehicle, a fire engine or an ambulance. This deployment trip of the starting vehicle covered with the virtual outer shell of the emergency vehicle is then faded in into at least one detection device of the autonomously driving motor vehicle, so that the autonomously driving motor vehicle is “tricked” into thinking that such an emergency vehicle is approaching the autonomously driving motor vehicle, for example. The detection device of the autonomously driving motor vehicle now transmits the displayed movement of the emergency vehicle covered with the virtual outer shell to a computer device, for example a neural network, whereupon the computer device analyzes the movement pattern of the original vehicle covered with the virtual outer shell of the emergency vehicle and appropriate driving maneuvers, for example evasive maneuvers, initiates. This makes it possible to drive through specific traffic scenarios with emergency vehicles in order to represent special traffic situations. Instead of driving millions of kilometers with data collecting vehicles in the hope of capturing such traffic situations with emergency vehicles, one can use the method according to the invention with any simple motor vehicle, i.e. original vehicle, to drive such emergency journeys inexpensively, quickly and easily on closed proving grounds and then through Fade in, for example, a detection optics of the autonomously driving motor vehicle.

The method according to the invention thus provides a comparatively simple method for a so-called “Do as I do” movement transfer, in which a source movement pattern of the starting vehicle is recorded and a skeleton of the starting vehicle is combined or covered with an outer shell of an emergency vehicle, in order then to be displayed in the detection device of the autonomously driving motor vehicle. In this way, the handling of such emergency vehicles can be trained much faster, more cost-effectively and thus more effectively than was possible with accidental encounters with such emergency vehicles.

In an advantageous development of the method according to the invention, the computer device has a neural network or is part of such a network. Such neural networks make it possible to train and learn how to deal with certain situations by evaluating sensor data, developing new connections, deleting existing connections and changing the weighting of different neurons or sensor data. A large number of emergency vehicle trips simulated with the original vehicle on a test site and fading in of these emergency trips in a detection device of an autonomously driving motor vehicle enables particularly fast, effective and inexpensive learning to deal with such emergency vehicles.

An image acquisition device, for example a camera, is expediently used as the acquisition device in the autonomously driving motor vehicle. Of course, other detection devices, such as radar devices, can also be included as an alternative or in addition to the emergency vehicle drive, which is initially simulated by the original vehicle and covered with the virtual outer shell on the basis of the skeleton detected in the process. Autonomously driving motor vehicles usually have a large number of cameras, so that for training in handling emergency vehicles, for example with a rear-view camera, the superimposed movement of the emergency vehicle covered with the virtual outer shell is superimposed. In this case, the autonomously driving motor vehicle is of the opinion that an emergency vehicle is approaching it from behind. Using the detected movement pattern of the starting vehicle covered with the virtual outer shell of the emergency vehicle, which the detection device forwards to the computer device of the autonomously driving motor vehicle, the latter can initiate a corresponding driving maneuver, for example an evasive maneuver or driving into a hard shoulder, using algorithms.

Further important features and advantages of the invention emerge from the subclaims, from the drawings and from the associated description of the figures on the basis of the drawings.

It goes without saying that the features mentioned above and those yet to be explained below can be used not only in the respectively specified combination, but also in other combinations or on their own, without departing from the scope of the present invention.

Preferred exemplary embodiments of the invention are shown in the drawings and are explained in more detail in the following description, with the same reference symbols referring to the same or similar or functionally identical components.

• 1 einen ersten Verfahrensschritt eines erfindungsgemäßen Verfahrens zum Trainieren eines Umgangs mit Einsatzfahrzeugen durch ein autonom fahrendes Kraftfahrzeug mit einem sich auf einem Prüfgelände bewegenden Ausgangsfahrzeug,
• 2 eine Darstellung wie in 1, jedoch mit extrahiertem Skelett und Bewegungsmuster des Ausgangsfahrzeugs,
• 3 eine Darstellung wie in 2, jedoch bei einem mit einer virtuellen Außenhülle eines Einsatzfahrzeuges überzogenem Ausgangsfahrzeug,
• 4 ein Einblenden des aus 3 mit der virtuellen Außenhülle des Einsatzfahrzeuges überzogenen Ausgangsfahrzeugs in eine Erfassungseinrichtung eines autonom fahrenden Kraftfahrzeugs.

They show, in each case schematically,

• 1 a first method step of a method according to the invention for training how to deal with emergency vehicles by an autonomously driving motor vehicle with a starting vehicle moving on a test site,
• 2 a representation as in 1 , but with the skeleton and movement pattern of the original vehicle extracted,
• 3 a representation as in 2 , but with a starting vehicle covered with a virtual outer shell of an emergency vehicle,
• 4th a display of the 3 with the initial vehicle covered with the virtual outer shell of the emergency vehicle into a detection device of an autonomously driving motor vehicle.

According to the 1 to 4th are different method steps of a method according to the invention for training how to deal with emergency vehicles 1 by an autonomously driving motor vehicle 2 described.

In the method according to the invention, first according to the 1 with an exit vehicle 3 , which can be designed as a normal motor vehicle, on a test site 4th an emergency drive of an emergency vehicle 1 simulated. The starting vehicle drives 3 the movement patterns that may occur during an emergency trip.

According to the 1 now become a skeleton 5 and a movement pattern 6th of the starting vehicle 3 recorded, whereupon in 3 the captured skeleton 5 with a virtual outer shell of an emergency vehicle 1 is covered. The movement pattern 6th will be out 2 continue to be maintained.

Then according to the 4th in at least one detection device 7th of autonomously driving motor vehicle 2 the mission drive with the virtual outer shell 9 of the emergency vehicle 1 coated starting vehicle 3 faded in. The autonomously driving motor vehicle is driving 2 on a normal road 11 . The detection device 7th now transmits the overlaid movement of the with the virtual outer shell 9 of the emergency vehicle 1 coated starting vehicle 3 to a computing device 8th of the autonomously driving motor vehicle 2 , whereupon this the movement or the movement pattern 6th the one with the virtual outer shell 9 of the emergency vehicle 1 coated starting vehicle 3 analyzes and initiates appropriate driving maneuvers, for example evasive action or approaching a hard shoulder, in order to avoid the emergency vehicle 1 To give priority.

With the method according to the invention, it is possible for the first time, so far only purely by chance when driving autonomous motor vehicles autonomously 2 Encounter situations with emergency vehicles occurring on the streets 1 now consciously bring about by such operational situations with a corresponding movement pattern 6th with a conventional motor vehicle, i.e. in the present case with the starting vehicle 3 , on a proving ground 4th be traveled and recorded. To the autonomously driving motor vehicle 2 then indicate that it is actually an emergency vehicle 1 acts, becomes the skeleton 5 of the starting vehicle 3 captured and with a virtual outer shell 9 overdrawn. Now, for example, visually as an emergency vehicle 1 recognizable, but only with the virtual outer shell 9 such an emergency vehicle 1 coated home vehicle 3 becomes together with the movement pattern carried out by it 6th now in the acquisition device 7th of the autonomously driving motor vehicle 2 faded in, whereupon this, for example by means of a neural network and a computer device 8th , how to deal with such an emergency vehicle 1 , for example initiating evasive maneuvers.

On the proving ground 4th can do a wide variety of emergency trips for a wide variety of emergency vehicles 1 , for example police vehicles, fire engines, ambulances etc., simulated and then with the for the respective emergency vehicle 1 associated outer shell 9 covered and the autonomously driving motor vehicle 2 be displayed. Training in handling such emergency vehicles 1 can thus be carried out significantly faster and more effectively and, in particular, also more cost-effectively than would be possible in normal road traffic, in which such encounters with emergency vehicles are only random and, for example, at a distance of several thousand kilometers 1 would occur.

The computing device 8th can have a neural network or be embedded in such a network and, in particular, also exchange data with other autonomously driving motor vehicles 2 carry out. The autonomously driving motor vehicle 2 can of course add further sensors or detection devices 7th exhibit to not just the emergency vehicle 1 , but also an environment or other motor vehicles 10 to record and when encountering the emergency vehicle 1 to coordinate the driving maneuvers to be initiated accordingly.

The detection device 7th in the autonomously driving motor vehicle 2 can for example be an image acquisition device, in particular a camera, but also a radar or the like.

QUOTES INCLUDED IN THE DESCRIPTION

This list of the documents listed by the applicant was 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.

Patent literature cited

• DE 102017126790 A1 [0002]

Claims (3)

Method for training the handling of emergency vehicles (1) by an autonomously driving motor vehicle (2), in which - an emergency drive of such an emergency vehicle (1) is initially simulated with an exit vehicle (3) on a test site (4), - a skeleton (5) and a movement pattern (6) of the starting vehicle (3) are recorded, - the detected skeleton (5) is covered with a virtual outer shell (9) of an emergency vehicle (1), - In at least one detection device (7) of the autonomously driving motor vehicle (2), the mission drive of the original vehicle (3) covered with the virtual outer shell (9) of the emergency vehicle (1) is displayed, - the detection device (7) transmits the superimposed movement pattern (6) of the starting vehicle (3) covered with the virtual outer shell (9) of the emergency vehicle (1) to a computer device (8), - The computer device (8) analyzes the movement pattern (6) of the starting vehicle (3) covered with the virtual outer shell (9) of the emergency vehicle (1) and initiates corresponding driving maneuvers. Procedure according to Claim 1 , characterized in that the computer device (8) has a neural network. Procedure according to Claim 1 or 2 , characterized in that an image acquisition device, in particular a camera, is used as the acquisition device (7) in the autonomously driving motor vehicle (2).

Images