DE102019201757A1 - Procedure and assistance system for monitoring the surroundings of an ego vehicle - Google Patents

Abstract

The invention relates to a method for monitoring the environment of an ego vehicle (1), using a sensor system (2) of an assistance system (3) at least one object (4) in the environment (5) of the ego vehicle (1) and object data of the respective object (4) are recorded for its classification, whereby by means of the assistance system (3) collision avoidance strategies, at least one cost function per collision avoidance strategy, at least one budget function, and at least one output and optimization function are provided, with a currently available budget required for collision avoidance is determined by means of the budget function depending on the recorded object data of the respective object (4), the collision avoidance strategies being prioritized by means of the output and optimization function taking into account the determined budget and the respective cost function, with the respective cost function being used to prioritize situational costs for the respective collision avoidance strategy can be determined, one of the prioritized collision avoidance strategies being selected and then executed in order to prevent a collision with the respective object (4). The invention also relates to an assistance system (3) for monitoring the surroundings of the ego vehicle (1).

Description

The present invention relates to a method and an assistance system for monitoring the surroundings of an ego vehicle. The invention also relates to a computer program product.

From the WO 2017/106 478 A1 discloses a system comprising a mobile machine and a computing device, wherein the mobile machine comprises a position determining device and the computing device is coupled to the position determining device. The computing device is configured to effect operation of the mobile machine according to a route map, the route map including information about field obstacles. The computing device receives location information from the position determining device. In addition, the computing device compares the location information with the information about the field obstacles, the information about the field obstacles having corresponding field coordinates of the field obstacles and identification information for each of the field obstacles. From this, it is automatically determined whether or not a change in the operation of the mobile machine is to be effected within a predetermined distance from one of the field coordinates on the basis of the comparison.

Based on the prior art, the present invention is based on the object of providing a method and an assistance system for monitoring the surroundings of an ego vehicle, with the aid of which collision avoidance is optimized. The object is achieved by the subject matter of claims 1 and 6. Further advantageous embodiments can be found in the dependent claims.

According to a method according to the invention for monitoring the surroundings of an ego vehicle, at least one object in the surroundings of the ego vehicle and object data of the respective object for its classification are recorded by means of a sensor system of an assistance system, with collision avoidance strategies, at least one cost function per collision avoidance strategy, at least one budget function using the assistance system , and at least one output and optimization function are provided, a currently available budget required for collision avoidance is determined by means of the budget function as a function of the recorded object data of the respective object, the collision avoidance strategies by means of the output and optimization function taking into account the determined budget and the respective cost function are prioritized, with situational costs for the respective collision avoidance strategy b be determined, wherein one of the prioritized collision avoidance strategies is selected, and wherein the selected collision avoidance strategy is executed in order to prevent a collision with the respective object. It is therefore a process in which the collision avoidance strategy is selected on the basis of a situational cost optimization. It is therefore no longer necessary to predefine the behavior of the assistance system in certain situations of the ego vehicle, but rather the assistance system itself decides which collision avoidance strategy is used.

Ego vehicles within the meaning of the invention are z. B. agricultural utility vehicles or machines (agricultural engineering) or utility vehicles in port areas that are operated at least partially autonomously in order to relieve a driver or operator of the ego vehicle and thus at least partially automate agriculture or port operations.

If the assistance system detects one or more objects in the vicinity of the ego vehicle, the respective object is classified by recording object data characterizing the respective object. The object data preferably include information about the type, size, position and / or mobility of the respective object. In other words, the sensor system detects whether it is a stationary object, for example a rock, or a moving object, for example a moving vehicle or ego vehicle. Furthermore, it is recorded where the object is located in the surroundings of the ego vehicle and in which direction it may be moving. The recorded object data about the respective object are stored in the assistance system and used to carry out the method.

Several collision avoidance strategies, at least one cost function per collision avoidance strategy, at least one budget function, and at least one output and optimization function are stored in the assistance system. The collision avoidance strategies preferably include an emergency stop of the ego vehicle, a change in the speed of the ego vehicle and / or a change in the route of the ego vehicle. The emergency stop includes braking or stopping to a standstill and, if necessary, deactivating the ego vehicle.

When the speed of the ego vehicle changes, a time interval at the potential collision point is changed. This is advantageous if the object lying in the vicinity of the ego vehicle is a moving vehicle, the ego vehicle and the further moving vehicle by changing a speed of the ego vehicle, hit or run over the potential collision point at different times.

By changing the route z. B. a drive around the respective object is initiated, wherein a distance of the ego vehicle to the potential collision point is increased, in particular maximized. The change in route is therefore suitable for B. for stationary, i.e. immovable objects.

Furthermore, a combination of several collision strategies or a change from a first to a second collision strategy is conceivable, which can be changed or adapted when the distance between the ego vehicle and the respective object is reduced. Of course, other collision avoidance strategies can also be defined depending on the application and stored in the assistance system.

One or more cost functions can be assigned to each collision avoidance strategy. Cost functions define virtual costs for one or more actions carried out in the respective collision avoidance strategy. For example, when carrying out a grain harvest, an unplanned bypassing an obstacle can trigger costs for an operator, since z. B. the areas bypassed are not harvested or other areas that are forcibly driven by the ego vehicle can be damaged. An emergency stop of the ego vehicle also causes virtual costs if z. B. Specialists have to arrive to manually restart the ego vehicle. The cost functions include, for example, a large number of different cost factors that are taken into account to determine the respective cost function.

The cost functions preferably include fixed cost factors and variable cost factors. Fixed cost factors are fixed values that can be precisely determined and cannot be changed. Using the example of the emergency stop, the costs of the arriving fitter cannot be changed and can accordingly be incorporated into the cost function as fixed cost factors. Fixed cost factors can also be fixed cost values for repairing a damaged or destroyed area. In contrast, variable cost factors are variable cost values that are variable depending on the workload and duration of the action. For example, because the object is driven around, it may be necessary to rework an area, such as mowing or threshing. Furthermore, when the ego vehicle is slowing down, delays in the operational sequence can result, which also incur costs.

The budget function is used to determine a current, i.e. situational, budget to avoid a potential collision. In a simplified example, the currently available budget is determined via the distance between the ego vehicle and the potential collision point. The closer the ego vehicle is to the collision point, that is to say the smaller the distance between the ego vehicle and the potential collision point, the greater the budget available for the collision avoidance strategy. In this case, an action or strategy for collision avoidance can be selected which is comparatively expensive, that is to say causes comparatively high virtual costs. For example, an emergency stop, which is associated with the costs of restarting manually, can be significantly more expensive than simply driving around the object or slowing down or accelerating the ego vehicle.

The output and optimization function defines how the collision avoidance strategies are selected based on the current budget. The collision avoidance strategies are prioritized, that is, they are sorted according to relevance, costs or other meaningful parameters. Various strategies for prioritizing the collision avoidance strategies can be useful, which are essentially dependent on the specific field of application of the ego vehicle. For example, outputs of the output and optimization function can be aimed at, which select collision avoidance strategies which cause the lowest possible costs. This is advantageous in agriculture, for example, where there is comparatively high cost pressure. Alternatively, outputs of the output and optimization function can be aimed at, which select collision avoidance strategies which generally cause higher costs. This can be advantageous in port operations, for example, where a high level of security must always be guaranteed.

The output and optimization function preferably includes a safety function in order to select one of the prioritized collision avoidance strategies on the basis of safety aspects. The safety function influences the selection of the collision avoidance strategy with regard to the stated strategy for prioritization.

Using the output and optimization function, which takes the budget function into account, and the cost functions, the assistance system determines a collision avoidance strategy and then executes it. The determination of the budget, the prioritization of the collision avoidance strategies and the selection of the collision avoidance strategy to be carried out can be carried out once or at certain time intervals around the To optimize the driving operation of the ego vehicle depending on the object data of the respective object and to ensure collision avoidance.

In an assistance system according to the invention for monitoring the environment of an ego vehicle, at least one object in the environment of the ego vehicle as well as object data of the respective object for its classification can be recorded by means of a sensor system, the assistance system collision avoidance strategies, at least one cost function per collision avoidance strategy, at least one budget function, and at least comprises an output and optimization function, wherein a currently available budget required for collision avoidance can be recorded by means of the budget function as a function of the recorded object data of the respective object, the collision avoidance strategies by means of the output and optimization function taking into account the determined budget and the respective Cost function can be prioritized, with situational costs for the respective collision avoidance strategy being determinable by means of the respective cost function, one of the priori based collision avoidance strategy can be selected and executed in order to prevent a collision with the respective object.

The assistance system or driver assistance system has a sensor system which comprises a large number of different sensor units and / or elements which are advantageous for monitoring the surroundings of the ego vehicle. The sensor system preferably comprises at least one radar sensor or a camera. This monitors the surroundings of the ego vehicle and, in the event of an object entering the sensible area of the sensor system, a corresponding signal is transmitted to the assistance system. Lidar sensors are also conceivable in order to detect objects lying or occurring in the vicinity of the ego vehicle.

The assistance system is preferably designed as a control device which evaluates the measurement data from the sensor system in order to initiate a collision avoidance strategy in the case of an object lying in the direction of travel of the ego vehicle. In other words, the assistance system is designed to monitor, optimize and control the driving operation of the ego vehicle.

The assistance system preferably comprises longitudinally and / or laterally acting actuators which can be used to carry out the selected collision avoidance strategy. A longitudinally acting actuator can, for example, be a braking mechanism, whereas a laterally acting actuator is to be understood, for example, as a steering device that is operatively connected to the wheels of the ego vehicle.

The method according to the invention can e.g. B. run by a computer or by a control unit or an assistance system. The method can thus be implemented in software. In this respect, the corresponding software is an independently salable product. The invention therefore also relates to a computer program product with machine-readable instructions which, when executed on a computer or on a control device, upgrade the computer and / or the control device to an operating logic of the assistance system for monitoring the surroundings of an ego vehicle or cause it to to carry out a method according to the invention.

• 1 eine stark vereinfachte schematische Ansicht eines Ego-Fahrzeugs mit einem erfindungsgemäßen Assistenzsystem zur Umfeldüberwachung,
• 2 eine schematische Darstellung eines Ausführungsbeispiels des Verfahrens gemäß der Erfindung, und
• 3 eine schematische Darstellung eines Verlaufs von Kostenfunktionen und einer Budgetfunktion in Abhängigkeit der Zeit.

A preferred exemplary embodiment of the invention is explained in more detail below with reference to the figures. Here shows

• 1 a greatly simplified schematic view of an ego vehicle with an assistance system according to the invention for monitoring the surroundings,
• 2 a schematic representation of an embodiment of the method according to the invention, and
• 3 a schematic representation of a course of cost functions and a budget function as a function of time.

According to 1 assigns an ego vehicle 1 an assistance system according to the invention 3 for monitoring the surroundings of the ego vehicle 1 on. A sensor system 2 of the assistance system 3 , comprising a radar sensor 7th , is intended to be an environment 5 of the ego vehicle 1 to monitor and objects occurring or present therein 4th capture. The sensors 2 captures specific object data to classify the respective object 4th such as the type, size, position and / or mobility of the respective object 4th , and stores this data in the assistance system 3 . The assistance system also includes 3 a longitudinal actuator 6a in the form of a braking device and a lateral actuator 6b in the form of a steering device.

In the present case, it is the object 4th the direction of travel of the ego vehicle 1 crossing agricultural machine 9 . The direction of movement of the agricultural machine 9 as well as the ego vehicle 1 are symbolized by an arrow. The assistance system 3 is presently set up for this purpose, the object data of the agricultural machine 9 , e.g. B. to record its direction and speed of movement and a potential collision point based on this object data 8th of the ego vehicle 1 with the agricultural machine 9 to investigate.

From the assistance system 3 are present several collision avoidance strategies, at least one cost function per collision avoidance strategy, a budget function and an output and optimization function are provided, with the assistance system 3 with the aid of these functions, a method according to the invention, as shown below in an exemplary embodiment 2 and 3 is described.

2 illustrates the sequence of an exemplary embodiment of the method according to the invention for monitoring the surroundings of the ego vehicle 1 to 1 . In one step 101 becomes the environment 5 of the ego vehicle 1 by means of the sensors 2 observed or scanned and the corresponding measurement data from the assistance system 3 evaluated. Used by the assistance system 3 an object 4th in the environment 5 of the ego vehicle 1 recognized, takes place in one step 102 an acquisition of the object data of the object 4th with regard to its type, size, position and / or mobility.

In one step 103 a currently available budget required for collision avoidance is determined by means of the budget function, taking into account the object data. The budget limits the selection of a suitable collision avoidance strategy with regard to its virtual costs, which as cost functions in the assistance system 3 are deposited. In other words, situational costs for the respective collision avoidance strategy are determined by means of the respective cost function. For each previously defined collision avoidance strategy, for example an emergency stop of the ego vehicle 1 , a change in the speed of the ego vehicle 1 and / or a change in the route of the ego vehicle 1 , a respective cost function is stored. The cost functions include both fixed cost factors and variable cost factors. Fixed cost factors are costs that are always the same regardless of the time of execution or the duration of the respective collision avoidance strategy. Variable cost factors describe costs for the collision avoidance strategy, which cause variable costs depending on the duration, the effort and / or the type of strategy.

Using the output and optimization function, in one step 104 the collision avoidance strategies are prioritized taking into account the budget determined and the respective cost function, the prioritization being essentially dependent on the area of application of the ego vehicle 1 is dependent. For this purpose, the output and optimization function includes a safety function in order to select one of the prioritized collision avoidance strategies based on safety aspects. For example, the greater the requirements for safe operation of the ego vehicle 1 are, the earlier collision avoidance strategies can be selected that cause higher virtual costs. In a subsequent step 105 one of the prioritized collision avoidance strategies is selected and done in one step 106 the execution of the selected collision avoidance strategy is initiated, with the in 1 shown actuators 6a , 6b be controlled in order to execute the respective collision avoidance strategy and a collision with the respective object 4th to prevent.

3 shows an exemplary course of two cost functions 11 , 12th and a budget function 10 depending on the time according to the description for step 104 , with the costs or the budget on the ordinate 14th and the time on the abscissa 13th are registered. Is the ego vehicle 1 for example far enough from the potential collision point 8th removed or there is no risk of collision between vehicles 1 , 9 before, there is also no selection and execution of a specific collision avoidance strategy. The budget function 10 is then below a first cost function 11 . The budget function becomes during driving 10 determined continuously taking into account the property data.

Recorded by the assistance system 3 a potential collision hazard, for example if both vehicles are 1 , 9 the potential collision point while driving 8th approximate, this is done by the sensors 2 recorded so that the budget function 10 increases. At a time T1 then becomes the first cost function 11 selected with the first collision avoidance strategy and the driving operation continued based on the first collision avoidance strategy. During further driving it can happen that the vehicles 1 , 9 the potential collision point 8th move closer so that the risk of collision increases. This can be seen in the increasing budget function 10 . The assistance system 3 is set up, for example, to select a second collision avoidance strategy, which is more expensive than the first collision avoidance strategy, in order to prevent the possible collision. Reached the budget function 10 at one after the first time T1 taking place second time T2 a defined value, is a second cost function 12th and accordingly a second collision avoidance strategy can be selected, the driving operation being continued on the basis of the second collision avoidance strategy.

It is alternatively possible that the assistance system 3 several potential collision points are determined so that a budget function is determined over all possible collision points. In addition, the output and optimization function can not only have one collision point 8th but optimize all theoretically possible collision points. In these cases, the ego vehicle can be driven 1 then be optimized if there are several objects 4th in the environment 5 of the ego vehicle 1 are located.

As an alternative, it is also conceivable that an active duration of cost functions of a respective collision avoidance strategy is taken into account. With increasing duration, the costs of these cost functions and thus also the respective strategies are classified as more and more expensive the longer they are used. Long-lasting collision avoidance strategies are a sign of ineffective functions, which are made more expensive by this method and at some point are no longer taken into account by the output and optimization function. In other words, the effectiveness of the respective collision avoidance strategy is checked by taking into account an active duration.

List of reference symbols

1

Ego vehicle

2

Sensors

3

Assistance system

4th

object

5

Environment of the ego vehicle

6a, 6b

Actuator

7th

Radar sensor

8th

Collision point

9

Agricultural machine

10

Budget function

11

First cost function

12th

Second cost function

13th

abscissa

14th

ordinate

101

Process step

102

Process step

103

Process step

104

Process step

105

Process step

106

Process step

T1

First point in time

T2

Second point in time

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

• WO 2017/106478 A1 [0002]

Claims (8)

Method for monitoring the surroundings of an ego vehicle (1), with at least one object (4) in the surroundings (5) of the ego vehicle (1) and object data of the respective object (4) being supplied by means of sensors (2) of an assistance system (3) its classification can be recorded, with the assistance system (3) - collision avoidance strategies, - at least one cost function per collision avoidance strategy, - at least one budget function, and - At least one output and optimization function are provided, with the following process steps being carried out: - Determination of a currently available budget required for collision avoidance by means of the budget function as a function of the recorded object data of the respective object (4); - Prioritization of the collision avoidance strategies by means of the output and optimization function taking into account the determined budget and the respective cost function, with situational costs for the respective collision avoidance strategy being determined by means of the respective cost function; - Selecting one of the prioritized collision avoidance strategies, - Execution of the selected collision avoidance strategy in order to prevent a collision with the respective object (4). Procedure according to Claim 1 , where the cost functions include fixed cost factors and variable cost factors. Method according to one of the preceding claims, wherein the collision avoidance strategies include an emergency stop of the ego vehicle (1), a change in the speed of the ego vehicle (1) and / or a change in the route of the ego vehicle (1). Procedure according to Claim 3 combined with Claim 2 , the object data including information about the type, size, position and / or mobility of the respective object (4). Method according to one of the preceding claims, wherein the output and optimization function comprises a safety function in order to select one of the prioritized collision avoidance strategies on the basis of safety aspects. Assistance system (3) for monitoring the surroundings of an ego vehicle (1), using a sensor system (2) at least one object (4) in the surroundings (5) of the ego vehicle (1) and object data of the respective object (4) for its classification are detectable, furthermore comprehensively - collision avoidance strategies, - at least one cost function per collision avoidance strategy, - at least one budget function, and - At least one output and optimization function, wherein a currently available budget required for collision avoidance can be recorded using the budget function as a function of the recorded object data of the respective object (4), the collision avoidance strategies using the output and optimization function taking into account the determined Budgets and the respective cost function can be prioritized, with the respective cost function being able to determine situational costs for the respective collision avoidance strategy, one of the prioritized collision avoidance strategies being selectable and executable in order to prevent a collision with the respective object (4). Assistance system (3) Claim 6 , characterized by longitudinally and / or laterally acting actuators (6a, 6b) which can be used to carry out the selected collision avoidance strategy. Computer program product containing machine-readable instructions which, when executed on a computer and / or on a control device, convert the computer and / or control device to an operating logic of the assistance system for monitoring the surroundings of an ego vehicle (1) according to the Claims 6 to 7th upgrade, and / or induce a method according to one of the Claims 1 to 5 execute.

Images