CN109211255B - Method for planning a route for a motor vehicle having an automatic vehicle system - Google Patents

The invention relates to a method for planning a route for a motor vehicle having an automatic vehicle system, which takes into account route-specific positionability and the performance of the automatic vehicle system, so that an adaptive routing can be implemented, wherein a feasibility analysis of at least one route is created, wherein the feasibility analysis comprises the following steps: selecting a route to be investigated from a starting point to a target point; determining an environmental condition on the selected route that is not associated with the vehicle; determining an expected detection rate of landmarks on the selected route suitable for vehicle localization by using environmental conditions not associated with the vehicle; determining a desired positioning accuracy on the selected route by using the desired detection rate; it is determined whether the desired positioning accuracy is sufficient to be supported by the automated vehicle system for traversing the selected route.

Method for planning a route for a motor vehicle having an automatic vehicle system

Technical Field

The invention relates to a method for planning a route for a motor vehicle having an automatic vehicle system. The invention further relates to a motor vehicle having an automatic vehicle system, wherein the vehicle system is designed to carry out a route planning method.

Background

Modern driving Assistance Systems (ADAS) and highly automated vehicle Systems for UAD (city autonomous driving) increasingly require detailed knowledge of the vehicle environment and situational awareness. Sensor measurement data is used as a basis for perceiving the vehicle environment. Objects can thus be extracted by means of a so-called detection algorithm, by means of which the vehicle environment can be described and analyzed. Modern environmental sensors, such as stereo video cameras or laser scanners, in combination with detection algorithms, can obtain a large amount of information from the vehicle environment, for example in the form of landmarks. These landmarks include traffic signs, traffic lights, lane markings, and the like. The detected object or detected landmark may in turn be used for vehicle localization. The performance of the entire automatic vehicle system is dependent here primarily on the performance of the environmental sensors.

The accuracy of vehicle positioning required by an automated vehicle system is not solely dependent on the performance of the environmental sensor system. In addition, the choice of environmental conditions and detection algorithms may also have a significant impact. The performance of the entire vehicle system is directly related to the route to be traveled and the environmental conditions known on the route.

Disclosure of Invention

The object of the invention is to provide a method for routing a motor vehicle having an automatic vehicle system, which takes into account the route-specific positionability and the performance of the automatic vehicle system, so that a situation-adapted routing can be achieved.

It is also an object of the invention to provide a motor vehicle with an automatic vehicle system which provides a situation-adaptive route planning.

To achieve this object, a method for planning a route for a motor vehicle having an automatic vehicle system is proposed, wherein a feasibility analysis of at least one route is created, wherein the feasibility analysis comprises the following steps:

-selecting a route to be investigated from a starting point to an end point,

-determining a vehicle-independent environmental condition on the selected route,

-determining an expected detection rate of landmarks suitable for vehicle localization on the selected route by using vehicle-independent environmental conditions,

-determining a desired positioning accuracy on the selected route by using the desired detection rate,

-determining whether the expected positioning accuracy is sufficient to be supported by the automatic vehicle system for driving through the selected route.

The automatic vehicle system may be a driver assistance system for driving assistance and/or highly automated driving and/or autonomous driving.

It is advantageous in the route planning method to take into account sufficiently precise route-specific positionability and overall system performance, in particular of the automated vehicle system. Routes along which satisfactory positioning cannot be achieved under known environmental conditions and with environmental sensor systems usable by vehicle systems may be classified as impassable. Such a route or route section may be bypassed. The method uses as input variables the environmental conditions along the route to be investigated, which are independent of the vehicle. Landmarks suitable for vehicle localization may be obtained from a localization map.

The method makes it possible to plan routes on which the required performance of the vehicle system is ensured with a high probability, so that a favorable influence on the robustness of the automatic vehicle system is made possible. In contrast to known route planning methods for automatic vehicle systems, vehicle-independent environmental conditions, i.e. environmental conditions which do not relate to the vehicle and/or the automatic vehicle system, are additionally used. However, the vehicle-related environmental conditions may also be included in the calculation of the expected positioning accuracy. The vehicle-related environmental condition may be, for example, sensor efficiency or robustness of a landmark detection algorithm.

The environmental sensor may be a stereo camera or a laser scanner or other suitable sensor device.

In this method, the step of determining the expected detection rate of landmarks suitable for vehicle localization is not necessarily performed in a separate method step, but may also be implicit in the determination of the expected localization accuracy. The key to the route planning method is to determine the expected location accuracy by using environmental conditions not associated with the vehicle on the selected route and then determine whether the expected location accuracy is sufficient to travel the selected route supported by the automated vehicle system. For example, it is also possible to determine the expected positional accuracy or to determine whether the expected positional accuracy is sufficient for the automatic vehicle system to drive through the selected route by means of a parameterized model and/or a neural network or other methods of machine training and/or statistical models. In the case of a parameterized model, the detection rate in the form of parameters may be explicitly or implicitly included in the method. If the positioning accuracy is determined by means of a neural network or the like, the expected detection rate is implicitly taken into account in the weights trained in the neural network. The detection rate is also included in the determination of the localization accuracy by statistical weighting in the statistical model. It is important to understand the present invention that the positioning accuracy depends implicitly or explicitly on the product of the detection rate (i.e. the percentage of landmarks detectable on the selected route) and the number of landmarks present on the selected route.

Furthermore, the detection rate, which is implicitly or explicitly entered into the method, may also depend on the type of landmark present on the selected route. For example, the detection rates of traffic signs, lane boundaries, traffic light signals, trees, or buildings may be different. In addition, the detection rate may also depend on the efficiency of the sensor device of the environmental sensor system. The detection algorithms used may also have different efficiencies for the detection rates of particular types of landmarks. By considering the detection rate explicitly or implicitly, all of the above factors can be combined individually or together to determine the expected location accuracy and then determine whether the expected location accuracy is sufficient to be supported by the autonomous driving system to travel through the selected route.

It is preferably provided that the vehicle-independent environmental condition is the traffic density of the selected route and/or the weather conditions and/or the road conditions.

Weather conditions, total and density of traffic, and road conditions may affect the detection performance of the sensor system, thereby affecting the detection rates of different types of landmarks. A high traffic density may in particular lead to a certain proportion of the landmarks on the selected route being at least temporarily obscured by the vehicle in front of the surroundings sensor system, whereby the detection rate and the positioning accuracy are reduced. Therefore, the correlation of the positioning accuracy with the current weather conditions and the total amount of traffic is advantageously taken into account. The environmental conditions not related to the vehicle, in particular the weather conditions and the total amount of traffic, can be determined, for example, by querying a weather database or a traffic database.

Provision is preferably made for determining an occlusion rate of the landmarks on the selected route, wherein the occlusion rate is preferably determined by using environmental conditions and/or a suitable landmark type, wherein the positioning accuracy and/or the detection rate is determined by using the occlusion rate.

The occlusion rate advantageously maps the impact of route-dependent and vehicle-independent environmental conditions on the positioning accuracy. Here, the occlusion rate may also depend on the type of landmark. Thus, landmarks located at lower elevations above the road surface are more likely to be obscured by increased traffic and vehicle populations for environmental sensor systems, which in turn reduces the detection rate of such landmarks. Conversely, landmarks that are disposed higher (e.g., traffic light signals) have a lower occlusion rate.

Occlusion rates may also be considered implicitly or explicitly in the method. In a parameterized model or neural network, implicit consideration may be made by parameters or training weights.

It is further preferably provided that the occlusion rate and/or the detection rate and/or the positioning accuracy are determined by using a parameterized model, wherein the parameterized model is preferably a machine training model, wherein the machine training model is further preferably trained by using a previous passability analysis, in particular by using a previous environmental condition and/or a previously determined detection rate and/or a previously determined occlusion rate and/or a previously determined positioning accuracy and/or a performance of the sensor device and/or a detection algorithm of the automated vehicle system.

The occlusion rate and/or detection rate and/or positioning accuracy may advantageously be determined by using a parameterized model or by using a neural network. In this case, the parameterized model or the neural network determines the occlusion rate and/or the detection rate and/or the positioning accuracy on the basis of a previously performed feasibility analysis, i.e. by using the results of the feasibility analysis performed before the method is currently performed.

It can therefore be provided that the parameterized model or the neural network undergoes a training phase before the feasibility analysis is carried out, in particular before the selection of the route to be investigated is carried out.

In a preferred embodiment, the occlusion rate is determined by using a parameterized model, wherein current weather information and/or the current amount of traffic on the selected route and/or the type of landmarks present on the selected route are used as input variables of the parameterized model.

The system is based on a model parameterized by a machine training method, in particular a neural network, which establishes a correlation between weather and traffic data and other information about the environment, an environment perception module or environment sensors used on the vehicle side and the resulting detection probabilities of the individual landmark types.

It is further preferably provided that the detection rate is determined by using the occlusion rate and/or the number, in particular the maximum number and/or the number density and/or the type of landmarks detectable on the selected route and/or the performance of the sensor device and/or the detection algorithm of the automatic vehicle system and/or the accuracy of the environmental conditions and/or the positioning data, in particular the accuracy of the GPS data.

The occlusion rate is used to determine a percentage of type-related landmarks that may be occluded by environmental conditions (e.g., increased total traffic) on the selected route. The detection rate then depends on the occlusion rate and, in addition, on the efficiency of the sensor or sensor arrangement. Furthermore, the detection rate may also depend on the efficiency of the detection algorithm, wherein different detection algorithms may be used, in particular for different types of landmarks.

It is further preferably provided that the positioning accuracy is determined by means of a statistical model, wherein the positioning accuracy is preferably determined by using an expected detection rate and/or an expected occlusion rate and/or a number density and/or a type of landmarks along the selected route.

The accuracy of the positioning depends explicitly or implicitly on the expected detection rate and the number or number density, in particular with respect to type, of landmarks along the selected route. Here, the detection rate may be affected by the occlusion rate.

It can further advantageously be provided that the determined positioning accuracy is compared with a preset threshold value, wherein, if the positioning accuracy is less than the threshold value, the driver of the motor vehicle is enabled to have the possibility of manually driving through the selected route and/or a new traffic lane analysis is carried out for another route.

In this method, the driver may be notified when a positioning accuracy threshold is undershot, the vehicle may not be able to traverse the selected route using the automated vehicle system, and a new route may then be planned.

Alternatively or additionally, the driver of the motor vehicle may have the opportunity to decide whether he wants to take the originally planned or selected route manually or wants to accept an automatic, possibly longer journey for this purpose.

Advantageously, the overall route of the automated vehicle system can be globally planned and re-planned at defined target points.

Provision may preferably be made for the best possible positioning to be taken into account in the determination of whether the desired positioning accuracy is sufficient for the automatic vehicle system to drive through the selected route and/or for the determination to be optimized to the shortest possible route with a sufficiently accurate positioning accuracy.

Furthermore, it can be provided that a sensor device and/or a detection algorithm suitable for driving through the selected route supported by the automatic vehicle system is determined, preferably if the positioning accuracy is greater than or equal to a threshold value.

It is advantageously determined by means of the method which sensor devices are suitable for which type of landmark to drive through the selected route. The system is based in particular on the use of a parameterized model, on the establishment of a correlation between weather data, traffic data and other information about the environment, sensor devices or environment awareness modules on the vehicle side and the resulting detection probabilities for the individual landmark types. By means of this method, together with information about the density of landmarks, which can be obtained, for example, from a localization map, and detectable landmark types known from the localization modules of the individual vehicle systems, it can be determined whether a certain route segment can be currently traveled by the vehicle system. In this case, detection algorithms of different expressions on the vehicle side are taken into account by the method, i.e., depending on the existing sensor device or depending on the existing detection algorithm, the respective vehicle system may travel in an area which is not available for other vehicle systems. For this purpose, it can advantageously be determined which sensor devices and/or which detection algorithms are used, in particular in which section of the route, to drive through the route.

The method uses weather data, traffic information, information about GPS accuracy, and information about the density of detectable landmarks along a route as input variables. The model, whose parameters are determined by the machine training method, is used to enable a determination to be made as to whether the selected route can be traveled through the route, and if so, which detection algorithm may be utilized to traverse the route.

A further solution to the problem on which the invention is based consists in providing a motor vehicle with an automatic vehicle system, wherein the automatic vehicle system is designed to carry out the above-described method.

It may further preferably be provided that the automatic vehicle system is designed for driving assistance and/or highly automated driving and/or automatic driving.

Drawings

The present invention is described in detail below with reference to the accompanying drawings. Wherein:

figure 1 shows a flow chart of a route planning method for a motor vehicle with an automatic vehicle system,

FIG. 2 shows an overview of the input variables for the training phase of the parameterized model.

Detailed Description

Fig. 1 shows a flow chart of a route planning method for a motor vehicle with an automatic vehicle system.

In a first optional step S1, in a training phase of the parameterized model, in particular of the neural network, correlations are established between these input variables and expected occlusion rates of the respective landmark types, preferably at respective known positions, based on information about weather data, detection rates of the various types of landmarks and other information sources such as traffic data. The position may be determined, for example, by GPS, dead reckoning, or vehicle location.

In the case of application, the method starts with the feasibility analysis in step S2. In step S2, a route to be reviewed is selected for transit using an automated vehicle system, which may include a navigation system, based on the start point and the destination point.

Subsequently, in step S3, an environmental condition that is not related to the vehicle on the selected route is determined. These environmental conditions may be the current total amount of traffic on the selected route or the current weather conditions, and are determined by querying a database. In addition, the expected landmarks are obtained from the positioning map.

In a subsequent step S4, a parameterized model is used to determine the occlusion rate of each desired landmark type currently expected on the route to be examined. In this case, the current total traffic and/or the current weather conditions and, if necessary, also the landmarks stored in the map for the selected route are used as input variables.

In step S5, the detection rate of the landmark suitable for vehicle positioning on the selected route is determined by using the environment condition not related to the vehicle. Here, the occlusion rate determined in step S4 is taken into account in the determination of the detection rate. Furthermore, vehicle-related conditions, such as the detection efficiency of different sensor devices (e.g. stereo cameras or laser systems), as well as the performance of possible detection algorithms for different types of landmarks, can be taken into account, in particular under known environmental conditions (weather, traffic, etc.).

Then, in step S6, a desired positioning accuracy on the selected route is determined by using the desired detection rate. In other words, a combination of a particular occlusion rate, detection rate, and number, number density, and type of landmarks present along the route may enable an estimation of the positioning accuracy by means of a statistical model.

For a specific landmark type, the positioning accuracy of the selected route is derived in particular from the product of the detection rate and the number density of landmarks present on the selected route or on the segments of the selected route.

In a further step S7, it is determined whether the desired positioning accuracy is sufficient for driving through the selected route, supported by the automatic vehicle system. Thus, the necessity of re-planning a route is decided based on the trafficability analysis. If this is necessary, it can be carried out according to the method, which restarts the feasibility analysis in step S2 again with a new selected route.

If the positioning accuracy is sufficient for the autonomous vehicle system to drive through the selected route, the sensor device and the detection algorithm to be used by the vehicle system are additionally specified in a further step S8 for the route classified as passable by the method. Further, a plan for switching detection algorithms along the selected route may be created to be responsive to changing landmark types.

For example, automated vehicle systems are routed along major roads. Sufficient environmental information is available along the route so that a sufficiently accurate positioning of the vehicle system can be achieved by detecting all landmarks of the "traffic sign" type.

However, a large amount of traffic is expected on the route during peak hours. An automated vehicle system implementing the method will determine a detection rate for the traffic condition and the planned route that is significantly below 100% due to an expected occlusion rate based on environmental conditions unrelated to the vehicle. However, the system determines a sufficiently good value, and in particular a value above a predetermined threshold, as the expected positioning accuracy. The system thus enables the route selected to be driven with assistance from the automated vehicle system.

In another case, strong precipitation may occur at the same time of day. The method re-determines the detection rate. Due to severe weather conditions, the detection rate is not high enough, which depends in particular on the increased occlusion rate caused by the severe weather conditions. Whereby it is impossible to obtain sufficiently accurate positioning accuracy. Alternative routes can then be selected in the context of the method and evaluated in the context of a further feasibility analysis. The new route passes through less traveled roads and thus has a lower occlusion rate. Alternatively, the driver may be informed via a human machine interface that a longer route has been planned and may now select between automatic travel on that route or manual travel on a shorter route.

Fig. 2 shows an overview of input variables of a parameterized model, in particular a training phase of a neural network.

In this training phase, the following are provided as input parameters in the neural network 14: