Method for determining a driving maneuver of a vehicle

Abstract

The invention relates to a method for determining a driving maneuver of a vehicle, wherein the driving maneuver is determined by means of a cost function, wherein the cost function is set up starting from support points in a solution space, characterized in that the number of support points is dynamically adapted to the width of the solution space.

Description

Prior ArtVarious methods exist for determining driving maneuvers and operating strategies for vehicles. Known hybrid operating strategies are adapted to topologies of hybrid drive trains and essentially take into account the course of a battery charge, influencing the torque distribution between internal combustion engine and electric machine.A coasting assist function maximizes the duration of the coasting state by estimating how long no propulsion torque will be demanded by the engine. Partially predictive data are used for this purpose.Driving strategy functions calculate speed trajectories that are intended to optimize comfort aspects. Fuel efficiency is not taken into account in this case.Known functions which react to vehicles traveling ahead, such as ACC, use information from the environment detection for predicting the behavior of the vehicle traveling ahead.Since the functions known from the prior art were each specifically developed for their specific application and partially influence the same manipulated variable (e.g. setpoint speed), their combination in a vehicle presents a challenge. Thus, a sail assistant would, for example, specify the state "sail" (i.e. essentially decoupling and switching off the internal combustion engine) as the setpoint value, but a cruise control function could simultaneously request an acceleration in order to maintain a preset desired speed.Functions which regulate a setpoint speed as a function of a route topology use an optimization algorithm whose solution space is defined as a speed band. However, fixed speed bands are unsuitable for taking account of non-fixed boundary conditions, such as a speed of a vehicle driving ahead.For energy-efficient planning of driving maneuvers, an optimization algorithm can be used that discretized a given multidimensional solution space and determines an optimum solution with respect to a cost function. In this case, a minimum of the cost function is generally sought. An intermediate step in the optimization is the creation of a cost matrix. This contains all results of the evaluation of the cost function and serves to search for the optimum solution. The discretisation of the solution space and thus that of the cost matrix decisively influences the quality of the result, but also the required resources such as e.g. computing capacity and storage space. There is therefore a need to provide a method which provides an optimum balance between resource requirement and quality of the result.Disclosure of the InventionAccording to the invention, a method for determining a driving maneuver of a vehicle is provided, which dynamically adjusts a discretizing of individual dimensions of a solution space to a width of the solution space. It is particularly advantageous to dynamically adapt the discretisation for each dimension of the solution space to the width of the solution space in the respective dimension.In an advantageous development, the method according to the invention can comprise one, more or all of the steps mentioned below:First, a maximum discretizing of the solution space and an initial declaring of a cost matrix take place.The solution space is divided into segments. A segment is a section, in particular a temporal section, of the solution space.A reduced discretisation per segment is determined. The reduced discretisation is dependent on the width of the solution space. For example, a narrow solution space can be discretized with fewer interpolation points than a wide solution space. In particular, a distance between the support points can be fixedly predefined, so that the number of support points results from the solution space width and the distance of the support points.A cost function for a full combinatorics of the degrees of freedom is calculated for all states with reduced discretisation in the current segment.The calculation is repeated for all segments.In a particularly advantageous development, the method according to the invention can additionally or alternatively comprise one, a plurality or all of the following steps:A processor load is determined for a past time interval, for example for the past 2 seconds.If the processor load thus determined exceeds a first threshold value, the discretizing is reduced further stepwise until the processor load is below the first threshold value. The reduction of the discretisation can in this case take place in particular by reducing the number of support points, in particular by presetting a greater distance between the support points.If the processor load falls below a second, lower threshold value, the discretization is compressed stepwise, in particular by using more support points, in particular by specifying a smaller distance between the support points. In this case, the second threshold value is smaller than the first threshold value.The method according to the invention is advantageously used in an optimizer or optimization module.In an advantageous development, the optimizer determines an optimum driving maneuver, in particular an optimum setpoint speed, in particular an optimum setpoint speed trajectory of the vehicle, starting from the solution space.In an advantageous development, the optimizer takes into account, in particular, a fuel consumption, an absolute acceleration, a deviation from an average speed and a deviation from an average acceleration when ascertaining the driving maneuver or the setpoint speed or the setpoint speed trajectory. In addition, a predicted travel duration and, in the case of a vehicle having an internal combustion engine that can be shut down (for example, in the case of a suitably configured hybrid vehicle), a number of starting processes of the combustion engine are taken into account. In an advantageous development of the method according to the invention, a distance from a vehicle driving ahead is also taken into account by the optimizer, for example by using a penalty function which assigns a penalty value to each time distance from the vehicle driving ahead.The determination of the solution space is preferably carried out by the method described below:Since important elements in road traffic, such as preceding vehicles, cannot be represented by a speed band, the solution space for a setpoint speed of a vehicle is expanded by a temporal dimension.In the temporal dimension, a t-band is advantageously defined, i.e. a band of functions which assign a (required) time to a distance covered or to a position indication. The t-band results from integration of a v-band defined by functions that associate a velocity with a route or position. The v-band can be determined by methods known from the prior art and takes into account in particular a route topology.A vehicle driving ahead can advantageously be taken into account by delimiting the t band with a predicted position-time trajectory of the vehicle driving ahead. In other words, the t-band may be limited such that the conversion of the target speed of the vehicle may not result in a tailgating to the preceding vehicle. In a particularly advantageous embodiment, the position-time trajectory of the vehicle driving in front is determined by integration of a predicted speed trajectory of the vehicle driving in front. In a further preferred embodiment, the speed trajectory of the vehicle driving ahead is determined by means of a combination of sensor data, in particular radar data, and an electronic horizon. The electronic horizon can take into account, for example, that the vehicle driving ahead takes into account a local maximum speed or that the vehicle driving ahead is expected to travel at a reduced speed on steep slopes or in tight curves.The determination of the solution space advantageously comprises the following steps:A speed of a vehicle driving ahead is predicted by assuming its current acceleration as constant for the prediction time based on its current speed. Thus, a velocity trajectory of the preceding vehicle is obtained. By integration, a position-time trajectory of the preceding vehicle is determined from the speed trajectory of the preceding vehicle.The solution space is adjusted to the current speed of the vehicle (ego vehicle) by selecting a beginning of the speed band (i.e. v band, i.e. the band between upper and lower speed raw limit) so that it comprises the current speed of the vehicle, advantageously in particular so that it corresponds to the current speed of the vehicle.A lower time raw limit is determined by integration from the upper speed raw limit.An upper time limit is determined by integrating the lower speed raw limit.Starting from the location-time trajectory of the vehicle driving ahead, the lower time raw limit is modified, in particular partially raised, in such a way that the lower time limit obtained in this way is above the location-time trajectory of the vehicle driving ahead.From the upper time limit and the lower time limit, a solution space is defined which is passed to the optimizer.An exemplary embodiment of the present invention is explained in more detail below with reference to the attached drawings. The following are shown:Brief Description of the DrawingsFIG. 1 shows a schematic signal flow for illustrating an exemplary embodiment of the method according to the invention; FIG. 2 shows a schematic illustration of an upper and lower speed raw limit; FIG. 3 shows a schematic illustration of an upper and lower time limit according to an advantageous refinement of the method according to the invention; FIG. 4 shows a schematic sequence of an exemplary embodiment of the method according to the invention.Embodiments of the InventionFIG. 1 shows a schematic signal flow for illustrating an exemplary embodiment of the method according to the invention. A powertrain management module (10) provides a current vehicle condition and transfers it to a solution space module (13) and an optimization module (14). A horizon module ( 11) provides an electronic horizon and transmits it to the solution space module ( 13). The electronic horizon comprises in particular a gradient of a road section lying ahead and an upper and a lower speed raw limit. Optionally, the electronic horizon may additionally comprise a curve radius. A surrounding area module ( 12) provides information about a vehicle driving ahead and transmits it to the solution space module ( 13) and a conversion module ( 16). The information about the vehicle driving ahead can be, in particular, its speed and the distance from the vehicle (ego vehicle). The solution space module (13) processes the information provided to it and derives therefrom an upper and a lower speed limit and / or an upper and a lower time limit and transmits these together with the gradient of the section ahead to the optimization module (14). A vehicle state estimator (15) estimates the current vehicle state and transmits it to the optimization module (14) and the solution space module (13). Advantageously, the current vehicle condition is taken into account in deriving the upper and lower speed limits, as well as the upper and lower time limits. The optimization module (14) can advantageously comprise a vehicle model. The optimization module ( 14) determines an optimized setpoint speed, in particular a setpoint speed trajectory for the vehicle, and transmits this to the conversion module ( 16). The conversion module converts the setpoint speed trajectory by suitable actuation of the corresponding actuators of the vehicle taking into account the information about the vehicle driving ahead provided by the surroundings module ( 12).In particular, for safety reasons, the information about the vehicle driving ahead can overwrite the implementation of the setpoint speed trajectory, for example when the vehicle driving ahead brakes strongly and its deceleration is not reflected fast enough by the signal chain comprising solution space module ( 13) and optimization module ( 14). Alternatively, the conversion module ( 16) can display the setpoint speed trajectory to a driver of the vehicle (ego vehicle), who then manually performs the conversion of the setpoint speed trajectory. This is advantageous in particular when no direct access to the required actuators by the conversion module is ensured.FIG. 2 shows a schematic representation of an upper ( 43) and lower ( 42) speed raw limit, as can be determined, for example, by methods known from the prior art. A location axis (40) indicates the location of the vehicle. A speed axis ( 41) assigns an upper and a lower speed raw value to each point of the position axis ( 40), so that a representation of the upper ( 43) and lower ( 42) speed raw limits is obtained from all upper and lower speed raw values. The upper and the lower speed raw limits define, for example, a speed solution space for an optimization algorithm.FIG. 3 shows a schematic illustration of an upper (52) and lower (55) time limit according to one aspect of the method according to the invention. A locus axis (50) represents the abscissa, a time axis (51) represents the ordinate. An upper time limit ( 52) can be obtained, for example, by integrating the lower speed raw limit ( 42). A lower time raw limit ( 53) can be obtained by integrating the upper speed raw limit ( 43). A location-time trajectory of a preceding vehicle ( 54) assigns each location a predicted time at which the preceding vehicle will be at that location. The lower time-raw limit ( 53) intersects the location-time trajectory of the preceding vehicle ( 54) in the illustrated example, and therefore the lower time-raw limit ( 53) is modified to obtain the lower time limit ( 55). Here, the lower time raw limit ( 53) is raised if necessary, so that the lower time limit ( 55) is located above the location-time trajectory of the vehicle traveling ahead ( 54). In this case, a safety distance from the vehicle driving ahead which is not to be undershot by the vehicle (ego vehicle) can advantageously be taken into account. The upper time limit (52) and the lower time limit (55) define a solution space for an optimizer.FIG. 4 shows a schematic sequence of an exemplary embodiment of the method according to the invention. In step 100, a maximum discretizing of the solution space and an initial declaring of a cost matrix take place. Step 110 is then carried out.In step 110, the solution space is divided into segments. A segment is a section, in particular a temporal section, of the solution space. Step 120 then follows.In step 120, a reduced discretisation per segment is determined. The reduced discretisation is dependent on the width of the solution space. For example, a narrow solution space can be discretized with fewer interpolation points than a wide solution space. For this purpose, a processor load is determined for a past time interval, for example for the past 2 seconds. If the processor load thus determined exceeds a threshold value, the discretisation is further reduced stepwise until the processor load is below the threshold value. The reduction of the discretisation can in this case take place in particular by reducing the number of support points, in particular by presetting a greater distance between the support points. If the processor load falls below a second, lower threshold value, the discretization is compressed stepwise, in particular by using more support points, in particular by specifying a smaller distance between the support points. Step 130 is then carried out.In step 130, a cost function for a full degree of freedom combinatorics is calculated for all states with reduced discretization in the current segment. Step 140 then follows.In step 140, the cost function of step 130 is optimized. On the basis of the optimization of the cost function, a driving maneuver is calculated; in particular, the driving maneuver can be a setpoint speed, in particular a setpoint speed trajectory of a vehicle. In a preferred development, the vehicle is operated with the calculated driving maneuver. If the driving maneuver is a setpoint speed trajectory, the vehicle is operated in such a way that its speed follows the setpoint speed trajectory.Steps 100 to 140 are preferably carried out in the optimization module (14).

Claims

Method for determining a driving maneuver of a vehicle, wherein the driving maneuver is determined by means of a cost function, wherein the cost function is set up starting from interpolation points in a solution space, characterized in that the number of interpolation points is dynamically adapted to the width of the solution space.Method according to Claim 1, characterized in that the number of interpolation nodes is adapted as a function of resources of a computing unit of the vehicle.Method according to Claim 2, characterized in that the number of nodes is reduced if a processor load of the arithmetic unit exceeds a first threshold value.Method according to either of Claims 2 and 3, characterized in that the number of nodes is increased if the processor load of the arithmetic unit falls below a second threshold value.Method according to Claims 3 and 4, characterized in that the first threshold value is above the second threshold value.Device, configured to carry out the method according to one of Claims 1 to 5.A computer program that causes a computing unit to perform the method according to any one of claims 1 to 5 when the computer program is executed by the computing unit.A storage medium on which the computer program according to claim 7 is stored.

Images