Determining a lane and lateral control for a vehicle

Abstract

Method for a vehicle (10), the method comprising: - capturing image data (6) by means of a camera (2) of the vehicle (10), - fusing the image data (6) with route data (1), which comprise information about a route on which the vehicle (10) is traveling, and creating fused data (7), wherein the fused data (7) are based more on the image data (6) in the vicinity of the vehicle (10), while the fused data (7) are based more on the route data (1), the further away from the vehicle (10) the location of the route is located, the properties of which the fused data (7) describe, and - Determining a lane for the vehicle (10) depending on the fused data (7).

Description

[0001] The present invention relates, on the one hand, to the determination of a lane of a vehicle and, on the other hand, to the determination of a steering angle for lateral control of a vehicle.

[0002] WO 2011 / 131 165 A1 discloses the determination of the road course for a motor vehicle. Sensor data from an environmental sensor system, which can display both a camera and a digital map with GPS, is provided to an evaluation unit. Further processing of the data obtained by the environmental sensor system is not described.

[0003] Furthermore, DE 10 2010 033 530 A1 discloses a method for lane centering of a vehicle, wherein the dynamic model for the lateral control only includes state variables that are available as a measured variable for state feedback.

[0004] Furthermore, DE 10 2011 107 196 A1 discloses a method and a system for vehicle lateral control using image data from a front and a rear camera.

[0005] Based on this prior art, the present invention aims to further process sensor data before it is used to determine a lane or to control the lateral movement of a vehicle.

[0006] According to the invention, this object is achieved by a method for a vehicle according to claim 1 and by a device according to claim 8. The dependent claims define preferred and advantageous embodiments of the present invention.

[0007] Within the scope of the present invention, a method for a vehicle is provided which comprises the following steps: • Capturing image data using a vehicle camera. • Merging this image data with route data, which includes information about a route on which the vehicle is traveling or stationary, in order to then create fused data through this fusion. The route data particularly includes curvature information and / or elevation data of the route and can be based on navigation data and stored road maps. The route data enables a preview of the road geometry ahead of the vehicle. The curvature information particularly includes information about a curvature or a curvature measure, for example, of a center line of the route. For example, the course of the route can be uniquely described using clothoids, with the route data providing curvature information for each location of these clothoids. The elevation data includes, for example, information about the elevation of certain points (e.g.the center line or the location points of the clothoid) of the route. The fused data describe, in particular, properties of the route, such as dimensions and a course of the route, so that a curvature or a curvature of the route or a lane of the vehicle within the route can be derived from the fused data. The fused data is based more on the image data near the vehicle, while it is based more on the route data at greater distances from the vehicle. In other words, the further away from the vehicle the point on the route whose properties the fused data describe is located, the more the fused data is based on the route data. • Determining the lane based on the merged data. The lane essentially corresponds to a lane tube, which has the same width as the vehicle and describes the path the vehicle is expected to take along the route or is planned to take. However, it is also possible for the lane to essentially correspond to a driving line, as is the case with a single-lane model.

[0008] The fusion of image data and route data to determine the lane represents an advantageous further processing of sensor data (i.e. the image data), thus solving the above-mentioned problem. By determining the lane based on the fused data, the accuracy of the image data in a close area in front of the vehicle is essentially combined with the accuracy of the route data in a more distant area in front of the vehicle. This combines the advantages of camera accuracy with the forward projection of predictive route data. The lane determined according to the invention can be used for automatic lateral control, but also for a lane keeping assistant. Automatic lateral control of a vehicle means that the driver does not have to intervene in the steering or monitor the steering. This means that automatic lateral control steers the vehicle autonomously.In contrast, the driver must monitor a lane departure warning system, which allows the driver to intervene in the steering. This means that the lane departure warning system assists the driver in steering the vehicle and only steers the vehicle independently if the driver does not intervene.

[0009] According to one embodiment of the invention, determining the lane comprises determining a curvature or a curvature profile of the route based on the fused data. In addition, in this embodiment, vehicle status data is measured. Based on the curvature and the measured status data, a steering angle of the vehicle is determined. A steering angle is understood, in particular, to be the average wheel steering angle of the steered axle of the vehicle in the force-free state.

[0010] This embodiment can be used, for example, for automatic lateral control of the vehicle.

[0011] The status data may include at least one of the following vehicle values: • The vehicle's sideslip angle. The sideslip angle is the angle between the vehicle's longitudinal axis and the vehicle's velocity vector at the vehicle's center of gravity. • The yaw rate of the vehicle. The yaw rate of the vehicle refers to the angular velocity of the vehicle's rotation around its vertical axis. The vertical axis or yaw axis refers to the vertical axis of the vehicle, or more precisely, the vehicle's fixed coordinate system. • A lateral deviation of the vehicle. The lateral deviation is the distance from the perpendicular of the vehicle's center to the lane perpendicular to the lane centerline. • A vehicle's yaw angle error. The yaw angle error corresponds to the difference between the vehicle's yaw angle and the lane marking yaw angle. This means that the yaw angle error corresponds to the angle of inclination between the vehicle's longitudinal axis and the lane marking. • A longitudinal speed of the vehicle. The longitudinal speed of the vehicle corresponds to the speed at which the vehicle moves in the direction of travel.

[0012] According to the invention, an actual steering angle of a steering system of the vehicle can be adjusted according to the steering angle.

[0013] The steering angle can be viewed as a target steering angle, whereby, for example, a vehicle's steering control system strives to ensure that the actual steering angle of the vehicle follows this target steering angle as closely as possible. In other words, according to the invention, the vehicle's steering is adjusted with respect to the steering angle determined according to the invention.

[0014] In particular, the state data is measured for a current one of several time steps. For the future of these time steps, the state data is determined depending on a model of the lateral dynamics of the vehicle, starting from the state data measured for the current time step and the curvature determined according to the invention. In other words, the state variables are predicted in discrete time steps over the prediction horizon, for example by means of a controller. If this prediction horizon comprises 50 time steps, for example, this leads to a look ahead of 2 s with a cycle time of 40 ms. This means that according to the invention the state variables are calculated for the next 50 time steps or 2 s. Of course, the number of time steps, the cycle time and thus the look ahead can be set as desired. The state variables orState data is measured in each current time step and serves as the starting point for the prediction.

[0015] By determining the future state data of the vehicle, the determination of the steering angle can advantageously be carried out better than if the future state data of the vehicle are not known.

[0016] According to a further embodiment of the invention, the steering angle is determined as a manipulated variable with the aid of a controller based on a target function depending on the state data and the curvature. The curvature is viewed as a disturbance variable for the controller. The disturbance variable or curvature is provided for all time steps (i.e. for the current and future time steps up to the prediction horizon), in particular as a vector. This predictive vector uses the curvature to depict the future route or the part of the route to be traveled by the vehicle in the future. In other words, the disturbance variable provided is the curvature of the route known over the prediction horizon of the controller, which is obtained from the fused data.

[0017] By using the curvature, which is also known for future time steps, as the disturbance variable, the invention makes it possible to react even before a lateral deviation from the center of the lane occurs. While, for example, a classic PID controller first requires a control deviation (in the case of a lane departure warning system, a lateral deviation from the center of the lane) to generate a reaction, the model predictive controller according to the invention can apply a compensating manipulated variable (i.e., a steering angle) even before the known disturbance occurs. Since, according to the invention, the camera's preview (i.e., the image data) is essentially expanded with the help of the predictive route data, the control quality is improved.

[0018] The model predictive controller implements model-based predictive control, which uses a model of the dynamic behavior of the controlled system during operation for control. This model enables the prediction of system behavior and can thus be used to adjust the manipulated variable even before a control deviation occurs.

[0019] The objective function can be used to determine a course of the steering angle over the time steps in such a way that, on the one hand, a temporal change in the steering angle is minimized and, on the other hand, a lateral deviation of the vehicle and a yaw angle error of the vehicle are minimized.

[0020] The control system according to the invention is therefore based on the optimization of the objective function. The objective function takes into account, on the one hand, a predicted center deviation or lateral deviation of the vehicle and, on the other hand, the predicted yaw angle error, which can be calculated from the determined manipulated variable curve or steering angle curve using the model. The steering angle curve determined using the objective function is weighted such that the output manipulated variables, i.e. the temporal changes in the steering angle, are as small as possible. The aim of the control system is therefore to bring the lateral deviation and the yaw angle error to zero with the least possible control input, i.e. with the smallest possible temporal changes in the steering angle.

[0021] In particular, the controller uses a single-track model to determine the state data in the future time steps depending on the currently measured state data and the curvature.

[0022] The single-track model corresponds to the single-track model presented by Dr. Riekert and Dr. Schunk in 1940. This single-track model uses the following simplifications and assumptions: • The overall center of gravity of the vehicle is at road level, so that even when cornering quickly there are no wheel load differences and no rolling movements. • The wheel contact points are merged axle by axle so that the vehicle only drives on one lane or line. • A change in the vehicle's driving speed is treated quasi-stationary, so that no tire circumferential forces occur. • Yaw and swim motion are the only degrees of freedom. • Lifting and pitching movements do not occur. • Any lateral acceleration occurring is less than 0.4 g. • As a result of a slip angle, neither caster nor restoring torque occurs.

[0023] Within the scope of the present invention, a device for a vehicle is also provided, wherein the device comprises a camera for capturing image data and control means. The control means are configured to generate fused data by fusing the image data with route data, which includes information about a route on which the vehicle is traveling. Furthermore, the control means are configured to determine a lane of the vehicle depending on the fused data.

[0024] The advantages of the device according to the invention essentially correspond to the advantages of the method according to the invention, which have been explained in detail above, so that a repetition is omitted here.

[0025] In particular, the device comprises sensor means for measuring or recording status data of the vehicle.

[0026] Finally, within the scope of the present invention, a vehicle is provided which comprises a device according to the invention.

[0027] The present invention is particularly suitable for use in motor vehicles. Of course, the present invention is not limited to this preferred application, as the present invention can also be used in aircraft.

[0028] In the present invention, image data and predictive route data are fused, which leads to an expanded route forecast compared to the prior art. This allows insights into future route routing to be determined, which in turn enables better adaptation of the control strategy for a lane departure warning system or for determining the steering angle. According to the invention, an online-capable, model-predictive controller can be used to calculate the manipulated variable (the steering angle), which takes into account the obtained predictive information (i.e., the fused data).

[0029] The present invention enables, for example, adaptive lane centering in a lane keeping assistant, which supports the driver as smoothly and almost continuously as possible. Compared to the prior art, particularly winding road sections can be negotiated with a significant increase in comfort. The invention allows predictive human driving behavior to be modeled much better than with a conventional PID controller. Thus, the present invention provides an immediate increase in comfort and confidence when using the device according to the invention.

[0030] In the following, the present invention is described using preferred embodiments of the invention with reference to the figures. In Fig. 1 shows a control system according to the invention. In Fig. 2 shows the flow chart of a method according to the invention. In Fig. 3 schematically shows a vehicle according to the invention with a device according to the invention.

[0031] In the Fig. In the inventive control system shown in Figure 1, the image data 6 captured by a camera 2 are merged 15 with predictive route data, which is stored, for example, in a navigation system, to form fused data 7. From this fused data 7, the curvature κ of the route or roadway on which the vehicle is traveling is determined. This curvature κ is considered a disturbance variable 14.

[0032] In addition to the disturbance variable 14, the model predictive control 12 is provided with state variables 13, which also include a longitudinal velocity v xof the vehicle. The state variables 13 are periodically measured or recorded by the vehicle's sensor means for the current time step and then determined for future time steps, in particular depending on the measured state variables 13 and the curvature κ using a single-track model. Depending on the disturbance variable 14 or the curvature κ and depending on the state variables 13, the control 12 determines a target steering angle δ. The control 12 according to the invention strives to keep the changes in the manipulated variable or the steering angle δ as small as possible. The aim of the control 12 is to determine the lateral deviation e y and the yaw angle error e ψ to be minimized as much as possible. The target steering angle δ is fed to a steering system 11 of the vehicle, which adjusts the actual steering angle to the target steering angle δ as quickly as possible.

[0033] In addition to the illustrated control 12, a control can also be used according to the invention that outputs a difference from the actual steering angle instead of the target steering angle δ. In this frequently used control, the actual steering angle must be supplied to the control as an input variable.

[0034] In Fig. 2 shows the flow chart of a method according to the invention.

[0035] In the first step S1, image data 6 is captured by a camera 2. In step S2, this image data 6 is fused with predictive route data 1 to form fused data 7. In step S3, state data 13 of the vehicle is captured, while in step S4, the curvature κ of the route is determined based on the fused data 7. Finally, in step S5, the steering angle δ is determined depending on the curvature κ and the state data 13.

[0036] In Fig.Figure 3 schematically shows a vehicle 10 according to the invention with a device 20 according to the invention. The device 20 according to the invention comprises, in addition to control means 3, a camera 2, sensor means 4, and a steering system 11. The control means 3 create fused data 7 from image data 6 acquired by the camera 2 and predictive route data 1. Depending on this fused data 7 and the status data 13 acquired by the sensor means 4, the control means 3 determines a steering angle, which is fed to the steering system 11. List of reference symbols 1 predictive route data 2 cameras 3 Tax resources 4 Sensor means 6 Image data 7 merged data 10 vehicles 11 Steering 12 model predictive control 13 state variables 14 Disturbance 15 Merger 20 Device δ steering angle β slip angle ψ yaw rate e y Lateral deviation e ψ Yaw angle error κ curvature v x Longitudinal speed

Claims

[1] A method for a vehicle (10), the method comprising: - capturing image data (6) by means of a camera (2) of the vehicle (10), - fusing the image data (6) with route data (1), which comprise information about a route on which the vehicle (10) is traveling, and creating fused data (7), wherein the fused data (7) are based more on the image data (6) in the vicinity of the vehicle (10), while the fused data (7) are based more on the route data (1), the further away from the vehicle (10) the location of the route is located, the properties of which the fused data (7) describe, and - Determining a lane for the vehicle (10) depending on the fused data (7). [2] Method according to claim 1, characterized by , that determining the lane comprises determining a curvature (κ) of the route depending on the fused data (7), that the procedure further includes: Measuring condition data (13) of the vehicle (10), and Determining a steering angle (δ) of the vehicle (10) depending on the curvature (κ) and the state data (13). [3] Method according to claim 2, characterized by that an actual steering angle of a steering system (11) of the vehicle (10) is set according to the steering angle (δ). [4] Method according to claim 2 or 3, characterized by , that the state data (13) are measured for a current one of several time steps, and that the state data (13) for future time steps are determined depending on a model of a lateral dynamics of the vehicle (10) based on the state data (13) of the current time step and the curvature (κ). [5] Method according to one of claims 2-4, characterized by , that the steering angle (δ) is determined as a control variable by means of a controller (12) based on a target function depending on the state data (13) and the curvature (κ), and that the curvature (κ) is a disturbance for the controller (12). [6] Method according to claims 4 and 5, characterized by that by means of the objective function a course of the steering angle (δ) over the time steps is determined in such a way that a temporal change of the steering angle (δ) is minimized and that a lateral deviation (e y ) of the vehicle (10) and a yaw angle error (e ψ ) of the vehicle (10) is minimized. [7] Method according to claim 5 or 6, characterized by that the controller (12) works with a single-track model to determine the state data (13) in the future time steps depending on the currently measured state data (13) and the curvature (κ). [8] Device for a vehicle (10), the device (20) comprising: a camera (2) for capturing image data (6), and Control means (3) for fusing the image data (6) with route data (1), which comprise information about a route on which the vehicle (10) is traveling, in order to determine fused data (7), and for determining a lane for the vehicle (10), wherein the fused data (7) is based more on the image data (6) in the vicinity of the vehicle (10), while the fused data (7) is based more on the route data (1) the further away from the vehicle (10) the location of the route is located, the properties of which the fused data (7) describe, wherein the device (20) is designed to carry out the method according to one of claims 1-7. [9] Device according to claim 8, characterized by that the device (20) comprises sensor means (4) for measuring status data (13) of the vehicle (10).

Images