Method for generating laterally offset tracks

Abstract

A method for generating a lateral offset trajectory for an at least partially automated mobile platform is proposed, comprising the steps of: providing a target lateral offset, inverting a provided dynamic model of the mobile platform, providing at least one constraint on a system quantity of the dynamic model for determining the lateral offset trajectory, determining a time series of lateral offset trajectory points for the inverted dynamic model based on the at least one constraint on the system quantity and the target lateral offset as input signals using a state variable filter, wherein each point of the time series of the lateral offset trajectory is analytically determined using a state variable filter, and determining a time series of values ​​of at least one adjustment quantity for the mobile platform for generating the lateral offset trajectory using the inverted dynamic model and the time series of lateral offset trajectory points as input signals for the inverted dynamic model.

Description

[Technical Field]

[0001] [Background technology]

[0002] Emergency evasive maneuvers to avoid accidents in road traffic can usually only be initiated at a very late stage, that is, just before a potential collision, which consequently imposes a high level of demands on the dynamics of the maneuver. On the one hand, in many situations, the objects involved can only be detected by sensors at a very late stage, for example, when a pedestrian emerges from behind an object onto the road. On the other hand, the sum of the uncertainty of the sensor and the uncertainty of the subsequent course of the detected situation is usually very high, for example, because it is difficult to predict sudden changes in the motion of a crossing pedestrian. Therefore, it is worthwhile to make full use of the dynamics used, especially vehicle dynamics. [Overview of the Initiative]

[0003] When planning a configuration for driver assistance functions to intervene in the vehicle's lateral dynamics, an appropriate vehicle trajectory is usually selected to ensure an optimal outcome in terms of intervention comfort, in which case, for example, lateral acceleration limits and / or lateral motion limits are controlled. To maintain controllability limits, the trajectory is selected so that the rough maximum yaw rate matches the controllability yaw rate limit. The result obtained from this is a yaw rate progression with a rough maximum value.

[0004] While driver manual intervention during automatic control interventions in vehicle dynamics can potentially introduce errors, ensuring the controllability of the vehicle through such manual intervention requires maintaining the limits on dynamic interventions detected within the framework of experimental studies on various actuators and vehicles. These limit values related to the actuator have been found to be rewritable in terms of vehicle dynamics as limit values independent of the respective vehicle systems, for example, limit values for yaw rate.

[0005] When developing such driver assistance functions, typically, actuator limit values specific to the system are derived from controllability values independent of the system, such as yaw rate. However, these limit values of the actuator that set the corresponding adjustment amounts and other restrictions on the system behavior, which can be described by so-called state quantities as necessary, are typically not directly considered in trajectory planning, but the function is realized additionally for the first time by an appropriate application of trajectory planning. This increases the application cost. When it is necessary to derive actuator-specific adjustment amount limit values from yaw rate limit values, directly considering the yaw rate limit values during the implementation of the function can be a problem depending on the situation. This typically cannot be done without simplification from differential or non-linear relationships. However, such a simplification can cause a significant deviation between the assumed vehicle dynamics and the actual vehicle dynamics. From this point, it may exceed the controllability limit value during the yaw rate of operation, and at this time, such an exceedance of the controllability limit value must be prevented by further increasing the additional application cost.

[0006] The development of such driver assistance functions that enable direct consideration of restrictions within the scope of trajectory planning is typically based on optimization methods. Therefore, usually, due to the high required computing power, it is only conditionally possible to incorporate it into the controller of a mass-produced mobile platform.

[0007] As a result, depending on the current level of technology, the potential for accident avoidance remains unutilized, because the maximum yaw rate or yaw rate limit is achieved at only one point in the orbital course.

[0008] From the perspective of the present invention, a method for generating a lateral offset trajectory, a method for providing a control signal, a method for using this method, a control device, a computer program product, and a machine-readable storage medium are proposed according to the configuration of the independent claims. Advantageous configurations are the subject of the dependent claims and the following description.

[0009] Throughout this specification of the present invention, the order of the method steps is shown in order to facilitate understanding of the method. However, those skilled in the art will know that many of the method steps can be carried out in other order to achieve equivalent or corresponding effects. In this sense, the order of the method steps can be changed as appropriate. Some constituent elements are provided with substitutes to improve readability or to make the correspondence more unambiguous, but this does not imply that there are any particular constituent elements.

[0010] A method for determining a lateral offset trajectory for at least partially automated mobile platforms has been proposed, and this method comprises the following steps. In one step, a target lateral offset is provided. In a further step, the provided dynamic model of the mobile platform is inverted. In a further step, at least one limit on the system quantity of the dynamic model is provided to determine the lateral offset trajectory. In a further step, a time series of lateral offset trajectory points is determined for the inverted dynamic model using a state-variable filter based on the at least one limit on the system quantity and the target lateral offset as an input signal, in which case each point of the time series of the lateral offset trajectory is analytically determined. In a further step, a time series of values ​​for at least one adjustment quantity for the mobile platform is determined to generate the lateral offset trajectory, using the inverted dynamic model and the time series of lateral offset trajectory points as an input signal for the inverted dynamic model.

[0011] From one perspective, state-variable filters have unlimited demand dynamics.

[0012] In other words, this method uses an inverted dynamic model to determine the amount of adjustment that will later be used as an input signal for a real dynamic system, such as a mobile platform. The lateral offset trajectory can be implemented by the mobile platform based on a time series of at least one adjustment amount.

[0013] Analytical determination of each point in a time series of a lateral offset trajectory using a state-variable filter should be interpreted broadly in relation to the constituent elements of analytical determination, and should be distinguished in particular from approximation and optimization methods. In particular, analytical determination involves determining the integral at least during the operating time and at each time point with a fixed step size. That is, analytical determination does not involve inductive determination of individual points in the sense of optimization methods. In particular, analytical determination involves the execution of n integrator chains, where n represents the system order characterizing the dynamic model. Such analytical determination of points in a time series of a lateral offset trajectory can be performed, for example, using a computer and correspondingly organized computer programs.

[0014] In particular, using a specific time series of values ​​for at least one adjustment variable, target setting for other control circuits that control the mobile platform can generate a lateral offset trajectory by setting, for example, a steering angle or steering angular velocity in response to input control.

[0015] This method, based on a differentially flat system and a state-variable filter for describing a dynamic model of a mobile platform, enables online trajectory planning that directly considers at least one system quantity limit, in particular adjustment quantity limits and / or state quantity limits, which can be directly considered in online trajectory planning.

[0016] In this case, the state-variable filter for online trajectory planning is the planar coordinate w zThe time series of the lateral offset trajectory points is determined by differentiating the target lateral offset n times with respect to time. Furthermore, this method can be used to provide a time series of values ​​for at least one adjustment variable, thereby enabling flatness-based input control and / or adjustment. In other words, this method allows at least one adjustment variable to be supplied in the form of a continuous value, and this continuous value can be used to control and / or adjust the mobile platform.

[0017] By using a differentially flat system to describe dynamic models and, in particular, state-variable filters, all system theoretical quantities can be parameterized, thereby allowing the dynamic model of a mobile platform to be inverted to construct control and / or tuning for the mobile platform. This is particularly applicable to controllable linear dynamic models of mobile platforms, and especially to nonlinear dynamic models if the dynamic model is differentially flat. Note that controllable linear models are always differentially flat, and if they are nonlinear, the flatness characteristics must be demonstrated separately.

[0018] In advantageous embodiments, this method allows for the direct consideration of all existing state and adjustment limit constraints of the mobile platform's dynamic model as a component of online trajectory planning, and / or provides flatness-based input control to maximize the utilization of automated avoidance functions. Consequently, closed planning of avoidance trajectories or lateral offset trajectories can be performed by optimally utilizing the controllability limits of state variables such as yaw rate and / or adjustment variables such as steering angular velocity. Additionally, further state variable limits and adjustment variable limits can be achieved, and for this purpose, optimization problems do not need to be solved during operation time, and therefore this method is compatible with the computing power of the set of controllers installed to implement this method.

[0019] In other words, this method yields the following: - Optimizing the use of avoidance functions by correspondingly generating lateral offset trajectories by making maximum use of the controllability limits of state variables such as yaw rate and / or adjustment variables such as steering angular velocity on the mobile platform. - A model-based closed approach to directly and systematically consider state variable limits and adjustment variable limits without the need to numerically solve the optimal control problem during operation time. -Since existing state variable limits and adjustment variable limits are already taken into account when a lateral offset trajectory occurs, there is a guarantee that the calculated trajectory can be changed. - Because calculations are performed using non-recursive algorithms, the numerical solution of the optimal control problem online can be omitted, thus avoiding a deterministic increase in computation time or real-time capacity. -If there is a change in the situation in which the avoidance trajectory has been adapted, this can be directly considered in the next calculation step using this method, enabling adaptive behavior with respect to the necessary lateral offset, the trajectory generated online, - An additional advantage is that the dynamic model used to generate the lateral offset trajectory can directly provide actuator-specific input control components for realizing avoidance maneuvers.

[0020] In this method, the dynamic model of the mobile platform can be described using a linear model described by a state-space representation given by equations 4.1 and 4.2.

[0021] TIFF2023545179000002.tif17170

[0022] Here, in equation 4.1 x (t)∈R n This represents the state vector of the dynamic model, TIFF2023545179000003.tif9170 is x This represents the time derivative of (t). u(t)∈R 1represents the input quantity of the dynamic model, where y(t)∈R 1 This represents the output amount.

[0023] A represents the system matrix that characterizes the dynamic behavior of the dynamic model. b This represents the system's input vector, c y represents the output vector of the corresponding system or dynamic model, and y represents the output quantity. The dynamic model of a mobile platform may be a dynamic model of the lateral dynamics of the mobile platform in particular, and this dynamic model is characterized in explicit function form by equations 4.3 and 4.4.

[0024] TIFF2023545179000004.tif101170

[0025] Equations 4.3 and 4.4 for the model of the lateral dynamics of the mobile platform are based on an extended linearized single-orbit model, in which case the lateral dynamics model is, in addition to the standard single-orbit model with floating angle (β) and yaw rate d / dtΨ(t) as state variables, corresponding to equations 4.3 and 4.4, with yaw angle Ψ(t) and lateral offset y L (t) is an extension of the state variable for steering angle δ(t). To describe the lateral dynamics model, this method uses the steering angular velocity instead of the steering angle δ(t) to generate a lateral offset trajectory. TIFF2023545179000005.tif10170 may be selected as the adjustment amount. That is, in the model, the steering angle δ(t) describes one state of the system, and the steering angular velocity TIFF2023545179000006.tif10170 characterizes one adjustment amount for the system.

[0026] The parameters in Matrix A represent the front tire slip stiffness c v , rear tire slip rigidity c h, let the mass of the mobile platform be \(m\), the vehicle speed be \(v\), and the distance from the rear axle of the mobile platform to the center of gravity be \(I\) h , the front axle \(I\) of the mobile platform v The distance from to the center of gravity is \(I\) v , and the moment of inertia of the mobile platform about the height direction axis is \(J\) z is represented by.

[0027] By inverting the dynamic model of the mobile platform, for example, directly from the preset output force process of the non-inverted model such as the time dependence of the lateral offset \(y\) L (t), the input force process of the non-inverted model such as the steering angular velocity TIFF2023545179000007.tif10170 can be calculated.

[0028] To invert the dynamic models 4.1 and 4.2, first convert this dynamic model to (differentially) planar coordinates corresponding to the linear control canonical form or equations 4.9 and 4.10 using the transformation rules 4.5 to 4.8 z Then, for the model of lateral dynamics corresponding to equations 4.3 and 4.4, a system of simultaneous equations of equations 4.11 and 4.12 can be obtained.

[0029] TIFF2023545179000008.tif28170 I

[0030] Equations 4.1 and 4.2 can be converted to planar coordinates and expressed according to the following equations 4.9 and 4.10.

[0031] ]> TIFF2023545179000009.tif15170

[0032] Here, the quantity \(z\) represents the variable for the (flat) output of the dynamic model of lateral dynamics. The transformation matrix \(T(4.5b, 4.5a)\) is the original coordinate xThe state described is in planar coordinates z Used to convert to Matrix A. R This characterizes the system of dynamic models in planar coordinates or in control standard forms. b R This describes the input vector of the transformed system. c R 'x' describes the output vector, 'y' describes the output quantity, and 'u' describes the corresponding input quantity of the dynamic model in planar coordinates.

[0033] TIFF2023545179000010.tif18170

[0034] Here, Q S is the controllability matrix, β is the scaling factor, and t is β times the last row of the inverted controllability matrix.

[0035] TIFF2023545179000011.tif100170

[0036] Here, z (x) is the (x)-th time derivative of z(t).

[0037] The parameterization selected according to equation 4.3 of the mobile platform's lateral dynamics model allows for several factors A during the transformation. R and c R T This becomes zero according to equations 4.13 and 4.14, which clearly reduces the complexity of solving the system of equations.

[0038] TIFF2023545179000012.tif24170

[0039] This allows the last line of the extended single-trajectory model of the lateral dynamics of the mobile platform to be simplified to 4.15, which can then be inverted in plane coordinate z according to equation 4.16.

[0040] TIFF2023545179000013.tif24170

[0041] In this way, for example, steering angular velocity The progression of adjustments like those in TIFF2023545179000014.tif10170 can be directly calculated using the corresponding derivative of the planar coordinate z of the dynamic model according to equation 4.16.

[0042] To determine the time series of the lateral offset trajectory point, the target lateral offset w in spatial coordinates is determined. y Corresponding to the target quantity, a reference signal w in planar coordinates z It can be converted to the following. The corresponding conversion is performed by filtering corresponding to equation 4.10, and the filtering equation 4.17 is obtained.

[0043] TIFF2023545179000015.tif61170

[0044] To enable the online execution of a lateral offset trajectory, the target lateral offset trajectory w y (t) may be set, and the target lateral offset trajectory will later be transformed in accordance with the filtering equation 4.17.

[0045] In particular, to generate a lateral offset trajectory that can be implemented online, for example, a target lateral offset value w y and target time t end A target lateral offset may be set, which will later be transformed in accordance with the filtration equation 4.17.

[0046] For online trajectory planning, the time series of lateral offset trajectory points for the inverted dynamic model is used, according to the input control equation 4.16, with a target lateral offset w z The nth time derivative of is determined from filter 4.17 using a state-variable filter.

[0047] Therefore, a state-variable filter, which can be described by equation 4.18 and has the order (n=5) of the extended single-orbit models 4.3 and 4.4, may be used in planar coordinates. In other words, the following applies: The state-variable filter plans the time series of lateral offset track points for the inverted extended single-track model. Therefore, the state-variable filter has the same system order (n=5 in the case considered) for the assumed section model or vehicle model. Furthermore, according to 4.18, no further model quantities / information from the single-track model are input into the unrestricted state-variable filter.

[0048] TIFF2023545179000016.tif10170

[0049] Corresponding to this, it becomes as follows:

[0050] TIFF2023545179000017.tif50170

[0051] Setting the required dynamics for trajectory planning using a state-variable filter is done by the filter coefficient (a F,0 ,…a F,4 This can be done by appropriately setting the parameters, for example, by full state feedback depending on the level of technology, or by designing a linear quadratic controller (LQR) under assumed system constraints, taking interval dynamics into account, i.e., taking mobile platforms into account. By setting the required dynamics for the state-variable filter, the generated lateral offset trajectory can be adapted to the behavior of the mobile platform.

[0052] In other words, for online trajectory planning, target lateral offset w y (t) or w z Under the setting of (t), the trajectory evolution z for the flatness-based inverted dynamic model. *(t) is the flat output of the dynamic model and its time derivative, i.e. z *(1) ,…,z *(5) ; z * =[ z * ,z *(1) ,z *(2) ,z *(3) ,z *(4) ] In response to this, the time series of the lateral offset trajectory point is calculated, and this trajectory progression corresponds to the target adjustment amount progression in equation 4.16. This is used as an input quantity for the inverted model used to calculate the adjustment amount progression after calculating TIFF2023545179000018.tif10170.

[0053] The determination of the time series of lateral offset orbital points for an inverted dynamic model can be performed using polytopic state constraints, such as those shown below, based on state system quantity constraints, such as yaw rate constraints, and is therefore a component of orbital planning.

[0054] Such quantities of k-dimensional polytopic state constraints correspond to equation 4.19, TIFF2023545179000019.tif24170 describes the constraints imposed by multiple boundaries (multiple hyperplanes) within state space. In this case, all state vector quantities lie on exactly one boundary. This is provided by TIFF2023545179000020.tif10170.

[0055] And F x This is a matrix that defines a linear combination of multiple restricted states. x (t) is the state vector in the original coordinates, vector g x The values ​​shown indicate the respective limits.

[0056] For the yaw rate constraint example considered here, the polytopic state constraint is reduced to a box state constraint according to equations 4.20 and 4.21. TIFF2023545179000021.tif8170 represents the limit of the state. This directly refers to TIFF2023545179000022.tif9170.

[0057] TIFF2023545179000023.tif19170

[0058] The constraints can be transformed into planar coordinates according to equations 4.22 and 4.23 by applying the transformation provisions to planar coordinates, which include the transformation matrix T described above for the transformation.

[0059] TIFF2023545179000024.tif18170

[0060] The selected parameters of system model 4.3 will result in several factors being applied during the conversion. TIFF2023545179000025.tif9170 corresponds to 4.24 and becomes 0.

[0061] TIFF2023545179000026.tif11170

[0062] Considering constraints within the scope of trajectory planning is the highest derivative of its flat output, i.e., the nth derivative z. * This is done with a limiting. The highest derivative z *(5) To generate a dependency between the state and the state constraint, the Taylor series over a short time length Δt is used to represent the state z * Perform an approximation. z *(5) The required limiting is the upper limit of the yaw rate. TIFF2023545179000027.tif9170 and lower limit For TIFF2023545179000028.tif8170, the calculation can be performed using equations 4.25 and 4.26.

[0063] TIFF2023545179000029.tif32170

[0064] Here, A P and b P This is equation 4.26b

[0065] This is provided by TIFF2023545179000030.tif38170.

[0066] In order to ensure the feasibility of an avoidance trajectory at the trajectory planning stage, an additional adjustment limit may be considered as a limit on the system quantity when determining the time series of lateral offset trajectory points with respect to steering angular velocity.

[0067] The limit on steering angular velocity as a system adjustment amount is the adjustment amount limit provided in plane coordinates according to the last line of system equation 4.3, corresponding to equations 4.15, 4.27, and 4.28. TIFF2023545179000031.tif9170 and The maximum allowable highest derivative z of a flat output including TIFF2023545179000032.tif9170 * This is done by calculating it.

[0068] TIFF2023545179000033.tif24170

[0069] A particular advantage of this method for generating lateral offset trajectories is that state variable limiting and adjustment variable limiting can be considered as components for generating the lateral offset trajectory. This allows for adherence to the controllability limits mentioned above, and enables the incorporation of state variable limits such as yaw rate.

[0070] In this way, the consideration of limitations on state variables and / or adjustment variables within the scope of orbital planning is taken into account by the highest derivative of its flat output, i.e., the nth derivative z * This can be done by limiting it.

[0071] In accordance with the methods described here, further state variable limits and adjustment variable limits can be implemented. In this case, to perform online trajectory planning, the system of differential equations 4.18 can be solved by numerical integration, for example, with a fixed step size during operation time, in which case, as already mentioned, the highest-order derivative of the flat output may be restricted.

[0072] In other words, the time series of lateral offset orbit points can be viewed as an inverted dynamic model with a state-variable filter, based on a limit of at least one system quantity corresponding to the switching system, because the saturation element 4.29 can be parameterized using the proposed limit functions 4.25, 4.26, 4.27, and 4.28 (see Figure 3 as a favorable embodiment).

[0073] TIFF2023545179000034.tif27170

[0074] Here, TIFF2023545179000035.tif9170 describes the unrestricted required dynamics of a time series of lateral offset orbital points, with respect to assumed system constraints, i.e., the allowed i.e., limited orbital trajectory. The limit functions 4.25, 4.26, 4.27, and 4.28 are wired together so that TIFF2023545179000036.tif10170 is obtained.

[0075] According to Equation 4.29, a priority order can be freely selected for the chosen limits of the adjustment and state variables, and this order can be realized via a series circuit of saturation elements. This prioritization has the advantage of automatically using the next lowest priority limit when compliance with a limit is physically impossible, thus still providing a solution.

[0076] From one perspective, it is proposed to analytically determine and / or calculate each point in the time series of a lateral offset trajectory using online numerical methods for solving differential equations and / or systems of differential equations.

[0077] From one perspective, it is proposed that a state-variable filter has a predetermined target dynamic, which in particular can be characterized by an extended single-orbit model of a mobile platform. The advantage of pre-setting target dynamics is that the desired dynamic behavior of the system can be parameterized and pre-defined. The extended single-orbit model can be used to consider all relevant states of the system. In this case, the single-orbit model is extended to the state of "lateral offset" in particular, as lateral offset orbits are planned for avoidance maneuvers.

[0078] From one perspective, the dynamic model of the mobile platform is transformed into planar coordinates, and it is proposed that the system of state-variable filters and the system of the dynamic model have the same system order.

[0079] In this case, the restricted state variable filter can be described by equations 4.18 and 4.29.

[0080] From one perspective, it is proposed to analytically determine each point in the time series of a lateral offset trajectory using numerical methods for solving differential equations. Advantageously, numerical methods, specifically online integration of simultaneous differential equations, allow for the determination of lateral offset trajectories and, in particular, the time series of lateral offset trajectory points, in a short time with less computational cost compared to optimization methods.

[0081] From one perspective, it is proposed to restrict at least one system quantity of a flatness-based dynamic model by polytopic state constraints of at least one system quantity of a state-variable filter. Advantageously, the polytopic state restriction imposes a box restriction, i.e., x <x max Not only can this be taken into consideration, but it also allows us to restrict any linear combination of multiple states, and in this way, for example, a position-dependent speed limit can be mapped or taken into consideration.

[0082] From one perspective, the unrestricted requirements dynamics are characterized by the time series of lateral offset trajectory points, and the filter coefficients (a) of the state-variable filter. F,0 ,…a F,4 It is proposed that a linear quadratic controller be pre-configured by full state feedback or design using ).

[0083] In particular, the required dynamics may take into account the section dynamics of a mobile platform, such as a vehicle. That is, the pole / time constants are selected manually and / or using the design method such that the planned trajectory has unlimited required dynamics.

[0084] By clearly considering the limitations of the system quantities, the required dynamics can be adjusted to be significantly more dynamic in methods for generating lateral offset trajectories, because the required dynamics are additionally limited under consideration of the limitations, and therefore feasibility is guaranteed.

[0085] From one perspective, it is proposed that the target lateral offset value be set within a certain time interval.

[0086] From one perspective, it is proposed that at least one limit on one system quantity of a flatness-based dynamic model corresponds to at least one limit on one tuning quantity and / or at least one limit on a state quantity of the flatness-based dynamic model. The advantages of this configured method have already been described.

[0087] From one perspective, it is proposed that the state-variable filter is limited depending on the priority and based on the limitation of one adjustment quantity of the dynamic model and / or the limitation of one state quantity of the dynamic model.

[0088] In other words, the limits on which the adjustment and state variables are selected are freely chosen in accordance with the prioritization, and this order can be realized via a series circuit of saturation elements. This prioritization has the advantage of automatically using the next lowest priority limit when compliance with the prioritized limits is physically impossible, thus still providing a solution.

[0089] From one perspective, it is proposed that at least one limited adjustment amount of a flatness-based dynamic model is the adjustment amount and / or the gradient and / or acceleration of the adjustment amount of at least one actuator affecting the lateral dynamics of the mobile platform. This allows the method to be adapted to various requirements for the control and / or adjustment or operation of the mobile platform.

[0090] From one perspective, it is proposed that at least one actuator controls the steering angle and / or at least one brake pressure and / or at least one wheel damper. By controlling different actuators, it is possible to adapt to the type of mobile platform or to a specific dynamic behavior pattern of the mobile platform.

[0091] From one perspective, it is proposed that at least one constraint on the state variables of the dynamic model is the floating angle and / or yaw angle and / or yaw rate and / or lateral acceleration and / or steering angle and / or lateral offset of the mobile platform. This allows the method to be adapted to various requirements for controlling and / or adjusting the dynamics of mobile platforms.

[0092] A method is proposed for providing a control signal to activate a vehicle that is at least partially automated, based on a time series of values ​​of at least one adjustment variable, and / or for providing a warning signal to alert the occupants of a vehicle, based on a time series of values ​​of at least one adjustment variable. Such control signals and / or warning signals can achieve a higher level of safety in mobile platforms that operate at least partially automatically.

[0093] The concept of "based on" should be understood broadly in relation to the constituent requirement of providing a control signal based on a time series of values ​​of at least one adjustment variable. This concept should be understood as the time series of values ​​of at least one adjustment variable being considered for any determination or calculation of the control signal, and in this case, not precluding the consideration of other input variables for this determination of the control signal. This also applies correspondingly to the provision of warning signals.

[0094] It is proposed that the methods described above be used to avoid accidents in road traffic.

[0095] A control device is proposed to be installed to implement one of the above methods for generating a lateral offset trajectory for at least partially automated mobile platforms.

[0096] From one perspective, a computer program is presented that, when executed by a computer, contains instructions that cause the computer to perform one of the methods described above. Such a computer program makes it possible to use the methods described above in various systems.

[0097] A machine-readable storage medium containing the aforementioned computer program is presented. The computer program is portable on such a machine-readable storage medium.

[0098] A mobile platform may be understood as a mobile system that is at least partially automated, and / or a driver assistance system. One example may be a vehicle that is at least partially automated, or a vehicle equipped with a driver assistance system. In this context, a system that is at least partially automated includes a mobile platform in terms of its at least partially automated functionality, but a mobile platform also includes vehicles and other mobile machines that include driver assistance systems. Other examples of mobile platforms may be a driver assistance system equipped with multiple sensors, or a mobile multi-sensor robot.

[0099] Several embodiments of the present invention are illustrated in relation to Figure 1 and will be described in more detail below. [Brief explanation of the drawing]

[0100] [Figure 1] This is a data flowchart for generating a lateral offset trajectory. [Figure 2]This is a data flowchart for online trajectory planning with a restricted state-variable filter. [Figure 3] This figure shows a cascade consisting of two saturation elements for prioritizing state variable limits and adjustment variable limits within the scope of orbital planning. [Figure 4] This figure shows the time course of the restricted highest-order derivative of the flat output of a dynamic model. [Figure 5a] This figure shows the change in steering angle over time. [Figure 5b] This figure shows the time course of a limited steering angular velocity. [Figure 6a] This is a comparison chart of avoidance trajectories. [Figure 6b] This is a comparison chart of the yaw rate over time. [Figure 7] This figure shows a simulation of one scenario of the avoidance function, which also includes a lateral offset trajectory. [Modes for carrying out the invention]

[0101] Figure 1 shows a flowchart overview of method 100 for generating a lateral offset trajectory 620 for a mobile platform that is at least partially automated. In step S1, the target lateral offset w y (t)110 is provided. In step S2, the target lateral offset w y (t)110 is converted to planar coordinates by filter 130 w z (t). For online trajectory planning 140, the target lateral offset is provided in planar coordinates to the state-variable filter 142 of the online trajectory planning 140. z (t) Adjustment amount 120 in step S4 Limitations of TIFF2023545179000037.tif13170 and state quantity 120 The constraints of TIFF2023545179000038.tif11170 are provided as input quantities. Furthermore, in step S3, the time series of the lateral offset orbital points is provided. z* (t n ) and the fifth time derivative for determining the time series of the lateral offset orbital points TIFF2023545179000039.tif13170 is determined as an input quantity for the flatness-based inverted dynamic model 150 and provided to the inverted dynamic model in step S5. Time series of lateral offset orbit points. z * (t n ) and the fifth time derivative Using TIFF2023545179000040.tif13170, in step S6, the flatness-based inverted dynamic model 150 determines at least one adjustment amount of the mobile platform. Determine the time series of values ​​for TIFF2023545179000041.tif13170. Determine at least one adjustment for the mobile platform. The time series with a value of 160 in TIFF2023545179000042.tif13170 can be used for input control to perform trajectory control of a mobile platform.

[0102] Figure 2 is a schematic diagram showing the information flow of the online trajectory planning 140 in Figure 1 in planar coordinates. In this case, the online trajectory planning 140 has a limiter 144 and an integrator chain 146, corresponding to an extended switchable state variable filter 140 with unlimited filter request dynamics 142.

[0103] At that time, the planar coordinate w is obtained from the target lateral offset. z (t)130, the highest-order time derivative of the flat output of the dynamic model is integrated by the integrator chain 146, limited by the limiter 144, using the preset required dynamics of the state-variable filter 142. Determine the unlimited request signal for TIFF2023545179000043.tif11170, and from here the orbit z * and z *(1) ,…z*(n) The nth time derivative is obtained, and the time series of the lateral offset orbital points is provided as an input to the flatness-based inverted dynamic model 150 of the mobile platform. At that time, this input is fed back to both the limiter 144 and the dynamics of the state-variable filter 142 for the next calculation step. The output signal of the online orbital planning 140 is, for example, an input control amount. For the calculation of TIFF2023545179000044.tif13170, a flatness-based inverted dynamic model 150 is provided. In this method, the system quantities are dynamically limited by limit functions 4.25, 4.26, 4.27, and 4.28, i.e., the filter dynamics are limited over time.

[0104] Figure 3 shows an overview of the data flow prioritization for limiting state variables and / or adjustment variables using a first saturation filter 144b and a second saturation filter 144d connected in series behind it, in which the input variable, i.e., the fifth derivative of z, is first limited using the first saturation filter 144b and / or the second saturation filter 144d. The first saturation filter 144b can limit state variables such as yaw rate, and the second saturation filter 144d, located behind the first saturation filter 144b in the information flow direction, can limit adjustment variables such as steering angular velocity. The yaw rate limiting section 144a limits the yaw rate and the trajectory of the flat output of the dynamic model z * Both (t) and its time derivative or the filtered state of equation 4.18 are provided. The steering speed limiter 144c limits the steering speed and the trajectory of the flat output of the dynamic model z * Both (t) and its time derivative are provided. In other words, in this way, prioritization of the limits can be achieved by a cascading arrangement of both limiters.

[0105] Figure 4 shows graph 400 plotting the fifth derivative of the flat output trajectory z with respect to time t using curve 450. The curves of the constraint imposed by state variable 410 and the adjustment variable 420 illustrate the general constraint on the fifth derivative. It can be seen that by making maximum use of state variable 410, the fifth derivative is set within the upper and lower limits of the adjustment variable 420.

[0106] Figure 5a shows an overview of an example of the progress of the adjustment amount of the steering angle 510 over time, as shown in graph 500a. Figure 5b then shows an overview of the corresponding progression of steering angular velocity 520 with upper and lower limits 525, as shown in graph 500b.

[0107] In Figure 6a, the lateral offset trajectory 620 generated using this method is compared with the lateral offset trajectory 610 generated by another method in Graph 600a. In this case, the latter was generated according to the level of technology, and the state constraints were not fully utilized to determine the trajectory. Error! Reference source not found. In this case, in graph 600a, the lateral offset y of the mobile platform is plotted within the same interval x, and it can be seen that the novel method can achieve an increase in lateral offset of approximately 20%.

[0108] Graph 600b in Figure 6b, which plots the yaw rate against time t for both the method 640 described here and the conventional method 630, clearly shows that by making maximum use of the maximum possible yaw rate within the orbital time range, corresponding to the method described here, improvements in orbital progression can be achieved.

[0109] Figure 7 outlines one traffic scenario, including a simulation of an avoidance function caused by a lateral offset trajectory resulting from a single person on the road. [Explanation of Symbols]

[0110] 110 Target Lateral Offset 120 System Volume 130 filters 140 Online Trajectory Planning 142 Variable-State Filter 144b First saturation filter 144d Second saturation filter 150 Dynamic Model 160 Adjustment amount

Claims

1. 1. A method for generating a lateral offset trajectory for an at least partially automated mobile platform, comprising: providing a target lateral offset (110) (S1); inverting the provided dynamic model of the mobile platform; providing (S4) at least one constraint on a system quantity (120) of the dynamic model to determine the lateral offset trajectory; determining a time series of lateral offset orbit points for the inverted dynamic model (150) using a state variable filter (140, 142) based on the at least one constraint on the system quantity (120) and the target lateral offset (110, 130) as an input signal; wherein analytically determining each point of the time series of the laterally offset trajectory; determining (S6) a time series of values ​​of at least one adjustment quantity (160) for the mobile platform to generate the laterally offset trajectory using the inverted dynamic model (150) and the time series of the laterally offset trajectory points as an input signal for the inverted dynamic model (150); A method comprising:

2. The method according to claim 1 , wherein the state variable filter (140, 142) has predetermined target dynamics, and in particular the predetermined target dynamics is characterized by an extended single-track model of the mobile platform.

3. 3. The method according to claim 1, wherein the dynamic model of the mobile platform is transformed into planar coordinates, in particular the system of state variable filters and the system of dynamic models have the same system order.

4. A method according to any one of claims 1 to 3, wherein each point of the time series of the laterally offset trajectories is determined analytically using a numerical solution of a differential equation.

5. 5. The method according to claim 1, further comprising constraining the at least one system quantity of the flatness-based dynamic model (150) by a polytopic state constraint of the at least one system quantity of the state variable filter (142).

6. 6. The method according to claim 1, wherein the at least one limitation of a system quantity of the dynamic model corresponds to at least one limitation of an adjustment quantity (120, 144b, 144d) of the dynamic model and / or at least one limitation of a state quantity (120, 144b, 144d).

7. 7. The method according to claim 6, wherein the state variable filter (142) is limited in dependence on a priority and based on a limitation of one adjustment quantity (120, 144b, 144d) of the dynamic model and / or based on a limitation of one state quantity (120, 144b, 144d) of the dynamic model.

8. 8. The method according to claim 6 or 7, wherein the at least one limited adjustment quantity (120, 144b, 144d) of the dynamic model is an adjustment quantity of at least one actuator affecting the lateral dynamics of the mobile platform and / or a gradient of said adjustment quantity and / or an acceleration of said adjustment quantity.

9. 9. The method according to claim 8, wherein said at least one actuator controls the steering angle and / or at least one brake pressure and / or at least one wheel damper.

10. 10. The method according to any one of claims 6 to 9, wherein the at least one constraint on the state quantities (120, 144b, 144d) of the dynamic model is a fly angle and / or a yaw angle and / or a yaw rate and / or a lateral acceleration and / or a steering angle and / or a lateral offset of the mobile platform.

11. A method for providing a control signal for activating an at least partially automated vehicle based on a time series of values ​​of at least one adjustment quantity (160) and / or a method for providing a warning signal for issuing a warning to an occupant of the vehicle based on a time series of values ​​of at least one adjustment quantity (160).

12. Use of a method according to any one of claims 1 to 10 for avoiding accidents in road traffic.

13. A control device arranged to carry out a method according to any one of claims 1 to 11.

14. A computer program comprising instructions which, when executed by a computer, cause the computer to carry out the method according to any one of claims 1 to 11.

15. A machine-readable storage medium having the computer program of claim 14 stored thereon.

Images