Vehicle control system

Abstract

To provide a vehicle control device which can predict behavior of a vehicle while reducing a load of calculation.SOLUTION: A vehicle control device 10 comprises: a road surface information acquisition unit 11 which acquires road surface information Ir indicating a state of a road surface R; and a prediction unit 12 which predicts behavior of a vehicle M by using control input. The prediction unit 12 adapts each of a plurality of control input candidates to a simple model representing two-dimensional behavior of the vehicle M, and predicts a travel track of a wheel when each of the simple models is assumed to travel on the road surface R. The prediction unit 12 acquires a road surface profile along each of the travel tracks based on the road surface information Ir acquired by the road surface information acquisition unit 11. The prediction unit 12 evaluates grounding property of the wheel regarding the road surface R that is varied according to the road surface profile and riding quality of the simple model so that a specific control input candidate is extracted and selected among the plurality of control input candidates which are adapted to the simple model.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a vehicle control device.

Background Art

[0002] Conventionally, for example, a vehicle control device disclosed in Patent Document 1 has been known. In the conventional vehicle control device, in model predictive control, in order to suppress the calculation load, the interval of prediction points is set to increase from the proximal side to the distal side of the vehicle. And in the conventional vehicle control device, by setting the weight of the prediction points on the distal side to be smaller than the weight of the prediction points on the proximal side of the vehicle, a predicted travel route that does not give the driver a sense of discomfort is calculated.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] By the way, when the vehicle travels off-road, the road surface input associated with the road surface displacement becomes larger compared to the paved road surface. As a result, in the vehicle, deterioration of the grounding performance and the riding comfort is caused, and an increase in the risk of falling and skidding is expected. For this reason, for example, in the model predictive control also performed by the conventional vehicle control device, by calculating and predicting the behavior of the vehicle and evaluating in advance the grounding performance, the riding comfort, the risk of falling, etc., it is considered that safer driving becomes possible.

[0005] However, the conventional vehicle control device does not take into account the road surface information regarding the shape of the road surface when calculating the predicted travel route in model predictive control. Also, in model predictive control, for example, it is not realistic to calculate and predict the behavior of the vehicle for all control inputs related to the control of the vehicle because the calculation load becomes high.

[0006] The objective of the present invention is to provide a vehicle control device that can predict vehicle behavior while reducing the computational load. [Means for solving the problem]

[0007] The vehicle control device of the present invention comprises a road surface information acquisition unit and a prediction unit. The road surface information acquisition unit acquires road surface information representing the condition of the road surface in front of the vehicle's direction of travel. The prediction unit predicts the behavior of the vehicle using control inputs that can be input in relation to vehicle control. In the vehicle control device of the present invention, the prediction unit acquires road surface information from the road surface information acquisition unit. The prediction unit also applies each of a plurality of control input candidates that can become control inputs to a simplified model that represents the behavior of the vehicle, and predicts the trajectory of the wheels of each simplified model assuming that the simplified model travels on the road surface represented by the road surface information. Then, based on the road surface information, the prediction unit acquires a road surface profile representing the road surface displacement along the trajectory corresponding to each control input candidate, and evaluates the contact of the wheels with the road surface and the ride comfort in the simplified model that change due to the road surface profile, thereby extracting and selecting a specific control input candidate from among the plurality of control input candidates applied to the simplified model. [Effects of the Invention]

[0008] According to the present invention, the prediction unit can narrow down the control input candidates by obtaining a road surface profile using a simplified vehicle model and evaluating the ground contact and ride comfort in advance. The prediction unit can then calculate and predict the vehicle's behavior using the narrowed-down specific control input candidates. Therefore, compared to calculating all vehicle behaviors using all control inputs, for example, the number of calculations can be reduced by narrowing down the number of control inputs, thus reducing the computational load on the prediction unit. [Brief explanation of the drawing]

[0009] [Figure 1] This is a schematic diagram of the vehicle and vehicle control device according to this embodiment. [Figure 2] This is a diagram illustrating the trajectory of a vehicle in a simplified model. [Figure 3] This diagram illustrates the vertical behavior of the sprung mass and unsprung mass in a simplified model. [Figure 4] Figure 1 is a flowchart of the control input determination program executed by the prediction unit. [Figure 5] Figure 1 is a flowchart of the control input candidate selection routine executed by the prediction unit. [Figure 6] This diagram illustrates the frequency characteristics of the sprung mass and unsprung mass displacements, and the frequency characteristics of the sprung mass and spring acceleration. [Modes for carrying out the invention]

[0010] Hereinafter, a vehicle control device 10, which is one embodiment of the present invention, will be described in detail with reference to the drawings. It should be noted that, in addition to the embodiments described below, the present invention can be implemented in various forms with various modifications and improvements based on the knowledge of those skilled in the art.

[0011] As shown in Figure 1, vehicle M has a body 1, wheels 2 positioned at the front, rear, left, and right, and a suspension system 3 that supports the body 1 and each wheel 2, and is a vehicle capable of autonomous driving, for example. Here, in addition to the left and right front wheels (wheels 2), the left and right rear wheels (wheels 2) of vehicle M may be steered, or each of the left, right, front, and rear wheels 2 may be steered independently. Vehicle M is also equipped with a sensor group 4. The sensor group 4 may include, for example, a front wheel steering angle sensor that detects the front wheel steering angle δf when the left and right front wheels are steered by the drive of an actuator of a front wheel steering device (not shown).

[0012] Furthermore, the sensor group 4 may include a rear wheel steering angle sensor that detects the rear wheel steering angle δr when the left and right rear wheels are steered by the actuators of the rear wheel steering device (not shown in the figure). In addition, the sensor group 4 may include wheel sensors that detect the wheel speed of each of the front, rear, left, and right wheels 2. The vehicle speed V can be calculated using the respective wheel speeds.

[0013] As shown in Figure 1, the vehicle control device 10 is mounted on the vehicle M. The vehicle control device 10 mainly consists of a microcomputer with a CPU, ROM, RAM, and various interfaces, and is configured to communicate with other control devices (not shown) mounted on the vehicle M. The vehicle control device 10 includes a road surface information acquisition unit 11 and a prediction unit 12.

[0014] The road surface information acquisition unit 11 acquires road surface information Ir representing the state of the road surface R in front of the vehicle M in the direction of travel. The road surface information acquisition unit 11 is equipped with, for example, a stereo camera and / or LiDAR (Light Detection and Ranging, or Laser Imaging Detection and Ranging), and acquires road surface information Ir including unevenness, undulations, ruts, etc. of the road surface R in front of the vehicle M in the direction of travel, that is, road surface displacement, which is the height H of the road surface R from a predetermined reference position B (see Figure 3 described later).

[0015] Furthermore, the road surface information acquisition unit 11 can, for example, pre-store a road surface information map that includes map information and road surface information Ir associated with the map information, and can acquire road surface information Ir when the vehicle M has traveled X meters or t seconds later, based on the direction of travel, the road surface information map, and the position information of the vehicle M. In addition, the road surface information acquisition unit 11 is equipped with a GNSS (Global Navigation Satellite System) receiver, and can acquire road surface information Ir from an external source (for example, a control system or cloud server) based on the position information of the vehicle M, which is traveling by autonomous driving.

[0016] The prediction unit 12 acquires road surface information Ir from the road surface information acquisition unit 11. Also, as will be described in detail later, the prediction unit 12 selects, as a control input candidate, a set of a plurality of control inputs that can be input in relation to the control of the vehicle M, and predicts the behavior of the vehicle using the selected control input candidate. Here, when the prediction unit 12 selects a control input candidate, as shown in FIGS. 2 and 3, each of the plurality of control input candidates that can be a control input is applied to a simple model 20 (for example, a two-dimensional model) that can represent the two-dimensional behavior of the vehicle M. The prediction unit 12 predicts the traveling trajectories Rt of the front and rear two wheels 22 of the simple model 20 when it is assumed that the simple model 20 travels on the road surface R represented by the road surface information Ir. Then, the prediction unit 12 acquires a road surface profile representing the road surface displacement of each road surface R along the traveling trajectory Rt corresponding to each control input candidate.

[0017] In this way, by acquiring (extracting) the road surface profile using the simple model 20, the prediction unit 12 can reduce the calculation load as compared with, for example, the case of calculating the traveling trajectories of the four wheels 2 of the vehicle M. Incidentally, when the prediction unit 12 predicts the traveling trajectory Rt, in other words, when predicting the traveling trajectory Rt in the horizontal plane, as the simple model 20, an equivalent two-wheel model (linear two-wheel model) shown in FIG. 2 can be used. Also, as will be described later, when the prediction unit 12 predicts the displacement and acceleration above and below the spring, etc., in other words, when predicting the vertical behavior (motion), as the simple model 20, a two-wheel model of a two-degree-of-freedom system shown in FIG. 3 can be used.

[0018] Also, the prediction unit 12 extracts and selects, from among the plurality of control input candidates applied to the simple model 20, a specific control input candidate that has good ground contact of the wheels 22 with respect to the road surface R and good riding comfort in the vehicle body 21, which change due to the road surface profile (road surface displacement) in the simple model 20. Further, the prediction unit 12 ranks the selected specific control input candidates in order of good ground contact of the wheels 22 and good riding comfort in the vehicle body 21 in the simple model 20.

[0019] Then, the prediction unit 12 applies the selected and ranked control input candidates in order from the highest rank to a detailed model (e.g., a three-dimensional model) that can represent the three-dimensional behavior of the vehicle M, and predicts the three-dimensional vehicle motion of the vehicle M. Further, the prediction unit 12 determines, as the final control input, a control input candidate for which the behavior of the vehicle M does not reach a predetermined state based on the predicted three-dimensional vehicle motion.

[0020] Thereby, the prediction unit 12 can predict the three-dimensional vehicle motion for the control input candidates narrowed down in number so that the grounding performance and riding comfort are good. For this reason, the prediction unit 12 is configured to reduce the number of calculations and reduce the calculation load. Further, the prediction unit 12 predicts the three-dimensional vehicle motion using the control input candidates in order from the highest rank. Thereby, the prediction unit 12 can determine, at an early stage, i.e., with a small number of calculations, a control input for which the grounding performance and riding comfort are good and the behavior of the vehicle M does not reach a predetermined state, and reduce the calculation load.

[0021] Next, referring to the flowchart shown in FIG. 4, a control input determination program executed by the prediction unit 12 (more specifically, the microcomputer constituting the prediction unit 12) will be described. The prediction unit 12 starts executing the control input determination program in step S10.

[0022] In the subsequent step S11, the prediction unit 12 acquires road surface information Ir in the forward direction of the traveling direction of the vehicle (the direction indicated by the arrow in FIG. 1) from the road surface information acquisition unit 11. That is, the prediction unit 12 acquires road surface information Ir including the road surface displacement (corresponding to the height H in FIG. 3) of the road surface R in front of the traveling direction of the vehicle M detected by the road surface information acquisition unit 11.

[0023] In the subsequent step S12, the prediction unit 12 executes a "control input candidate selection routine" shown in FIG. 5. As shown in FIGS. 2 and 3, the control input candidate selection routine is a routine for selecting a plurality of control inputs with good grounding performance and good riding comfort using a simple model 20 that represents the two-dimensional behavior of the vehicle M and ranking each control input.

[0024] Here, in the following explanation, "sprung mass" refers to the side of the vehicle body 1 supported by the suspension device 3 of vehicle M, and the side of the vehicle body 21 supported by the suspension device 23 of the simplified model 20. Also, in the following explanation, "unsprung mass" refers to the side of the wheel 2 supported by the suspension device 3 of vehicle M, and the side of the wheel 22 supported by the suspension device 23 of the simplified model 20.

[0025] The prediction unit 12 starts executing the control input candidate selection routine in step S100. Then, in the following step S101, the prediction unit 12 sets the value of variable j, which distinguishes each of the control input candidates consisting of multiple control inputs that can be set in the vehicle M, to "1" (j=1,2,...,N). Here, as shown in Figure 1, the control input candidates include control inputs that can be obtained from the sensor group 4, such as the front wheel steering angle δf, the rear wheel steering angle δr, and the vehicle speed V, and N control input candidates are set.

[0026] In the subsequent step S102, the prediction unit 12 applies each control input constituting the j-th control input candidate to the simplified model 20 and calculates and predicts the time-series travel trajectory Rt (hereinafter also simply referred to as "wheel position travel trajectory Rt") at the positions of the front and rear wheels 22. As a result, the prediction unit 12 can calculate the "wheel position travel trajectory Rt" of the simplified model 20, assuming that the simplified model 20 travels on the road surface R represented by the road surface information Ir, as shown in Figures 2 and 3.

[0027] In the subsequent step S103, the prediction unit 12 calculates a time-series road surface profile along the "travel trajectory Rt of the wheel position". In other words, in step S103, the prediction unit 12 calculates a road surface profile that represents the unevenness of the road surface R, i.e., the road surface displacement, over time, as shown in Figure 3, causing the vehicle body 21 and wheels 22 to move vertically.

[0028] Specifically, as described above, the prediction unit 12 acquires road surface information Ir in step S11. Therefore, the prediction unit 12 extracts and calculates the road surface displacement at the positions that each wheel 22 passes through, corresponding to the "travel trajectory Rt of the wheel position," as a time-series road surface profile.

[0029] In the subsequent step S104, the prediction unit 12 performs a Fourier transform on the time-series road surface profile. That is, the prediction unit 12 performs a Fourier transform on the time-series road surface profile, or in other words, the road surface displacement in the time domain, to convert it into a road surface profile in the frequency domain, i.e., the road surface displacement in the frequency domain. Hereinafter, the converted road surface displacement in the frequency domain will be referred to as the "frequency characteristics of the road surface displacement".

[0030] In the subsequent step S105, the prediction unit 12 calculates the actual sprung mass and unsprung mass frequency characteristics of the simplified model 20. As described above, assuming that the simplified model 20 travels on a road surface R represented by a road surface profile, each wheel 22 passes over irregularities on the road surface R, and the displacement and acceleration in relation to the road surface displacement are obtained as the behavior of the sprung mass (vehicle body 21) and unsprung mass (wheels 22). In other words, in this case, the sprung mass and unsprung mass frequency characteristics, including the "frequency characteristics of road surface displacement," are obtained.

[0031] Here, when evaluating the quality of the wheel 22's contact with the road surface R and the quality of the ride comfort in the vehicle body 21, it is necessary to exclude the influence of the "frequency characteristics of road surface displacement." In other words, when evaluating the quality of contact and ride comfort, it is necessary to use the actual, or in other words, pure, displacement and acceleration frequency characteristics for each of the sprung mass (vehicle body 21) and unsprung mass (wheel 22).

[0032] Therefore, as shown in Figure 6, the prediction unit 12 calculates, for example, the frequency characteristics of the sprung mass and unsprung mass displacements with respect to the experimentally calculable road surface displacement (hereinafter simply referred to as "displacement / road surface displacement frequency characteristics"). The prediction unit 12 also calculates, for example, the frequency characteristics of the vertical acceleration of the sprung mass and unsprung mass with respect to the experimentally calculable road surface displacement (hereinafter simply referred to as "acceleration / road surface displacement frequency characteristics"). Note that acceleration can also be calculated by taking the second derivative of the displacement.

[0033] Then, as shown in Figure 6, the prediction unit 12 multiplies the calculated "frequency characteristics of displacement / road surface change" and the "frequency characteristics of road surface displacement". This allows the prediction unit 12 to calculate the frequency characteristics of the pure sprung and unsprung displacements that actually occur in the simplified model 20, with the influence of road surface displacement excluded (hereinafter simply referred to as "actual displacement frequency characteristics").

[0034] Similarly, the prediction unit 12 multiplies the calculated "frequency characteristics of acceleration / road surface displacement" by the "frequency characteristics of road surface displacement". This allows the prediction unit 12 to calculate the frequency characteristics of the pure sprung mass and unsprung mass that actually occur in the simplified model 20, with the influence of road surface displacement removed (hereinafter simply referred to as "actual acceleration frequency characteristics").

[0035] Next, in step S106, the prediction unit 12 calculates the partial overall near the resonance frequency for the "actual displacement frequency characteristics" and the "actual acceleration frequency characteristics" in order to evaluate ground contact and ride comfort. Specifically, as shown enclosed by a thick solid rectangle in the "actual displacement frequency characteristics" of Figure 6, the prediction unit 12 calculates the partial overall, which is the sum of power within a predetermined frequency range specified near the resonance frequencies of the sprung mass and unsprung mass.

[0036] Similarly, as indicated by the thick solid line square in the "Actual Acceleration Frequency Characteristics" of FIG. 6, the prediction unit 12 calculates the partial overall of the sum of the powers in a predetermined frequency range specified near the resonance frequencies above and below the spring. Then, the prediction unit 12 associates the calculated values of the partial overall of the displacements above and below the spring and the values of the partial overall of the accelerations above and below the spring with the variable j, and executes the process of step S107.

[0037] In step S107, the prediction unit 12 determines whether the value of the variable j has reached the number N of control input candidates. That is, if the value of the variable j is equal to the value N, in other words, if the partial overall has been calculated for all N control input candidates, the prediction unit 12 determines "Yes" in step S107 and executes the process of step S108.

[0038] In step S108, the prediction unit 12 selects K (K < N) from the top in ascending order of the values for the "values of the partial overall of displacements" and the "values of the partial overall of accelerations" calculated N each for above and below the spring. Hereinafter, the processing for each of the selected K "values of the partial overall of displacements" and "values of the partial overall of accelerations" will be described. Note that the "values of the partial overall of displacements" and the "values of the partial overall of accelerations" can be evaluated such that the smaller the value, the better the grounding and riding comfort.

[0039] The prediction unit 12 selects K each from the N "values of the partial overall of displacements above the spring" and the N "values of the partial overall of displacements below the spring" in ascending order. Then, the prediction unit 12 selects the control inputs corresponding to each of the ranked K "values of the partial overall of displacements above the spring" and the control inputs corresponding to each of the ranked K "values of the partial overall of displacements below the spring" as specific "control input candidates".

[0040] Furthermore, the prediction unit 12 ranks the K selected "control input candidates" in ascending order of their corresponding "partial overall value of sprung displacement" and "partial overall value of unsprung displacement". In this way, the prediction unit 12 can rank the selected "control input candidates" with respect to sprung and unsprung displacement.

[0041] Furthermore, the prediction unit 12 selects K values ​​in ascending order from N "partial overall values ​​of acceleration on the sprung mass" and N "partial overall values ​​of acceleration on the unsprung mass". The prediction unit 12 then selects the control inputs corresponding to each of the K ranked "partial overall values ​​of acceleration on the sprung mass" and each of the K ranked "partial overall values ​​of acceleration on the unsprung mass" as specific "control input candidates".

[0042] Furthermore, the prediction unit 12 ranks the K selected "control input candidates" in ascending order of their corresponding "partial overall value of sprung mass acceleration" and "partial overall value of unsprung mass acceleration". In this way, the prediction unit 12 can rank the selected "control input candidates" with respect to sprung mass and unsprung mass acceleration.

[0043] Here, generally speaking, for a vehicle M (including the simplified model 20), the smaller the displacement of the unsprung mass, specifically the wheels 2 (wheels 22), near the resonant frequency, the better the ground contact can be evaluated. Also, generally speaking, for a vehicle M (including the simplified model 20), the smaller the acceleration of the sprung mass, specifically the vehicle body 1 (vehicle body 21), near the resonant frequency, the better the ride comfort can be evaluated. Therefore, in step S106, the prediction unit 12 can omit the selection and ranking of sprung mass displacements and the selection and ranking of unsprung mass accelerations.

[0044] On the other hand, in step S107, the prediction unit 12 determines "No" if the value of variable j is not the value N, and executes the process in step S109. In step S109, the prediction unit 12 increments variable j by "1". Then, the prediction unit 12 uses the "j+1"th control input candidate to execute each step process from step S102 onward.

[0045] In step S108, the prediction unit 12 selects and ranks K specific "control input candidates" from the N "control input candidates". Then, in step S110, it terminates the execution of the "control input candidate selection routine". The prediction unit 12 then returns to step S12 of the "control input determination program" and executes the processing of each step from step S13 onwards.

[0046] In step S13 of the "control input determination program," the prediction unit 12 sets the value of variable i, which distinguishes the K selected specific "control input candidates," to "1" (i=1,2,...,K). Then, the prediction unit 12 executes the process in step S14.

[0047] In step S14, the prediction unit 12 predicts the three-dimensional vehicle motion of vehicle M using a detailed model that represents the behavior of vehicle M. In predicting the three-dimensional vehicle motion, the prediction unit 12 obtains the i-th "control input candidate" from the K "control input candidates" in order of priority.

[0048] The prediction unit 12 then uses the acquired i-th "control input candidate," such as the front wheel steering angle δf, the rear wheel steering angle δr, and the vehicle speed V, as state variables to predict the three-dimensional vehicle motion, including roll, pitch, heave (bouncing), and yaw behavior, that occurs in the vehicle M. Note that the prediction of three-dimensional vehicle motion can be performed using well-known calculation methods, such as prediction (simulation) using three-dimensional equations of motion or characteristic equations, so a detailed explanation is omitted.

[0049] Next, in step S15, the prediction unit 12 evaluates a predetermined state of vehicle behavior, including vehicle motion behavior and vehicle driving behavior, based on the predicted three-dimensional vehicle motion. Here, an example of a predetermined state of vehicle behavior is a state in which vehicle M overturns when driving off-road.

[0050] In this case, the prediction unit 12 evaluates the magnitude of the tipping risk, which represents the likelihood of tipping over. The magnitude of the tipping risk is evaluated, for example, based on the positional relationship between the zero moment point (ZMP), which is represented as the intersection of the direction in which the resultant force of gravity and inertial force at the vehicle's center of gravity points with the ground, and the sides forming an arbitrarily set support polygon (such as the distance between the ZMP and the sides).

[0051] Furthermore, an example of a predetermined state of vehicle behavior is the state in which vehicle M skids while driving off-road. In this case, the prediction unit 12 evaluates the magnitude of the skid risk, which represents the possibility of skidding occurring. The magnitude of the skid risk is evaluated, for example, based on the relationship between the magnitude of the force in the tire shear direction predicted during cornering of vehicle M and the size of the friction circle.

[0052] Furthermore, a predetermined state of vehicle behavior is when vehicle M travels along a target route by autonomous driving or the like. Path deviation The state can be illustrated as an example. In this case, the prediction unit 12 evaluates the magnitude of the path deviation risk, which represents the difference between the target driving path and the predicted driving path of vehicle M predicted based on, for example, the control input. The path deviation risk is evaluated based on the difference in the distance between the target driving path and the predicted driving path, or the difference in path length.

[0053] In this embodiment, the prediction unit 12 is shown as an example of evaluating each of the following as predetermined states of vehicle behavior: rollover risk, skidding risk, and path deviation risk. However, it goes without saying that the items that the prediction unit 12 evaluates as predetermined states of vehicle behavior can be increased, decreased, or changed as needed.

[0054] For example, the prediction unit 12 can evaluate only the risk of tipping over, only the risk of skidding, or only the risk of deviating from the intended path as predetermined states of the vehicle's behavior. Alternatively, the prediction unit 12 can evaluate both the risk of tipping over and the risk of skidding, or both the risk of tipping over and the risk of deviating from the intended path, or both the risk of skidding and the risk of deviating from the intended path as predetermined states of the vehicle's behavior.

[0055] Furthermore, the prediction unit 12 can, of course, also evaluate risks and evaluation items other than the example of rollover risk as predetermined states of vehicle behavior. In this case, the prediction unit 12 can, for example, use the risk of vehicle M heave (bouncing) behavior or vehicle M yaw behavior when driving in ruts as an example of an evaluation item.

[0056] In the subsequent step S16, the prediction unit 12 determines, based on the risk assessment results, whether, for example, vehicle M will tip over and skid when driving off-road. Alternatively, the prediction unit 12 determines, for example, whether vehicle M will deviate from its path when driving on-road. In other words, the prediction unit 12 determines whether the predicted behavior of vehicle M will reach a predetermined state.

[0057] Specifically, when the prediction unit 12 determines whether or not a rollover will occur, it determines "No" and executes the process in step S17 because the vehicle M will not roll over if the ZMP remains located inside the support polygon. Alternatively, when the prediction unit 12 determines whether or not a skid will occur, it determines "No" and executes the process in step S17 because the vehicle M will not skid if the magnitude of the force in the tire shear direction does not exceed the friction circle.

[0058] Furthermore, when the prediction unit 12 determines whether or not a path deviation occurs in vehicle M, if the difference in the distance between the target driving path and the predicted driving path, or the difference in path length, is less than a predetermined distance, it determines that vehicle M has not deviated from its path and executes the process in step S17. In other words, in step S16, the prediction unit 12 determines "No" and executes the process in step S17 only if the behavior of vehicle M does not reach a predetermined state.

[0059] If the result in step S16 is "No," in other words, if the vehicle M does not reach a predetermined state, the prediction unit 12, in step S17, ultimately determines the i-th specific "control input candidate" that does not cause a risk of tipping over as the "control input." Here, as described above, the prediction unit 12 predicts the three-dimensional vehicle motion using the "control input candidates" in order of their ranking from the K specific "control input candidates," that is, in order of best ground contact and ride comfort. Therefore, the control input determined in step S18 following the "No" determination in step S16 is a control input that provides good ground contact and ride comfort for the vehicle M, and in which the vehicle M does not reach a predetermined state (does not cause risk).

[0060] On the other hand, in step S16, when the prediction unit 12 determines whether or not the vehicle M will overturn, if ZMP is located outside the support polygon, the vehicle M will overturn, so it determines "Yes" and executes the process in step S18. Alternatively, when the prediction unit 12 determines whether or not the vehicle M will skid, if the magnitude of the force in the tire shear direction exceeds the friction circle, skid will occur, so it determines "Yes" and executes the process in step S18.

[0061] Alternatively, when the prediction unit 12 determines whether or not a path deviation occurs in vehicle M, if the difference in the distance between the target driving path and the predicted driving path or the difference in path length is greater than or equal to a predetermined distance, it determines that vehicle M is deviating from its path and executes the process in step S18. In other words, in step S16, if the behavior of vehicle M reaches a predetermined state, the prediction unit 12 determines "No" and executes the process in step S18.

[0062] In step S18, the prediction unit 12 determines whether the value of variable i is one of the selected "control input candidates" (number K). If the value of variable i is K, the prediction unit 12 determines "Yes" and executes the process in step S19.

[0063] In step S19, the prediction unit 12 predicts the three-dimensional vehicle motion of vehicle M using K specific "control input candidates" and determines that there is a possibility of rollover risk, skidding risk, and path deviation risk occurring. Therefore, it selects the control input that minimizes the evaluation values ​​for rollover risk, skidding risk, and path deviation risk. In other words, the prediction unit 12 selects the "control input candidate" with the lowest probability of rollover risk, etc. occurring, that is, the "control input candidate" in which the vehicle M's behavior is least likely to reach a predetermined state, and then executes step S17.

[0064] If the determination in step S16 is "Yes", the prediction unit 12, in step S17, finally determines the "control input candidate" that makes the vehicle M least likely to reach a predetermined state as the "control input". In this case, as described above, the prediction unit 12 predicts the three-dimensional vehicle motion using all of the K "control input candidates" as control inputs.

[0065] Therefore, the control input adopted in step S19 and determined in step S17 may have a lower ranking among the "candidate control inputs." Consequently, the determined control input may result in a slight deterioration in ground contact and ride comfort, but it is the control input that makes it least likely for the vehicle M to reach the predetermined state.

[0066] On the other hand, in step S18, the prediction unit 12 determines "No" if the variable i is not among the number K of "control input candidates" and executes the process in step S20. In step S20, the prediction unit 12 increments the variable i by "1". Then, the prediction unit 12 uses the "i+1"th "control input candidate," in other words, a "control input candidate" with a lower rank than the "control input candidate" used in the previous program execution, and executes the processes in each step from step S14 onward.

[0067] Then, in step S17, when the prediction unit 12 finally determines the control input, it temporarily terminates the execution of the "control input determination program" in step S21. Then, at predetermined intervals, or when execution is necessary, the prediction unit 12 restarts the execution of the control input determination program in step S10.

[0068] Here, the final determined control input is output from the vehicle control device 10 to each device, such as the front wheel steering device, the rear wheel steering device, or the engine control device or motor control device. Accordingly, each device controls the operation of the corresponding equipment according to the acquired control input. As a result, the vehicle M can drive in accordance with the three-dimensional vehicle motion predicted by the prediction unit 12, that is, in a manner that suppresses the occurrence of risks and prevents the behavior from reaching a predetermined state.

[0069] As can be understood from the above explanation, the vehicle control device 10 is composed of a road surface information acquisition unit 11 and a prediction unit 12. The road surface information acquisition unit 11 acquires road surface information Ir, which represents the state of the road surface R in front of the vehicle M in the direction of travel. The prediction unit 12 predicts the behavior of the vehicle M using control inputs that can be input in relation to the control of the vehicle M. The vehicle control device 10 then has the prediction unit 12 apply each of the multiple control input candidates that can become control inputs to a simplified model 20 capable of representing the behavior of the vehicle M, predict the travel trajectory of the wheels 22 of the simplified model 20 assuming that each simplified model 20 travels on the road surface R represented by the road surface information Ir, acquire the road surface profile of each road surface R along the travel trajectory corresponding to each control input candidate, and extract and select from the multiple control input candidates a control input candidate that has good contact between the wheels 22 and the road surface R and good ride comfort in the simplified model 20, which changes due to the road surface profile.

[0070] According to this, the prediction unit 12 can narrow down the control input candidates by obtaining a road surface profile using a simplified model 20 of the vehicle M and evaluating the ground contact and ride comfort in advance. Then, the prediction unit 12 can apply the narrowed-down specific control input candidates to the detailed model and calculate and predict the behavior of the vehicle M (three-dimensional vehicle motion). Therefore, for example, compared to calculating the behavior (three-dimensional vehicle motion) of all vehicles M using all control inputs, the computational load can be reduced by narrowing down the number of control inputs. [Explanation of Symbols]

[0071] M...Vehicle, 1...Vehicle body, 2...Wheels, 3...Suspension system, 4...Sensor group, 10...Vehicle control device, 11...Road surface information acquisition unit, 12...Prediction unit, 20...Simplified model, 21...Vehicle body, 22...Wheels, 23...Suspension system, R...Road surface, Rt...Tracking trajectory, Ir...Road surface information

Claims

1. A road surface information acquisition unit acquires road surface information representing the condition of the road surface in front of the vehicle's direction of travel, It includes a prediction unit that predicts the behavior of the vehicle using control inputs that can be input in relation to the control of the vehicle, The prediction unit, The road surface information is acquired from the road surface information acquisition unit. Each of the multiple control input candidates that can be the control inputs is applied to a two-dimensional model capable of representing the two-dimensional behavior of the vehicle, and the trajectory of the wheels of each of the two-dimensional models is predicted assuming that each of the two-dimensional models travels on the road surface. Based on the road surface information, a road surface profile is obtained that represents the road surface displacement of the road surface along the travel trajectory corresponding to each of the control input candidates. By evaluating the wheel's contact with the road surface and the ride comfort in the two-dimensional model, which change due to the road surface profile, a specific number of control input candidates are extracted and selected from among the multiple control input candidates applied to the two-dimensional model in order of their suitability for contact with the road surface and ride comfort. The selected control input candidates are applied to a three-dimensional model capable of representing the three-dimensional behavior of the vehicle to predict the three-dimensional vehicle motion, which is the behavior of the vehicle. Vehicle control system.

2. The prediction unit ranks the selected control input candidates in order of their ground contact and ride comfort, The three-dimensional vehicle motion is predicted by applying the ranked control input candidates to the three-dimensional model in order from the highest-ranked to the lowest-ranked. The vehicle control device according to claim 1.

3. Based on the predicted three-dimensional vehicle motion, the prediction unit ultimately determines as the control input candidate a control input such that the predicted behavior of the vehicle does not result in the vehicle overturning when driving off-road, does not result in the vehicle skidding when driving off-road, and when the vehicle is driving the target driving path by autonomous driving, the distance between the target driving path and the predicted driving path of the vehicle predicted based on the control input, or the path length, does not exceed a predetermined distance. The vehicle control device according to claim 1 or 2.

Images