JP2010049719A - Device for estimating driving intention, driving operation assist device for vehicle, and vehicle equipped with driving operation assist device for vehicle - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a driving operation assist device for a vehicle for estimating a driving intention of a driver with high precision and for controlling a reaction force according to the intention of the driver. <P>SOLUTION: The driving operation assist device for a vehicle calculates a risk potential in the periphery of one's own vehicle, and calculates an instruction value of an operation reaction force to be generated to an accelerator pedal on the basis of the risk potential. In addition, the instruction value of the reaction force is corrected on the basis of a driving intention state estimated by a device for estimating a driving intention as the driving intention of an actual driver. When the driving intention of the actual driver is estimated to be a lane change, the instruction value of the reaction force to the accelerator pedal is gradually decreased on the basis of elapsed time from when the driver intended the lane change. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a driving intention estimation device that estimates a driving behavior intention of a driving driver and a vehicle driving operation assistance device that assists a driver in accordance with the driving intention.

  A conventional driving intention estimation apparatus estimates driving intention using a driver's gaze behavior (see, for example, Patent Document 1). This apparatus projects the driver's line-of-sight direction on the front projection plane, and estimates the driver's driving intention using the line-of-sight direction frequency distribution in a plurality of divided areas on the projection plane.

JP 2002-331850 A

  The above-described conventional device can estimate the driver's intention of driving behavior using the driver's line-of-sight direction, the gaze gaze frequency, and the like. However, the driver's gaze behavior is affected by differences in the driving environment of the vehicle, and there is also a problem that the driver's individual differences are large and the accuracy of intention estimation varies. Is desired.

  A driving intention estimation device according to the present invention includes a vehicle surrounding state detection unit that detects a vehicle surrounding state of a host vehicle, a driving operation amount detection unit that detects a driving operation amount by an actual driver, and a plurality of driving intentions. Virtual driver driving operation amount calculation that calculates the amount of driving operation necessary for each virtual driver to fulfill the driving intention based on the vehicle surrounding state of the host vehicle detected by the vehicle surrounding state detection means for different virtual drivers. Driving intention sequence generation for dynamically determining driving intention sequences of a plurality of virtual drivers in a predetermined time interval by determining means and driving intentions of a plurality of virtual drivers in each time frame of a past predetermined time interval including the present time And a series of approximate degrees of driving operation amounts of a plurality of virtual drivers and actual driving operation amounts in a predetermined time interval ( Is calculated for each driving intention sequence generated by the driving intention sequence generation means, and the driving operation amount series approximation degree calculation means. The driving intention estimation means for estimating the actual driving intention of the driver by comparing a plurality of driving operation amount series approximation degrees, and the driving intention state of the actual driver's driving intention estimated by the driving intention estimation means Driving intention state determination means for estimating

  A vehicle driving operation assistance device according to the present invention includes a driving intention estimation device, an obstacle detection means for detecting an obstacle situation around the own vehicle, and a risk potential around the own vehicle based on a detection result by the obstacle detection means. A risk potential calculating means for calculating the operating force, an operating reaction force calculating means for calculating an operating reaction force of the accelerator pedal based on the risk potential calculated by the risk potential calculating means, and an operating reaction force that generates an operating reaction force on the accelerator pedal. Force generation means, and correction means for correcting an operation reaction force generated by the accelerator pedal based on the driving intention and the state of the driving intention estimated by the driving intention estimation device.

  According to the driving intention estimation device of the present invention, the driving intention of the virtual driver and the actual driving operation of the driver are compared in series to estimate the driving intention and the state of the driving intention. Can be estimated.

  According to the vehicle driving operation assisting device of the present invention, the operation reaction force generated by the accelerator pedal is corrected based on the estimated driving intention and the state of the driving intention, so that the operation reaction force that meets the actual driver's expectation. Control can be performed.

1 is a system diagram of a driving intention estimation device according to a first embodiment of the present invention. The flowchart which shows the process sequence of the driving intention estimation process in 1st Embodiment. The figure explaining the calculation method of the driving operation amount of a virtual driver. The figure which shows the example of the dynamic driving | operation intention series of a virtual driver. (A) (b) The figure explaining the setting method of the dynamic driving intention series of a virtual driver. The figure explaining the calculation method of the driving intention elapsed time in a dynamic driving intention series. The system diagram of the driving assistance device for vehicles by a 2nd embodiment. The block diagram of the vehicle carrying the driving operation assistance apparatus for vehicles shown in FIG. The figure which shows the structure of an accelerator pedal and its periphery. The flowchart which shows the process sequence of the driving operation assistance control process in 2nd Embodiment. The figure which shows the relationship between risk potential and reaction force increase amount. The figure which shows the relationship between estimated driving intention elapsed time and a time constant. (A) The figure which shows a mode that the own vehicle changes lane, (b) The figure which shows the time change of the accelerator pedal reaction force at the time of lane change. The flowchart which shows the process sequence of the driving operation assistance control process in the modification of 2nd Embodiment. The figure which shows the relationship between estimated driving intention elapsed time and a time constant. The system diagram of the driving assistance device for vehicles by a 3rd embodiment. The flowchart which shows the process sequence of the driving operation assistance control process in 3rd Embodiment. The figure which shows the relationship between estimated driving intention elapsed time and a time constant. (A) The figure which shows a mode that the own vehicle changes lane, (b) The figure which shows the time change of the accelerator pedal reaction force at the time of lane change. The flowchart which shows the process sequence of the driving operation assistance control process in the modification of 3rd Embodiment. The figure which shows the relationship between estimated driving intention elapsed time and a time constant.

<< First Embodiment >>
A driving intention estimation apparatus according to a first embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a system diagram showing a configuration of a driving intention estimation device 1 according to the first embodiment of the present invention. First, the configuration of the driving intention estimation device 1 according to the first embodiment will be described.

  The driving intention estimation device 1 includes a vehicle surrounding state detection unit 10 that detects a state around the host vehicle, a vehicle state detection unit 20 that detects a vehicle state, and a driving operation amount detection unit that detects a driving operation amount by a driver's operation. 30, a virtual driver driving intention sequence dynamic determination unit 40, a virtual driver operation amount calculation unit 50, a virtual driver driving operation amount approximation degree calculation unit 60, a driving intention estimation unit 70, and a driving intention state determination unit 80 It has.

  The driving intention estimation device 1 sets a plurality of virtual drivers having driving intentions, and compares the actual driving operation of the driver with the driving operation of the virtual driver. Then, based on how close the actual driving operation of the driver and the driving operation of the virtual driver are, the actual driving intention of the driver such as lane keeping or lane change is estimated. In order to estimate the driving intention, the driving intention sequence of the virtual driver in the predetermined time from the present to the past is dynamically determined. Then, a series of approximate degrees of driving operation is calculated using the dynamically determined driving intention sequence, and the actual driving intention of the driver is estimated. Further, the estimated actual driving intention state of the driver is determined.

  The vehicle surrounding state detection unit 10 includes, for example, a front camera and a yaw angle sensor that acquire the front road condition of the host vehicle as an image, and detects the lateral position of the host vehicle in the lane, the yaw angle of the host vehicle with respect to the lane, and the like. To do. The vehicle surrounding state detection unit 10 also includes an image processing device that performs image processing on an image signal acquired by the front camera.

  The vehicle state detection unit 20 includes a vehicle speed sensor that detects the vehicle speed, for example. The driving operation amount detector 30 includes, for example, a steering angle sensor incorporated in the steering system, and detects the steering angle of the host vehicle.

  The virtual driver driving intention sequence dynamic determination unit 40, the virtual driver operation amount calculation unit 50, the virtual driver driving operation amount approximation degree calculation unit 60, the driving intention estimation unit 70, and the driving intention state determination unit 80 are each configured by a microcomputer, for example. Is done. Alternatively, in a controller composed of a CPU and CPU peripheral components such as a ROM and a RAM, each can be configured by a software form of the CPU.

  The virtual driver driving intention sequence dynamic determination unit 40 dynamically determines driving intention sequences of a plurality of virtual drivers in a predetermined time in the past from the present. Here, the dynamic determination of the driving intention series of the virtual driver means that the driving intention of the virtual driver, and further, the number of virtual drivers, that is, the driving intention series changes with time. Therefore, for example, even if the current driving intention is a right lane change, there is a possibility that a virtual driver whose past driving intention is lane keeping is set.

  The virtual driver driving operation amount calculation unit 50 calculates a driving operation amount necessary for a plurality of virtual drivers given different driving intentions to fulfill their driving intentions. The virtual driver driving operation amount approximation degree calculation unit 60 calculates the driving operation amount of the virtual driver calculated by the virtual driver driving operation amount calculation unit 50 and the actual driving operation amount of the driver detected by the driving operation amount detection unit 30. The degree of approximation is calculated for each driving intention sequence determined by the virtual driver driving intention sequence dynamic determination unit 40.

  The driving intention estimation unit 70 estimates the driving intention of the actual driver by comparing the degree of approximation for each driving intention series regarding the driving operation amount calculated for the plurality of virtual drivers and the actual driver. The driving intention state determination unit 80 determines the estimated actual driving intention state of the driver.

  Below, operation | movement of the driving intention estimation apparatus 1 by 1st Embodiment is demonstrated in detail using FIGS. FIG. 2 is a flowchart showing a processing procedure of a driving intention estimation processing program in the driving intention estimation device 1. FIG. 3 is a diagram illustrating a method for calculating a driving operation amount of a virtual driver, FIGS. 4 and 5A and 5B are diagrams illustrating a method for setting a dynamic driving intention sequence of a virtual driver, and FIG. It is a figure explaining the state determination method of a driving intention. The processing content of the processing shown in FIG. 2 is continuously performed at regular intervals (for example, 50 msec).

  In step S101, the current lateral position y of the host vehicle detected by the vehicle surrounding state detection unit 10 and the yaw angle ψ of the host vehicle are read. As shown in FIG. 3, the lateral position y in the lane is a lateral distance from the lane center line of the own lane to the own vehicle center point O, and the yaw angle ψ is the rotation angle of the own vehicle with respect to the straight direction of the own lane. It is.

  In step S102, driving operation amounts Oid of a plurality of virtual drivers are calculated. Here, three virtual drivers with driving intentions of lane keeping (LK), right lane change (LCR), and left lane change (LCL) are set. Then, a driving operation amount Oid necessary for each virtual driver to fulfill its driving intention is calculated. Here, the steering angle θid of the steering operation performed by the virtual driver is calculated as the driving operation amount Oid. A method for calculating the driving operation amount Oid of the virtual driver will be described below.

(1) When the driving intention of the virtual driver is lane keeping In order to calculate the steering angle θid of the virtual driver, first, the front reference point LK (i) when the driving intention of the virtual driver is lane keeping is set, and the front The lateral position p_lk of the reference point LK (i) is calculated. The number of the forward reference points LK (i) is arbitrary, but here, a case where two forward reference points LK1 and LK2 are set on the front-rear direction center line of the host vehicle will be described as an example. As shown in FIG. 3, the distance px (i) from the vehicle center point O to the forward reference points LK1, LK2 is set to px (1) = 10 m and px (2) = 30 m, for example (px = {10 m , 30m}). The distance px (i) can be set according to the vehicle speed, for example.

The lateral distance lat_pos (px (i)) from the center line of the lane in which the host vehicle currently travels to the forward reference point LK (i) is the distance px (y) between the yaw angle ψ of the host vehicle and the forward point LK (i). Depending on i), for example, it can be calculated based on the image signal from the front camera. The lateral position p_lk (px (i)) of the forward reference point LK (i) in the case of lane keeping can be expressed by the following (Equation 1).
p_lk (px (i)) = lat_pos (px (i)) i = {1, ..., n} (Formula 1)
Here, n = 2.

Using the lateral position p_lk (px (i)) of the forward reference point LK (i), the steering angle θid_lk of the virtual driver in the case of lane keeping can be calculated from the following (Equation 2).
θid_lk = Σ {a (i) × p_lk (px (i))} (Formula 2)
Here, a (i) is a weighting coefficient for weighting the lateral position p_lk (px (i)) at the forward reference point LK (i), and an appropriate value is set in advance.

(2) When the driving intention of the virtual driver is the right lane change In order to calculate the steering angle θid of the virtual driver, the forward reference point LCR (i) when the driving intention of the virtual driver is the right lane change is set. FIG. 3 shows an example in which two front reference points LCR1 and LCR2 are set in front of the host vehicle.

The lateral position p_lcr (px (i)) of the forward reference point LCR (i) in the case of changing the right lane is the left and right of the forward reference point LK (i) in the case of maintaining the lane, as expressed by the following (Equation 3). It can be calculated by adding the offset amount lc_offset_lcr to the direction distance lat_pos (px (i)).
p_lcr (px (i)) = lat_pos (px (i)) + lc_offset_lcr i = {1, ..., n} (Formula 3)
Here, n = 2. The offset amount lc_offset_lcr is set in advance to an appropriate value, for example, lc_offset_lcr = -1.75, in order to set the lateral position p_lcr (px (i)) of the forward reference point LCR (i) in the case of changing the right lane.

Using the lateral position p_lcr (px (i)) in the lane of the forward reference point LCR (i), the steering angle θid_lcr in the case of changing the right lane can be calculated from the following (Equation 4).
θid_lcr = Σ {a (i) × p_lcr (px (i))} (Formula 4)
Here, a (i) is a weighting coefficient for weighting the lateral position p_lcr (px (i)) in the lane at the forward reference point LCR (i), and an appropriate value is set in advance.

(3) When the driving intention of the virtual driver is the left lane change In order to calculate the steering angle θid of the virtual driver, the forward reference point LCL (i) when the driving intention of the virtual driver is the left lane change is set. FIG. 3 shows an example in which two front reference points LCL1 and LCL2 are set in front of the host vehicle.

The lateral position p_lcl (px (i)) of the forward reference point LCL (i) in the case of the left lane change is expressed by the left and right of the forward reference point LK (i) in the case of lane keeping, as expressed by the following (Equation 5). It can be calculated by adding the offset amount lc_offset_lcl to the direction distance lat_pos (px (i)).
p_lcl (px (i)) = lat_pos (px (i)) + lc_offset_lcl i = {1, ..., n} (Formula 5)
Here, n = 2. The offset amount lc_offset_lcl is set in advance to an appropriate value, for example, lc_offset_lcl = 1.75, in order to set the lateral position p_lcl (px (i)) of the forward reference point LCL (i) when changing the left lane.

Using the lateral position p_lcl (px (i)) in the lane of the forward reference point LCL (i), the steering angle θid_lcl of the virtual driver in the case of changing the left lane can be calculated from the following (Equation 6).
θid_lcl = Σ {a (i) × p_lcl (px (i))} (Expression 6)
Here, a (i) is a weighting coefficient for weighting the lateral position p_lcl (px (i)) in the lane at the forward reference point LCL (i), and an appropriate value is set in advance.

  Thus, after calculating the driving operation amount Oid of the virtual driver in step S102, the process proceeds to step S103. In step S103, the current steering angle θrd detected by the driving operation amount detector 30 is read as the actual driving operation amount Ord of the driver.

  In step S104, a dynamic driving intention sequence of the virtual driver is generated. Here, as shown in FIG. 4, a driving intention sequence of a virtual driver is generated by combining possible driving intentions in m time frames from a current time t to a past time point (t−m + 1). For example, the driving intention series L1 has the driving intention of the lane keeping LK from the past time (t−m + 1) to the time (t−1), but the driving intention changes to the right lane change LCR at the current time t. Represents a changed virtual driver. The driving intention series L2 represents a virtual driver whose driving intention has changed from the lane keeping LK to the right lane change LCR at a time point (t-1) one frame before, and the driving intention series L3 is a past time point (tm + 3). Represents a virtual driver whose driving intention has changed from the lane keeping LK to the right lane changing LCR.

  Below, the setting method of the dynamic driving intention series of a virtual driver is demonstrated. As the driving intention sequence from the past time point (t−m + 1) to the time (t−1), the driving intention sequence set at the time (t−1) is continuously used. The driving intention of the virtual driver at the current time t is dynamically determined based on the driving intention of the virtual driver and the generation rule of the driving intention sequence at the time (t−1) one frame before.

  As shown in FIG. 5A, when the driving intention of the virtual driver at time (t-1) is lane keeping LK, the possible driving intention of the virtual driver at time t is lane keeping LK, right lane change. LCR and left lane change LCL. Therefore, the driving intention sequence that was one until time (t−1) is increased to three at time t. At this time, the three driving intention sequences respectively inherit the driving intention sequences from the branch source time (t−m + 1) to the time (t−1).

  On the other hand, when the driving intention of the virtual driver at time (t-1) is a lane change, the virtual driver's driving intention at time t is only a lane change in the same direction. As shown in FIG. 5B, when the driving intention of the virtual driver at time (t-1) is the right lane change LCR, the only possible virtual driver driving intention at time t is the right lane change LCR. . Specifically, when it is determined that the host vehicle continues to change the right lane at time t, the driving intention is the right lane changing LCR, and the driving intention sequence is continued. On the other hand, for example, when the lane change to the right adjacent lane is completed and the lane change is not executed at time t, the driving intention sequence itself is extinguished.

  As a lane change continuation determination method, for example, when the driving intention at time (t-1) is the right lane change LCR using the lateral position of the host vehicle and the traveling lane detection result, the host vehicle is still in the same lane at time t. When driving, it is determined that the lane change is continuing. On the other hand, if the vehicle is traveling in a lane different from the time (t-1) at time t, it is determined that the lane change is not being executed. When the driving intention at time (t-1) is the left lane change LCL, the driving intention at time t is set in the same manner.

  As described above, in step S104, a driving intention sequence is newly formed or disappeared based on the driving intention of the virtual driver at time (t-1) and the vehicle surrounding state of the host vehicle at the current time t. A plurality of driving intention sequences between the time points (t−m + 1) are generated.

  Note that the m time frames when generating the driving intention sequence are always the m time frames closest to the present, and when generating the driving intention sequence in the next period (t + 1), time (t + 1) To m time frames from the past time (t−m + 2).

  In Step S105, the virtual driver's driving operation amount Oid for each driving intention calculated in Step S102 and the actual driver's driving operation amount Ord detected in Step S103 are used to calculate the virtual driver's driving operation amount approximation degree Pid. calculate. Here, the degree of approximation Pid_lk, Pid_lcr, and Pid_lcl when the driving intention is lane keeping, right lane change, and left lane change are collectively expressed as Pid. Similarly, the steering angles θid_lk, θid_lcr, θid_lcl of the virtual driver when the driving intention is lane keeping, right lane change, and left lane change are collectively expressed as θid.

The virtual driver driving operation amount approximation degree Pid is a normalized (normalized) value of the steering angle θid of the virtual driver with respect to a normal distribution in which the steering angle θrd of the actual driver is an average value and the predetermined value ρrd is a standard deviation. The log probability can be calculated from the following (Equation 7).
Pid = log {Probn ((θid−θrd) / ρrd)} (Expression 7)
Here, Probn is a probability density conversion function for calculating the probability that a given sample is observed from a population represented by a normal distribution.

  As described above, in step S105, the approximate degree Pid_lk in the case of maintaining the lane, the approximate degree Pid_lcr in the case of changing the right lane, and the approximate degree Pid_lcl in the case of changing the left lane are calculated using (Equation 7). Thereafter, the process proceeds to step S106.

In subsequent step S106, the virtual driver driving operation amount sequence approximation degree Pids is calculated for each of the dynamic driving intention sequences of the plurality of virtual drivers generated in step S104. Here, as shown in FIG. 6, m driving operation amount approximation degrees Pid (j) (t−m + 1) from the past time point (t−m + 1) belonging to each dynamic driving intention series j to the current time point t. ) To Pid (j) (t), a series approximation degree P (j) ids of the dynamic driving intention series j is calculated. It is assumed that the past driving operation amount approximation degree Pid is stored in a memory (not shown). The dynamic virtual driver driving operation amount sequence approximation degree P (j) ids can be calculated from the following (Equation 8).
P (j) ids = ΠPid (j) (t−i + 1) (Equation 8)
Here, Π is the virtual driver driving operation amount approximation degree Pid at the past time point (t−m + 1) from the virtual driver driving operation amount approximation degree Pid (j) (t) at the current time t in the dynamic driving intention series j. (j) represents the sum of products obtained by integrating all of (t-m + 1).

As described above, in step S106, the degree of approximation P (j) ids of each dynamic driving intention sequence j is calculated using (Equation 8), and then the process proceeds to step S107.
In step S107, as expressed in the following (formula 9), the driving intention sequence j of the virtual driver having the maximum value of the dynamic virtual driver driving operation amount sequence approximation degree P (j) ids calculated in step S106 is set to the maximum. Is selected as the driving intention series Lmax having the likelihood of. Then, the current driving intention of the virtual driver having the driving intention series Lmax is selected as the actual driver's driving intention λrd.
λrd = max {Pids_lk, Pids_lcr, Pids_lcl} (Formula 9)

  In step S108 and subsequent steps, the state of the actual driver's driving intention λrd estimated in step S107 is determined. For example, even if it is estimated that the driving intention is a lane change at the current time t, how long the actual driver has recalled the lane change differs depending on the driving intention series j.

  Specifically, as shown in the series L1 in FIG. 4, when the driving intention changes to the right lane change LCR at the current time t, and as shown in the series L2, the right lane changes at the time (t-1) one frame before. When changing to LCR, or when recalling the right lane change LCR from the past time (t−m + 3) as shown in the series L3, the time from when the actual driver recalled the right lane change LCR And the state of the estimated driving intention λrd is different. Therefore, an actual driver estimates the elapsed time etlc from recalling the driving intention λrd to the current time t, and determines the state of the driving intention λrd estimated in step S107.

First, in step S108, the current time t is substituted as a variable k for searching backward from the current time tlc after the appearance of the driving intention λrd. In step S109, the driving intention sequence Lmax of the virtual driver having the maximum likelihood is referred to, and the driving intention λdid (k) of the driving intention sequence Lmax at time k is expressed by the following (Equation 10), step S107. It is determined whether or not the driving intention λrd estimated in step 1 is satisfied.
λrd = Lmax (λdid (k)) (Equation 10)

  If an affirmative determination is made in step S109 and the driving intention λdid (k) of the driving intention sequence Lmax at time k is the same as the estimated driving intention λrd, the process proceeds to step S110. In step S110, it is determined whether all m driving intentions λdid (k) in the driving intention series Lmax have been searched. If a negative determination is made in step S110, the process proceeds to step S111. In step S111, the time (k-1) one frame before is set as the time k. Thereafter, the process returns to step S109, and it is determined again whether or not the driving intention λdid (k) of the driving intention sequence Lmax at the time k is the driving intention λrd.

  If the determination in step S109 is negative and the driving intention λdid (k) of the driving intention sequence Lmax at time k is not the driving intention λrd estimated in step S107, the process proceeds to step S112. For example, when the driving intention λrd is the right lane change LCR, if the driving intention λdid (k) of the driving intention sequence Lmax at the time k becomes the lane keeping LK, a negative determination is made in step S109 and the process proceeds to step S112. Further, when the determination in step S110 is affirmative and the search for the driving intention λdid (k) is performed up to the first time frame (t−m + 1) of the driving intention sequence Lmax, the process proceeds to step S112.

In step S112, the elapsed time from the appearance of the driving intention λrd for the first time in the driving intention series Lmax to the present time, that is, the estimated driving intention elapsed time etlc of the actual driver is calculated from the following (formula 11).
etlc = (t−k) × ΔT (Equation 11)
In (Expression 11), ΔT is a sampling period of the driving intention sequence generation process, and is, for example, a time interval between the current time t and the time (t−1) one frame before. The time k is the time when the driving intention λrd appears in the driving intention series Lmax.

  As shown in FIG. 6, in the series L1, since the driving intention of the right lane change LCR appears at the current time t, the time t is substituted into the variable k in (Equation 11). Thereby, the estimated driving intention elapsed time etlc in the series L1 is estimated to be 0 sec. In the series L2, since the driving intention of the right lane change LCR appears at the time point (t-1) one frame before, the time (t-1) is substituted into (Equation 11), and the estimated driving intention elapsed time etlc is set to ΔTsec. Estimated. Similarly, in the series L3, since the intention of the right lane change LCR has appeared since the past time (t−m + 3), the estimated driving intention elapsed time etlc is estimated to be ΔT × (m−3) sec.

After calculating the estimated driving intention elapsed time etlc as described above, the process proceeds to step S113.
In step S113, the actual driving intention λrd of the driver estimated in step S107 and the estimated driving intention elapsed time etlc calculated in step S112 are output. Thus, the current process is terminated.

Thus, in the first embodiment described above, the following operational effects can be achieved.
(1) The driving intention estimation device 1 dynamically generates driving intention sequences of a plurality of virtual drivers in a past predetermined time interval including the current time t, and the driving operation amount Oid of the virtual driver for each driving intention sequence j. A driving operation amount series approximation degree P (j) ids representing a series approximation degree with the actual driving operation amount Ord of the driver is calculated. Then, the actual driving intention λrd of the driver is estimated by comparing a plurality of driving operation amount series approximation degrees P (j) ids, and the state of the estimated driving intention λrd is estimated. Thus, by estimating the state of the driving intention λrd, it is possible to determine the actual state of the driver when the driving intention estimation unit 70 estimates that the actual driver's intention is λrd.
(2) The driving intention state determination unit 80 estimates an elapsed time etlc after the estimated driving intention λrd is recalled by an actual driver as the state of the driving intention λrd. This makes it possible to accurately estimate the state of the driving intention λrd, that is, how long the actual driver intended the driving action.
(3) The lane change intention estimated by the drive intention estimation unit 70 by estimating the elapsed time (lane change intention elapsed time) etlc after the actual driver recalled the lane change in the drive intention state determination unit 80 You can know how certain it is. In the above-described first embodiment, the estimated driving intention elapsed time etlc is calculated for the estimated driving intention λrd, but the estimated driving intention elapsed time etlc is calculated only when the lane change intention is estimated. It is also possible.

  In the first embodiment described above, the steering angles θrd and θid are used as the actual driving operation amount Ord of the driver and the driving operation amount Oid of the virtual driver. However, it is not limited to this, For example, the operation amount of an accelerator pedal can also be used as a driving operation amount. In this case, the pedal operation amount Sid of the virtual driver can be calculated based on the degree of approach to the preceding vehicle represented by the inter-vehicle distance between the host vehicle and the preceding vehicle, the inter-vehicle time THW, and the like. The actual driver's driving intention is estimated based on the degree of approximation between the actual driver's pedal operation amount Srd and the virtual driver's pedal operation amount Sid.

  In the first embodiment described above, in order to calculate the driving operation amount Oid of the virtual driver, two forward reference points are set for each driving intention as shown in FIG. Can be set arbitrarily.

<< Second Embodiment >>
A vehicle driving assistance device according to a second embodiment of the present invention will be described with reference to the drawings. FIG. 7 is a system diagram showing a configuration of the vehicle driving assistance device 100 according to the second embodiment of the present invention, and FIG. 8 is a configuration diagram of a vehicle on which the vehicle driving assistance device 100 is mounted. .

First, the configuration of the vehicle driving operation assistance device 100 will be described.
The laser radar 110 is attached to the front grill or bumper of the vehicle, and scans the front area of the host vehicle by irradiating infrared light pulses in the horizontal direction. The laser radar 110 measures the reflected wave of the infrared light pulse reflected by a plurality of front reflectors (usually the rear end of the preceding vehicle), and determines the inter-vehicle distance to the preceding vehicle from the arrival time of the reflected wave. Detect relative speed. The detected inter-vehicle distance and relative speed are output to the controller 150. The forward area scanned by the laser radar 110 is about ± 6 deg with respect to the front of the host vehicle, and a forward object existing in this range is detected.

  The front camera 120 is a small CCD camera, a CMOS camera, or the like attached to the upper part of the front window, and detects the state of the front road as an image. The image signal from the front camera 120 is subjected to image processing by the image processing device 130 and output to the controller 150. The detection area by the front camera 120 is about ± 30 deg in the horizontal direction with respect to the front-rear direction center line of the vehicle, and the front road scenery included in this area is captured as an image.

The vehicle speed sensor 140 detects the vehicle speed of the host vehicle by measuring the number of wheel rotations and the number of rotations on the output side of the transmission, and outputs the detected host vehicle speed to the controller 150.
Furthermore, the actual driver's driving intention λrd and the driving intention elapsed time etlc estimated by the driving intention estimation apparatus 1 of the first embodiment described above are input to the controller 150.

  The controller 150 includes a CPU and CPU peripheral components such as a ROM and a RAM. The controller 150 configures a risk potential calculation unit 151, an accelerator pedal reaction force command value calculation unit 152, and an accelerator pedal reaction force command value correction unit 153, for example, depending on the software form of the CPU.

  The risk potential calculation unit 151 uses the host vehicle speed, the inter-vehicle distance, the relative vehicle speed relative to the preceding vehicle input from the laser radar 110 and the vehicle speed sensor 140, and the image information around the vehicle input from the image processing device 130. The surrounding risk potential RP is calculated. The accelerator pedal reaction force command value calculation unit 152 calculates a command value FA of the accelerator pedal reaction force generated by the accelerator pedal 160 based on the risk potential RP calculated by the risk potential calculation unit 151.

  The accelerator pedal reaction force command value correction unit 153 is based on the driving intention estimation result input from the driving intention estimation device 1 and the driving intention elapsed time etlc, and is calculated by the accelerator pedal reaction force command value calculation unit 152. The force command value FA is corrected. The accelerator pedal reaction force command value FAc corrected by the accelerator pedal reaction force command value correction unit 153 is output to the accelerator pedal reaction force control device 170.

  The accelerator pedal reaction force control device 170 controls the accelerator pedal operation reaction force according to a command value from the controller 150. As shown in FIG. 9, the accelerator pedal 160 is connected to a servo motor 180 and an accelerator pedal stroke sensor 181 via a link mechanism. The servo motor 180 controls the torque and the rotation angle in accordance with a command from the accelerator pedal reaction force control device 170, and arbitrarily controls the operation reaction force generated when the driver operates the accelerator pedal 160. The accelerator pedal stroke sensor 181 detects the stroke amount S of the accelerator pedal 160 converted into the rotation angle of the servo motor 180 via the link mechanism.

  Note that the normal accelerator pedal reaction force characteristic when the accelerator pedal reaction force control is not performed is set so that, for example, the accelerator pedal reaction force increases linearly as the stroke amount S increases. The normal accelerator pedal reaction force characteristic can be realized by the spring force of a torsion spring (not shown) provided at the center of rotation of the accelerator pedal 160, for example.

  Next, the operation of the vehicle driving assistance device 100 according to the second embodiment will be described in detail with reference to FIG. FIG. 10 is a flowchart showing the processing procedure of the driving operation assistance control program in the controller 150. This processing content is continuously performed at regular intervals (for example, 50 msec).

  In step S201, an environmental state quantity representing the traveling environment around the host vehicle detected by the laser radar 110, the front camera 120, and the vehicle speed sensor 140 is read. Specifically, the inter-vehicle distance D between the host vehicle and the preceding vehicle, the preceding vehicle speed V2, and the host vehicle speed V1 are read. In step S202, a risk potential RP around the host vehicle is calculated based on the travel environment read in step S201. Here, in order to calculate the risk potential RP around the host vehicle, a margin time TTC and an inter-vehicle time THW for the preceding vehicle are calculated.

The margin time TTC is a physical quantity indicating the current degree of proximity of the host vehicle with respect to the preceding vehicle. The allowance time TTC is the number of seconds after which the inter-vehicle distance D becomes zero when the current traveling state continues, that is, when the host vehicle speed V1, the preceding vehicle speed V2, and the relative vehicle speed Vr (Vr = V2-V1) are constant. And a value indicating whether or not the preceding vehicle is in contact. The margin time TTC is obtained by the following (Equation 12).
TTC = −D / Vr (12)

  The smaller the margin time TTC value, the closer the contact with the preceding vehicle, and the greater the degree of approach to the preceding vehicle. For example, when approaching a preceding vehicle, it is known that most drivers start a deceleration action before the margin time TTC becomes 4 seconds or less.

The inter-vehicle time THW is an effect when it is assumed that the degree of influence on the margin time TTC due to a change in the vehicle speed of the assumed vehicle ahead, that is, the relative vehicle speed vr changes when the host vehicle follows the preceding vehicle. It is a physical quantity indicating the degree. The inter-vehicle time THW is expressed by the following (formula 13).
THW = D / V1 (Formula 13)

  The inter-vehicle time THW is obtained by dividing the inter-vehicle distance D by the own vehicle speed V1, and indicates the time until the own vehicle reaches the current position of the preceding vehicle. The greater the inter-vehicle time THW, the smaller the predicted influence level with respect to the surrounding environmental changes. That is, when the inter-vehicle time THW is large, even if the vehicle speed of the preceding vehicle changes in the future, the degree of approach to the preceding vehicle is not greatly affected, and the margin time TTC does not change so much. When the own vehicle follows the preceding vehicle and the own vehicle speed V1 = the preceding vehicle speed V2, the inter-vehicle time THW can be calculated using the preceding vehicle speed V2 instead of the own vehicle speed V1 in (Equation 13).

Then, the risk potential RP for the preceding vehicle is calculated using the calculated margin time TTC and the inter-vehicle time THW. The risk potential RP for the preceding vehicle can be calculated using the following (Equation 14).
RP = a / THW + b / TTC (Formula 14)

  As shown in (Formula 14), the risk potential RP is a physical quantity that is continuously expressed from the margin time TTC and the inter-vehicle time THW. Here, a and b are constants for appropriately weighting the inter-vehicle time THW and the margin time TTC, and appropriate values are set in advance. The constants a and b are set to, for example, a = 1 and b = 8 (a <b).

  In step S203, the stroke amount S of the accelerator pedal 160 detected by the accelerator pedal stroke sensor 181 is read. In step S204, an accelerator pedal reaction force command value FA is calculated based on the risk potential RP calculated in step S202. First, the reaction force increase amount ΔF corresponding to the risk potential RP is calculated.

  FIG. 11 shows the relationship between the risk potential RP for the preceding vehicle and the reaction force increase amount ΔF. As shown in FIG. 11, when the risk potential RP is less than or equal to the minimum value RPmin, the reaction force increase amount ΔF is set to zero. This is to avoid annoying the driver by increasing the accelerator pedal reaction force FA when the risk potential RP around the host vehicle is very small. As the minimum value RPmin, an appropriate value is set in advance.

In the region where the risk potential RP exceeds the minimum value RPmin, the reaction force increase amount ΔF is set to increase exponentially according to the risk potential RP. The reaction force increase amount ΔF is expressed by the following (Equation 15).
ΔF = k · RP ⁿ (Expression 15)
Here, the constants k and n are different depending on the vehicle type and the like, and are appropriately set in advance so that the risk potential RP can be effectively converted into the reaction force increase amount ΔF based on the results obtained by a drive simulator or a field test. Keep it.

  Further, the accelerator pedal reaction force command value FA is calculated by adding the reaction force increase amount ΔF calculated according to (Equation 15) to the normal reaction force characteristic corresponding to the accelerator pedal stroke amount S.

  In step S205, the driving intention λrd estimated by the driving intention estimation device 1 is read, and it is determined whether or not the driver's driving intention λrd estimated in step S206 is a lane change. If the driver's driving intention λrd is a lane change, the process proceeds to step S207. In step S207, the estimated driving intention elapsed time etlc calculated by the driving intention estimation device 1 is read.

  In step S208, the accelerator pedal reaction force command value FA calculated in step S204 is corrected based on the estimated driving intention elapsed time etlc input in step S207, that is, the elapsed time etlc since the actual driver recalled the lane change. To do. Specifically, the accelerator pedal reaction force command value FA calculated in step S204 is subjected to filter processing such as a low-pass filter to be attenuated.

The corrected accelerator pedal reaction force command value FA can be expressed using the following (Equation 16). In (Equation 16), the corrected reaction force command value FA is expressed as FAc as a control reaction force command value.
FAc = gf (FA)
= Kf · {1 / (1 + Tsf_etlc)} · FA (Expression 16)
Here, kf is an appropriately set constant, and Tsf_etlc is a time constant of the low-pass filter. The time constant Tsf_etlc is set as a function of the actual driver's estimated driving intention elapsed time etlc as shown in the following (Equation 17).
Tsf_etlc = f (etlc) (Expression 17)

  FIG. 12 shows the relationship between the estimated driving intention elapsed time etlc and the time constant Tsf_etlc. As shown in FIG. 12, the time constant Tsf_etlc is set to become smaller as the elapsed time etlc becomes longer. In this way, the accelerator pedal reaction force can be quickly attenuated by decreasing the time constant Tsf_etlc as the elapsed time etlc from the actual driver recalls the driving intention.

  On the other hand, if it is determined in step S206 that the driving intention λrd estimated by the driving intention estimation device 1 is lane keeping, the process proceeds to step S209, and the accelerator pedal reaction force command value FA calculated in step S204 is used for control. Is set as the command value FAc.

  In step S210, the accelerator pedal reaction force command value FAc set in step S208 or S209 is output to the accelerator pedal reaction force control device 170. The accelerator pedal reaction force control device 170 controls the servo motor 180 in accordance with a command input from the controller 150. Thus, the current process is terminated.

  The operation of the vehicle driving assistance device 100 will be described with reference to FIGS. FIG. 13A shows a state in which the host vehicle changes lanes from the driving lane to the overtaking lane on the left side, and FIG. 13B shows the time change of the accelerator pedal reaction force command value FAc when changing the lane. Show. When the driver's intention to change lanes is detected at time t = ta, the accelerator pedal reaction force command value FAc gradually decreases.

  Here, when the estimated driving intention elapsed time etlc of the actual driver is t1, the accelerator pedal reaction force command value FAc gradually decreases as shown by a solid line in FIG. On the other hand, when the estimated driving intention elapsed time etlc is t2 longer than t1 (see FIG. 12), the accelerator pedal reaction force command value FAc greatly decreases as indicated by a broken line.

  In this way, the accelerator pedal reaction force command value FAc is lowered according to the actual driver's estimated driving intention elapsed time etlc. The longer the elapsed time etlc from when the driver intends to change the lane to when the driver's intention to drive is assumed to be the lane change, the faster the accelerator pedal reaction force command value FAc decreases. It is possible to perform accelerator pedal reaction force control that meets the driver's expectation without obstructing the vehicle.

Thus, in the second embodiment described above, the following operational effects can be achieved.
(1) The controller 150 calculates the risk potential RP based on the obstacle situation around the host vehicle, and performs accelerator pedal operation reaction force control based on the risk potential RP. At this time, the operation reaction force generated by the accelerator pedal 160 is corrected based on the estimated driving intention λrd and the state of the driving intention λrd. Thus, the accelerator pedal reaction force control meeting the driver's expectation can be performed while reflecting the actual driving intention of the driver while notifying the driver of the risk potential RP around the host vehicle as the operation reaction force.
(2) When the estimated driving intention λrd is a lane change, the controller 150 corrects the operation reaction force generated by the accelerator pedal 160 based on the lane change intention elapsed time etlc. As a result, the accelerator pedal reaction force control suitable for the driver's intention can be performed when changing the lane.
(3) The controller 150 includes an accelerator pedal reaction force command value correction unit 153 that corrects the relationship between the risk potential RP and the operation reaction force. The accelerator pedal reaction force command value correction unit 153 decreases the operation reaction force as the lane change intention elapsed time etlc becomes longer. As a result, when the actual driver intends to change the lane for a long time and expects a quick decrease in the accelerator pedal reaction force, the accelerator pedal reaction force control meeting the driver's expectation can be performed.

-Modification of the second embodiment-
Here, the accelerator pedal reaction force command value FA is corrected only when the estimated driving intention elapsed time etlc of the actual driver exceeds the predetermined value etlco. Hereinafter, how the accelerator pedal reaction force command value FA is corrected will be described with reference to FIG. FIG. 14 is a flowchart showing the processing procedure of the driving operation assistance control program in the controller 150. This processing content is continuously performed at regular intervals (for example, 50 msec). The processing in steps S301 to S307 is the same as the processing in steps S201 to S207 described in the flowchart of FIG.

  In step S308, it is determined whether or not the estimated driving intention elapsed time etlc of the actual driver read in step S307 exceeds a predetermined value etlco. If the determination in step S308 is affirmative and the elapsed time etlc exceeds the predetermined value etlco, the process proceeds to step S309, and the accelerator pedal reaction force command value FA calculated in step S304 is corrected based on the estimated driving intention elapsed time etlc.

Specifically, the control reaction force command value FAc is calculated by the following (Equation 18).
FAc = gf (FA)
= Kf · {1 / (1 + Ksf × Tsf_etlc)} · FA (Expression 18)
In (Expression 18), Ksf is a coefficient relating to the time constant Tsf_etlc. The coefficient Ksf is set as a function of the estimated driving intention elapsed time etlc of the actual driver as shown in (Equation 19) below.
Ksf = f (etlc) (Equation 19)

  FIG. 15 shows the relationship between the estimated driving intention elapsed time etlc and the coefficient Ksf. As shown in FIG. 15, the coefficient Ksf decreases as the estimated driving intention elapsed time etlc increases in a region exceeding the predetermined value etlco. Note that the coefficient Ksf = 1 at the predetermined value etlco. Thereby, the time constant term (Ksf × Tsf_etlc) becomes smaller as the estimated driving intention elapsed time etlc becomes longer.

  On the other hand, if a negative determination is made in step S308, the process proceeds to step S310, and the accelerator pedal reaction force command value FA calculated in step S304 is set as it is as a reaction force command value FAc for control. In the subsequent step S311, the reaction force command value FAc set in step S309 or step S310 is output to the accelerator pedal reaction force control device 170. Thus, the current process is terminated.

  As described above, by correcting the accelerator pedal reaction force command value FAc in accordance with the estimated driving intention elapsed time etlc, the actual driver's intention to change the lane is detected as in the second embodiment described above. In this case, it is possible to attenuate the accelerator pedal reaction force and perform reaction force control that matches the driver's intention. Further, if the estimated driving intention elapsed time etlc is less than the predetermined value etlco, that is, if the elapsed time etlc from when the actual driver recalls the lane change until the driving intention is estimated to be a lane change is short, the accelerator The pedal reaction force command value FA is not corrected. As a result, the probability of correcting the reaction force command value FA based on an erroneous driving intention estimation result is reduced, and unnecessary pedal reaction force attenuation against the driver's intention is reduced, which reduces annoyance to the driver. It becomes possible to do.

  In the second embodiment described above, it is also possible to multiply the preset time constant Tsf_etlc by the coefficient Ksf and set the coefficient Ksf according to the estimated driving intention elapsed time etlc.

<< Third Embodiment >>
A vehicle driving assistance device according to a third embodiment of the present invention will be described with reference to the drawings. FIG. 16 is a system diagram showing a configuration of a vehicle driving assistance device 200 according to the third embodiment of the present invention. In FIG. 16, portions having the same functions as those of the second embodiment described above are denoted by the same reference numerals. Here, differences from the second embodiment will be described.

  In the third embodiment, when the driver's intention to change lanes is detected by the driving intention estimation device 1, the risk potential RP around the host vehicle is corrected. Therefore, the controller 150A of the vehicle driving assistance device 200 includes a risk potential calculation unit 151, a risk potential correction unit 154, and an accelerator pedal reaction force command value calculation unit 155.

  Next, the operation of the vehicle driving assistance device 200 according to the third embodiment will be described in detail with reference to FIG. FIG. 17 is a flowchart showing the processing procedure of the driving operation assistance control program in the controller 150A. This processing content is continuously performed at regular intervals (for example, 50 msec). The processing in steps S401 and S402 is the same as the processing in steps S201 and S202 described in the flowchart of FIG.

  In step S403, the actual driver's driving intention λrd estimated by the driving intention estimation apparatus 1 is read. In step S404, it is determined whether or not the estimated driving intention λrd of the driver is a lane change. If the driver's driving intention λrd is a lane change, the process proceeds to step S405. In step S405, the estimated driving intention elapsed time etlc calculated by the driving intention estimation device 1 is read.

  In step S406, the risk potential RP calculated in step S402 is corrected based on the estimated driving intention elapsed time etlc input in step S405, that is, the elapsed time etlc since the actual driver recalled the lane change. Specifically, the risk potential RP calculated in step S402 is subjected to filter processing such as a low-pass filter to be attenuated.

The corrected risk potential RP can be expressed using the following (formula 20). In (Equation 20), the risk potential after correction actually used for control is represented by RPc.
RPc = gp (RP)
= Kp · {1 / (1 + Tsp_etlc)} · RP (Expression 20)
Here, kp is an appropriately set constant, and Tsp_etlc is a time constant of the low-pass filter. The time constant Tsp_etlc is set as a function of the actual driver's estimated driving intention elapsed time etlc as shown in the following (Equation 21).
Tsp_etlc = f (etlc) (Formula 21)

  FIG. 18 shows the relationship between the estimated driving intention elapsed time etlc and the time constant Tsp_etlc. As shown in FIG. 18, the time constant Tsp_etlc is set to become smaller as the elapsed time etlc becomes longer. As described above, the longer the elapsed time etlc after the actual driver recalls the driving intention, the smaller the time constant Tsf_etlc, the lower the risk potential RP and the quicker the accelerator pedal reaction force can be attenuated. .

  On the other hand, if it is determined in step S404 that the driving intention λrd estimated by the driving intention estimation device 1 is lane keeping, the process proceeds to step S407, and the risk potential RPc calculated in step S402 is directly used as the control risk potential RPc. Set as.

  In step S408, the accelerator pedal stroke amount S detected by the accelerator pedal stroke sensor 181 is read. In step S409, the accelerator pedal reaction force command value FA is calculated based on the risk potential RPc set in step S406 or S407. The calculation process of the accelerator pedal reaction force command value FA is the same as the process in step S204 of FIG.

  In step S410, the accelerator pedal reaction force command value FA calculated in step S409 is output to the accelerator pedal reaction force control device 170. Thus, the current process is terminated.

  The operation of the vehicle driving assist device 200 will be described with reference to FIGS. 19 (a) and 19 (b). FIG. 19A shows a state in which the host vehicle changes lanes from the driving lane to the overtaking lane on the left side, and FIG. 19B shows the time change of the accelerator pedal reaction force command value FA when the lane is changed. Show. When the driver's intention to change lanes is detected at time t = ta, the accelerator pedal reaction force command value FA gradually decreases.

  Here, when the estimated driving intention elapsed time etlc of the actual driver is t1, the accelerator pedal reaction force command value FA gradually decreases as shown by the solid line in FIG. On the other hand, when the estimated driving intention elapsed time etlc is t2 longer than t1 (see FIG. 18), the accelerator pedal reaction force command value FA greatly decreases as indicated by a broken line.

  In this way, the risk potential RP is attenuated according to the estimated driving intention elapsed time etlc of the actual driver, and the accelerator pedal reaction force command value FA is lowered. The longer the elapsed time etlc from when the driver intends to change the lane to when the driver's intention to drive is assumed to be the lane change, the faster the risk potential RP is attenuated and the accelerator pedal reaction force command value FAc is reduced. Therefore, accelerator pedal reaction force control that meets the driver's expectation can be performed without hindering the driver's pedal operation.

As described above, in the third embodiment described above, the following operational effects can be obtained in addition to the effects of the second embodiment described above.
The controller 150A includes a risk potential correction unit 154 that corrects the relationship between the obstacle situation and the risk potential RP. The risk potential correction unit 154 decreases the risk potential as the lane change intention elapsed time etlc becomes longer. As a result, when the actual driver intends to change the lane for a long time and expects a quick decrease in the accelerator pedal reaction force, the accelerator pedal reaction force control meeting the driver's expectation can be performed.

-Modification of the third embodiment-
Here, the risk potential RP is corrected only when the estimated actual driving intention time etlc of the driver exceeds the predetermined value etlco. Hereinafter, how the risk potential RP is corrected will be described with reference to FIG. FIG. 20 is a flowchart showing a processing procedure of a driving operation assistance control program in the controller 150A. This processing content is continuously performed at regular intervals (for example, 50 msec). The processing in steps S501 to S505 is the same as the processing in steps S401 to S405 described in the flowchart of FIG.

  In step S506, it is determined whether or not the estimated driving intention elapsed time etlc of the actual driver read in step S505 exceeds a predetermined value etlco. If the determination in step S506 is affirmative and the elapsed time etlc exceeds the predetermined value etlco, the process proceeds to step S507. In step S507, the risk potential RP calculated in step S502 is corrected based on the estimated driving intention elapsed time etlc.

Specifically, the risk potential RPc for control is calculated by the following (Equation 22).
RPc = gp (RP)
= Kp · {1 / (1 + Ksp × Tsp_etlc)} · RP (Formula 22)
In (Expression 22), Ksp is a coefficient relating to the time constant Tsp_etlc. The coefficient Ksp is set as a function of the estimated driving intention elapsed time etlc of the actual driver as shown in the following (Equation 23).
Ksp = f (etlc) (Equation 23)

  FIG. 21 shows the relationship between the estimated driving intention elapsed time etlc and the coefficient Ksp. As shown in FIG. 21, the coefficient Ksp decreases as the estimated driving intention elapsed time etlc increases in a region exceeding the predetermined value etlco. Note that the coefficient Ksp = 1 at the predetermined value etlco. Thereby, the time constant term (Ksp × Tsp_etlc) becomes smaller as the estimated driving intention elapsed time etlc becomes longer.

  On the other hand, if a negative determination is made in step S506, the process proceeds to step S508, and the risk potential RP calculated in step S502 is set as it is as the risk potential RPc for control. The processing in subsequent steps S509 to S511 is the same as the processing in steps S408 to S410 in FIG.

  As described above, by correcting the risk potential RP according to the estimated driving intention elapsed time etlc, as in the third embodiment described above, when an actual driver's lane change intention is detected, The reaction force control suitable for the driver's intention can be performed by attenuating the accelerator pedal reaction force. Furthermore, if the estimated driving intention elapsed time etlc is less than the predetermined value etlco, that is, if the elapsed time etlc from when the actual driver recalls the lane change until the driving intention is estimated to be a lane change is short, the risk The potential RP is not corrected. As a result, the probability of correcting the risk potential RP based on an erroneous driving intention estimation result is lowered, and unnecessary pedal reaction force attenuation contrary to the driver's intention is reduced so as to reduce annoyance to the driver. Is possible.

  In the third embodiment described above, it is also possible to multiply the preset time constant Tsp_etlc by the coefficient Ksp and set the coefficient Ksp according to the estimated operation intention elapsed time etlc.

  In the second and third embodiments described above, the risk potential RP is calculated using the margin time TTC and the inter-vehicle time THW between the host vehicle and the preceding vehicle. However, the present invention is not limited to this. For example, the reciprocal of the margin time TTC can be used as the risk potential.

  In the first to third embodiments described above, the vehicle ambient state detecting unit 10 as the vehicle ambient state detecting unit, the driving operation amount detecting unit 30 as the driving operation amount detecting unit, and the virtual as the virtual driver driving operation amount calculating unit. As a driver driving operation amount calculation unit 50, a virtual driver driving intention sequence dynamic determination unit 40 as a driving intention sequence generation means, a virtual driver driving operation amount approximation degree calculation unit 60 as a driving operation amount sequence approximation degree calculation means, and a driving intention estimation means The driving intention estimation unit 70 and the driving intention state determination unit 80 are used as driving intention state determination means. Further, the laser radar 110, the front camera 120 and the vehicle speed sensor 140 are used as the obstacle detection means, the risk potential calculation section 151 as the risk potential calculation means, and the accelerator pedal reaction force command value calculation sections 152 and 155 as the operation reaction force calculation means. The accelerator pedal reaction force control device 170 was used as the operation reaction force generating means. Further, the controllers 150 and 150A are used as the correction means, the accelerator pedal reaction force command value correction unit 153 is used as the operation reaction force correction means, and the risk potential correction unit 154 is used as the risk potential correction means. However, the present invention is not limited to these, and another type of millimeter wave radar or the like can be used as the obstacle detection means. It is also possible to use a steering reaction force control device that generates a steering reaction force as the operation reaction force generating means.

1: Driving intention estimation device 10: Vehicle ambient state detection unit 20: Vehicle state detection unit 30: Driving operation amount detection unit 40: Virtual driver driving intention sequence dynamic determination unit 50: Virtual driver driving operation amount calculation unit 60: Virtual driver Driving operation amount approximation degree calculation unit 70: driving intention estimation unit 80: driving intention state determination unit 100, 200: vehicle driving operation assisting device 150, 150A: controller 170: accelerator pedal reaction force control device

Claims (10)

Vehicle surrounding state detecting means for detecting the vehicle surrounding state of the host vehicle;
Driving operation amount detection means for detecting the driving operation amount by an actual driver;
For a plurality of different virtual drivers given driving intentions, driving required for each virtual driver to fulfill the driving intentions based on the vehicle surrounding state of the host vehicle detected by the vehicle surrounding state detection means A virtual driver operation amount calculation means for calculating an operation amount;
A driving intention sequence that dynamically determines the driving intention sequence of the plurality of virtual drivers in the predetermined time interval by respectively determining the driving intention of the plurality of virtual drivers in each time frame of a past predetermined time interval including the present time. Generating means;
A series approximation degree (hereinafter referred to as a driving operation quantity series approximation degree) of the driving operation quantity of the plurality of virtual drivers and the actual driving operation quantity of the actual driver in the predetermined time interval is the driving intention series. Driving operation amount series approximation degree calculating means for calculating for each driving intention series generated by the generating means;
Driving intention estimation means for estimating the driving intention of the actual driver by comparing the plurality of driving operation quantity series approximation degrees calculated by the driving operation quantity series approximation degree calculation means;
A driving intention estimation device comprising: a driving intention state determination unit that estimates a state of the driving intention with respect to the driving intention of the actual driver estimated by the driving intention estimation unit. In the driving intention estimation device according to claim 1,
The driving intention state determination means estimates an elapsed time from when the driving intention estimated by the driving intention estimation means is recalled by the actual driver as the driving intention state. Estimating device. In the driving intention estimation device according to claim 2,
The driving intention of the plurality of virtual drivers is lane keeping and lane change,
The driving intention estimation device, wherein the driving intention state determination means estimates a lane change intention elapsed time after the actual driver recalls a lane change. Vehicle surrounding state detecting means for detecting the vehicle surrounding state of the host vehicle, driving operation amount detecting means for detecting a driving operation amount by an actual driver, and a plurality of different virtual drivers given driving intentions A virtual driver driving operation amount calculating means for calculating a driving operation amount necessary for each virtual driver to perform the driving intention based on a vehicle surrounding state of the host vehicle detected by the state detecting means; Driving intention sequence generation means for determining the driving intentions of the plurality of virtual drivers in each time frame of a past predetermined time interval and dynamically generating the driving intention sequences of the plurality of virtual drivers in the predetermined time interval; A system of the driving operation amount of the plurality of virtual drivers and the actual driving operation amount of the actual driver in the predetermined time interval A driving operation amount sequence approximation degree calculating means for calculating the approximate degree of approximation (hereinafter referred to as driving operation amount series approximation degree) for each driving intention sequence generated by the driving intention sequence generating means, and the driving operation amount The driving intention estimation means for estimating the driving intention of the actual driver by comparing the plurality of driving operation amount series approximation degrees calculated by the series approximation degree calculation means, and the driving intention estimation means estimated by the driving intention estimation means About the driving intention of an actual driver, a driving intention estimation device having driving intention state determination means for estimating the driving intention state;
Obstacle detection means for detecting an obstacle situation around the host vehicle;
Risk potential calculating means for calculating a risk potential around the host vehicle based on a detection result by the obstacle detecting means;
Based on the risk potential calculated by the risk potential calculation means, an operation reaction force calculation means for calculating an operation reaction force of an accelerator pedal,
Operation reaction force generating means for generating the operation reaction force on the accelerator pedal;
Vehicular driving operation assistance comprising correction means for correcting the operation reaction force generated by the accelerator pedal based on the driving intention and the state of the driving intention estimated by the driving intention estimation device apparatus. The vehicle driving operation assistance device according to claim 4,
The driving intention of the plurality of virtual drivers is lane keeping and lane change,
The driving intention state determination means estimates a lane change intention elapsed time after the actual driver recalls a lane change,
When the driving intention estimation device estimates that the driving intention is a lane change, the driving intention estimation device includes a correction unit that corrects the operation reaction force generated by the accelerator pedal based on the lane change intention elapsed time. A driving operation assisting device for a vehicle. The vehicle driving assistance device according to claim 5,
The correction means is an operation reaction force correction means that corrects a relationship between the risk potential and the operation reaction force, and the operation reaction force correction means decreases the operation reaction force as the lane change intention elapsed time becomes longer. A driving operation assisting device for a vehicle. The vehicle driving operation assistance device according to claim 6,
The driving reaction assisting device for a vehicle, wherein the operation reaction force correcting means corrects a relationship between the risk potential and the operation reaction force when the lane change intention elapsed time is a predetermined value or more. The vehicle driving assistance device according to claim 5,
The correction unit is a risk potential correction unit that corrects a relationship between the obstacle situation and the risk potential, and the risk potential correction unit decreases the risk potential as the lane change intention elapsed time becomes longer. A driving operation assisting device for a vehicle. The vehicle driving operation assistance device according to claim 8,
The risk potential correcting means corrects a relationship between the obstacle situation and the risk potential when the lane change intention elapsed time is a predetermined value or more. A vehicle comprising the vehicular driving assist device according to any one of claims 4 to 9.

Images