JP2011186878A - Traveling object travel route generating device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a traveling object travel route generating device which does not give sense of incongruity to a driver by making an avoidance direction to an obstacle on a calculated obstacle avoidance route coincide with an obstacle avoidance direction set on the basis of a driving operation. <P>SOLUTION: In (a), an obstacle avoidance direction X1 based on a driving operation and an initial target avoidance route candidate X2 are set. In (b), a virtual obstacle 32a for generating a second risk potential is set at a position opposite from a detected obstacle 32 across the target avoidance route candidate X2. In (c), the target avoidance route candidate X2 is corrected into a target avoidance route candidate X3 on the basis of a second risk potential generated on the basis of the virtual obstacle 32a. In (d), a new virtual obstacle 32b is set, the second risk potential is regenerated on the basis of the new virtual obstacle 32b, and the target avoidance route candidate X3 is corrected into the final target travel route X4 on the basis of the second risk potential. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a mobile body travel route generation device that generates a travel route for a mobile body such as a vehicle to travel while appropriately avoiding an obstacle existing in front of the movement direction.

Conventionally, for example, a technique described in Patent Document 1 is known as a technique for setting a moving path of a moving body.
The movement path setting technique proposed by this Patent Document 1 generates a probability potential field that represents the probability that an obstacle may exist based on the position information of the detected obstacle,
After adding a gradient toward the target position to this stochastic potential field, search for a route toward the target position.
Thus, the obstacle avoidance behavior of the moving body can be realized without predicting the fluctuation of the sensor data while predicting the future action of the dynamic obstacle.

JP 2003-241836 A

According to this proposed technology, a potential field is formed based on the existence probability of an obstacle, and a path is generated based on the gradient of the potential field.
The passage direction (avoidance direction) of the generated path with respect to the obstacle is determined in a direction in which the gradient of the potential field is large.

However, when the driver drives the moving body, the passing direction desired by the driver and the direction in which the gradient of the potential field is large do not necessarily match.
When the above-described proposed technology is used, depending on the situation, the moving body may be moved in a direction different from the driver's intention, and the concern that the driver may feel uncomfortable cannot be eliminated.

  The present invention, after explicitly setting the obstacle avoidance direction, checks the avoidance direction with respect to the obstacle of the avoidance route candidate when generating the avoidance route, and the obstacle avoidance direction of the avoidance route candidate is the above-mentioned When the direction is opposite to the set direction, the avoidance route candidate is corrected so that the obstacle avoidance direction becomes the set direction as described above to generate a travel route of the moving body, thereby causing the above-mentioned problem of uncomfortable feeling. An object of the present invention is to propose a moving body travel route generation device that can wipe the sway.

  For this purpose, the moving body travel route generating apparatus according to the present invention includes an obstacle detection unit, an obstacle avoidance direction setting unit, a target avoidance route candidate setting unit, a first route evaluation unit, and an avoidance as follows. The configuration includes a direction difference degree determination unit, a virtual obstacle setting unit, a second route evaluation unit, a target avoidance route candidate correction unit, and an avoidance route determination unit.

The obstacle detection means detects the current position of the obstacle in front of the moving body driven by the driver.
The obstacle avoidance direction setting means determines an avoidance direction necessary for the mobile body to avoid the obstacle position detected by the obstacle detection means in a direction crossing the traveling direction of the mobile body based on the driver's operation. Set.
The target avoidance path candidate setting means sets a target avoidance path candidate necessary for avoiding the obstacle position detected by the obstacle detection means in a direction intersecting the moving body traveling direction.

The first route evaluation means calculates the degree of approach between the target avoidance route candidate and the detected obstacle based on the obstacle position information from the obstacle detection means.
The avoidance direction difference degree determination means is a difference degree between the avoidance direction set by the obstacle avoidance direction setting means and the avoidance direction of the target avoidance path candidate set by the target avoidance path candidate setting means with respect to the detected obstacle. Is determined.

The virtual obstacle setting means is based on the determination result of the avoidance direction difference degree determination means, and has moved from the obstacle position to the direction intersecting the moving body traveling direction for the target avoidance route candidate having a large difference degree. Set virtual obstacles.
The second route evaluation means calculates the degree of approach between the virtual obstacle and the target avoidance route candidate.

The target avoidance route candidate correction means corrects the target avoidance route candidate based on the evaluation value by the first route evaluation means or the second route evaluation means.
The avoidance route determination means determines an avoidance route based on the corrected target avoidance route candidate and the avoidance direction set by the obstacle avoidance direction setting means, and sets this as a moving body travel route.

In the above-described moving body travel route generation device according to the present invention, the obstacle avoidance direction setting unit sets the obstacle avoidance direction, and the target avoidance route candidate setting unit sets the obstacle avoidance target avoidance route candidate. If there is a possibility that the avoidance direction of the target avoidance path candidate for the obstacle and the avoidance direction of the obstacle set by the obstacle avoidance direction setting means may be different, the target avoidance path candidate correction means selects the target avoidance path candidate. The avoidance route is determined by correcting the avoidance direction so as to be directed to the obstacle avoidance direction set by the obstacle avoidance direction setting means.
For this reason, according to the moving body travel route generating apparatus of the present invention, the obstacle avoidance direction of the moving object can be surely matched with the set obstacle avoidance direction, and thus move in a direction different from the driver's intention. The above-mentioned problem of giving the driver a sense of incongruity without evading the body can be eliminated.

1 is a schematic plan view showing a four-wheeled vehicle 1 provided with a moving body travel route generating device according to a first embodiment of the present invention as viewed from above. FIG. 2 is a functional block diagram of the microprocessor in FIG. 2 is a flowchart showing a main routine of an obstacle avoidance path generation program executed by a microprocessor in FIG. FIG. 4 is a flowchart showing a subroutine related to an obstacle avoidance route calculation process in the main routine of FIG. FIG. 5 is a plan view illustrating the running state of the host vehicle used in explaining the obstacle avoidance route generation program of FIGS. 6 is an explanatory diagram showing a coordinate system set in order to define state quantities of the own vehicle and an obstacle in the example of FIG. FIG. 7 is an explanatory diagram showing the risk potential of the host vehicle with respect to obstacles and road boundaries obtained in the host vehicle driving situation of FIGS. It is a characteristic diagram which illustrates the nonlinear function which modeled tire lateral force. FIG. 4 is a characteristic diagram illustrating the second risk potential used in the obstacle avoidance route generation calculation according to FIGS. 3 and 4, (a) is a characteristic diagram when the obstacle avoidance direction of the vehicle is on the left side, (b) These are characteristic diagrams in the case where the obstacle avoidance direction of the own vehicle is on the right side. 5 and 6 are explanatory diagrams sequentially showing the calculation process of the obstacle avoidance route generation process according to FIGS. 3 and 4, in which (a) is an explanatory diagram showing an initial state at the start of calculation; ) Is an explanatory diagram when the second risk potential is applied. (C) is an explanatory diagram showing a final calculation result. FIG. 11 is an explanatory diagram showing a calculation result of the obstacle avoidance route generation process shown in FIG. 10, where (a) is a time chart showing a time-series change of the target front wheel turning angle, and (b) is an obstacle avoidance route of the own vehicle. FIG. FIGS. 5 and 6 are explanatory diagrams showing iterative calculation of the obstacle avoidance route generation process shown in FIGS. 3 and 4 in the traveling state of the vehicle, (a) is an explanatory diagram showing an initial state at the start of calculation, and (b) Is an explanatory diagram when the virtual obstacle position is set due to a mismatch between the obstacle avoidance direction based on the driving operation and the avoidance direction of the target avoidance route candidate, and (c) shows that the target avoidance route candidate is a virtual obstacle (D) shows the target avoidance route candidate is not appropriate even by the correction, and after the virtual obstacle position is reset, the target based on the reset virtual obstacle is displayed. It is explanatory drawing which shows the final calculation result when an avoidance path | route candidate is correct | amended again. FIG. 10 is a plan view illustrating the traveling state of the host vehicle used in explaining the obstacle avoidance route generation processing according to the second embodiment of the present invention. FIG. 13 is a time chart showing time-series changes in virtual obstacle positions related to the first obstacle and the second obstacle when the obstacle avoidance route generation process of the second embodiment is performed in the host vehicle traveling state of FIG. (A) is a time chart when the vehicle is set to avoid the first obstacle to the left and then the second obstacle to the left. (B) is the vehicle's first obstacle. (C) is a time chart when the second obstacle is set to avoid to the right after avoiding the object to the left, (c) shows the second obstacle after the vehicle avoids the first obstacle to the right (D) shows a time chart when the vehicle is set to avoid the first obstacle to the right and then the second obstacle to the right. It is a time chart. FIG. 13 is an explanatory diagram showing a result of performing the obstacle avoidance route generation process of the second example in the host vehicle traveling state of FIG. 13, (a) after the host vehicle avoids the first obstacle to the left, An explanatory diagram when setting to avoid the second obstacle to the left, (b) shows the vehicle avoiding the first obstacle to the left and then avoiding the second obstacle to the right (C) is an explanatory diagram when the vehicle is set to avoid the first obstacle to the right and then the second obstacle to the left. It is explanatory drawing at the time of setting so that a vehicle might avoid the 2nd obstacle to the right after avoiding the 1st obstacle to the right. FIG. 10 is a plan view illustrating the traveling state of the host vehicle used in describing the obstacle avoidance route generation processing according to the third embodiment of the present invention. FIG. 16 shows the calculation result when the obstacle avoidance route generation processing of the third embodiment is performed in the host vehicle traveling state of FIG. 16, (a) is a time chart showing the time series change of the obstacle moving speed parameter, b) is an explanatory diagram showing the movement status of the obstacle and the avoidance route of the own vehicle with respect to the moving obstacle.

Hereinafter, embodiments of the present invention will be described in detail based on examples shown in the drawings.
<First embodiment>
FIG. 1 shows a four-wheeled vehicle 1 having a moving body travel route generating apparatus according to a first embodiment of the present invention as viewed from above, in which 2 is a vehicle body, 3L and 3R are front left and right wheels, 4L and 4R are the left and right rear wheels, respectively.
In the vehicle 1 (moving body) shown in FIG. 1, the steering force from the steering wheel 5 that is steered by the driver is transmitted to the front wheel steering system via the steering column 6 and the steering gear box 7, and this steering system The front wheels 3L and 3R are steered.

As shown in Fig. 1, a set of two cameras 11L and 11R are installed facing forward in the vehicle body 2, and the road conditions ahead of the vehicle are photographed using these, and obstacles, road boundaries, lane markings, etc. Is detected.
Therefore, the cameras 11L and 11R correspond to the obstacle detection means in the present invention.
The two cameras 11L and 11R are separated from each other in the vehicle width direction and arranged approximately at the same level in the vehicle interior, so that not only the direction from the subject vehicle to the subject but also the distance from the subject vehicle to the subject is maintained. It shall be detectable.

The left and right front wheels 3L and 3R are non-driving wheels, and wheel speed sensors 12L and 12R for detecting these wheel speeds Vw are provided as vehicle speed sensors.
The vehicle speed sensors 12L and 12R can be composed of rotary encoders attached to the left and right front wheels 3L and 3R, and generate pulse signals with a pulse width proportional to the rotation of the left and right front wheels 3L and 3R. v can be measured.

The vehicle body 2 is provided with a yaw rate sensor 13 for detecting the yaw rate γ of the vehicle and an acceleration sensor 14 for detecting the acceleration α.
The yaw rate sensor 13 can use a known device configured using a crystal resonator or a semiconductor, and the acceleration sensor 14 can use a known device configured using a piezoelectric element or the like.
Note that the acceleration sensor 14 detects acceleration in a specific direction generated in the vehicle body 2, and in particular, detects the lateral acceleration α generated in the lateral direction of the vehicle body 2.

The steering column 6 is provided with a steering angle sensor 15 that detects the steering angle θ _s of the steering wheel 5 and a steering torque sensor 16 that detects the steering torque Ts from the steering wheel 5.
The steering angle sensor 15 detects the steering angle θ _s from the rotation angle of the steering shaft (not shown) in the steering column 6, and the steering torque sensor 16 is also to the steering shaft (not shown) in the steering column 6. It is assumed that the steering torque Ts is detected from this torque.

Here, the vehicle 1 in FIG. 1 not only steers the left and right front wheels 3L and 3R by the steering wheel 5, but also a steering shaft (not shown) in the steering column 6 by a steering assist motor 17 that performs a power steering function at this time. 2), the left and right front wheels 3L, 3R can be additionally steered.
The steering assist motor 17 is driven and controlled by a microprocessor 19 via a motor controller 18.

The microprocessor 19 is an integrated circuit including an A / D conversion circuit, a D / A conversion circuit, a central processing unit, a memory, and the like, and information from the cameras 11L and 11R and the various sensors 12L and 12R. , 13, 14, 15 and 16 are input.
Based on these input information, the microprocessor 19 executes the stored program in the memory to perform signal processing of the input information, and at the same time, the target travel route (avoidance route), that is, the left and right front wheels 3L and 3R. A time series signal of the target turning amount is calculated.

  The motor controller 18 responds to the calculated target turning amount every moment from the microprocessor 19 and applies a steering assist torque to a steering shaft (not shown) in the steering column 6 via the steering assist motor 17. By assigning, the left and right front wheels 3L and 3R are steered so as to be the target steer amount.

<Avoidance route control system>
FIG. 2 shows the above-described microprocessor 19 by a functional block diagram. The microprocessor 19 generally includes a sensor signal processing unit 21, an obstacle avoidance direction setting unit 22, a target avoidance route candidate setting unit 23, The target avoidance route candidate storage unit 24, the first risk potential generation unit 25, the second risk potential generation unit 26, the avoidance route correction unit 27, and the command value output processing unit 28.

The sensor signal processing unit 21 integrally processes information from the cameras 11L and 11R input to the microprocessor 19 and information from the various sensors 12L, 12R, 13, 14, 15, and 16, and The information is converted into information relating to the motion state, the driver's operation state, the position of the obstacle, and the road boundary, and the information is developed on the same coordinate system.
Although what kind of information is specifically converted will be described later, for the conversion process, many methods are already known, including methods for detecting obstacles and road boundaries by image processing, Detailed description is omitted here.

The obstacle avoidance direction setting unit 22 constitutes the obstacle avoidance direction setting means in the present invention, and the obstacle avoidance direction to be satisfied by the travel route generated from the operation state of the driver detected by the sensor signal processing unit 21. Set up.
The target avoidance route candidate setting unit 23 constitutes a target avoidance route candidate setting means in the present invention, sets any suitable avoidance route candidate prior to the calculation of the obstacle avoidance route,
The set target avoidance route candidate is stored in a predetermined area of the target avoidance route candidate storage unit 24.

The first risk potential generation unit 25 constitutes the first route evaluation means in the present invention, and based on the position of the obstacle detected by the sensor signal processing unit 21, an obstacle on an arbitrary position on the road A process of constructing a function that quantitatively represents the magnitude of the degree of approach to the is performed.
The second risk potential generation unit 26 constitutes a second path evaluation unit in the present invention, and the obstacle avoidance direction setting unit 22 sets the position of the obstacle detected by the sensor signal processing unit 21. A process of constructing a risk potential function in which a small risk evaluation value is assigned to the road area in the obstacle avoidance direction is performed.

  The avoidance route correction unit 27 constitutes target avoidance route candidate correction means in the present invention, reads out the target avoidance route candidate from the target avoidance route candidate storage unit 24, and further receives the first risk from the first risk potential generation unit 25. Reads the potential risk or the second risk potential from the second risk potential generator 26, corrects the target avoidance route candidate so that the risk evaluation value of the target avoidance route candidate becomes smaller, and calculates a new target avoidance route candidate Perform the process.

Switching between the first risk potential and the second risk potential to be read into the avoidance route correction unit 27, and the iterative process of the correction calculation of the target avoidance route candidate,
The target avoidance route candidate storage unit 24, the first risk potential generation unit 25, the second risk potential generation unit 26, and the avoidance route correction unit 27 are controlled by an avoidance route calculation unit 29.

That is, the obstacle avoidance direction setting unit based on the obstacle avoidance direction of the target avoidance route candidate calculated by the target avoidance route candidate setting unit 23 based on the position of the obstacle detected by the sensor signal processing unit 21, and the operation of the driver The obstacle avoidance direction set by 22 is compared.
If the obstacle avoidance direction of the target avoidance route candidate does not match the obstacle avoidance direction based on the driver's operation, the second risk potential from the second risk potential generation unit 26 is read into the avoidance route correction unit 27. ,
When the obstacle avoidance direction of the target avoidance route candidate matches the obstacle avoidance direction based on the driver's operation, the first risk potential from the first risk potential generation unit 25 is read into the avoidance route correction unit 27. .

The correction result of the avoidance route correction unit 27 after the iterative route calculation unit 29 has executed a predetermined number of iterations is the target avoidance route,
At the same time, the avoidance route correction unit 27 calculates the target front wheel turning amount every moment necessary for the vehicle 1 to travel along the target avoidance route as will be described later, and this target front wheel turning amount is processed as a command value output process. Sent to part 28.

The command value output processing unit 28 sequentially reads the time-series target front wheel turning amount that has been sent, and calculates the steering assist torque command value necessary to realize the read target turning amount. The command value is output to the motor controller 18.
The motor controller 18 supports the driver's steering operation so that the traveling route of the vehicle becomes the target avoidance route by drivingly controlling the steering assist motor 17 so that the steering assist torque command value is generated.

<Avoidance path generation calculation>
A control program executed by the microprocessor 19 for performing the above processing will be described in detail below with reference to FIGS.
However, to make the explanation easier to understand, the vehicle 1 is traveling on the straight road 31 between the left and right road boundaries 31L and 31R as shown in FIGS. The explanation will be developed for each specific example.

  FIG. 3 is a main routine that is repeatedly executed at predetermined time intervals. First, in step S1, information from the cameras 11L and 11R and signals from the various sensors 12L, 12R, 13, 14, 15, and 16 are micro-processed. The data is read into the memory of the processor 19, and based on these, the vehicle motion state and the obstacle information are calculated as values on an arbitrarily set coordinate system as follows.

Although the coordinate system can be determined arbitrarily, in this embodiment, as shown in FIGS. 5 and 6 (FIG. 6 shows a state after the vehicle 1 further travels from the position of FIG. 5), the vehicle on the road Set the X axis along the direction of travel and the Y axis along the direction perpendicular to the X axis.
A case will be described where the current position of the vehicle 1 shown in FIG. 5 is set to a coordinate system in which the origin of the X coordinate is set and the origin of the Y coordinate is set near the center line of the road.
By setting the coordinate system as described above, the position of the vehicle (center point) can be expressed in the form of (X, Y) = (x, y) as shown in FIG.

Further, as physical quantities representing the motion state of the host vehicle 1, the yaw angle θ, the vehicle speed v, the vehicle body side slip angle β, the yaw rate γ, and the front wheel steering angle δ are important motion physical quantities.
Among these physical movement quantities, the vehicle speed v can be approximated by the wheel speeds of the non-driven wheels, so that the detection values of the vehicle speed sensors 12L and 12R attached to the left and right front wheels 3L and 3R can be used as described above.

  The detected value of the yaw rate sensor 13 can be used as the yaw rate γ, and the yaw angle θ can be determined by determining an appropriate initial value and integrating the output value of the yaw rate sensor 13, or if it is a straight road, the road boundary It can be obtained by estimating the angle formed between 31L and 31R and the direction in which the vehicle 1 is facing by image processing.

の演算により求めることができる。
なお、車両縦方向の速度v_xを車速ｖで近似し、車両横方向の速度v_yを加速度センサ14の検出値である車両横加速度αの積分によって求めれば、上記の(1)式からすべり角βの近似値を得ることができる。
これ以外にも、車輪速Vw、ヨーレートγ、横加速度α等の信号からオブザーバによって、より精度良くすべり角βを推定する技術も知られているので、そのような手法を用いてすべり角βを得てもよい。 The slip angle β is defined as v _{x in} the longitudinal direction of the vehicle and v _{y in the} lateral direction of the vehicle.

Can be obtained by the following calculation.
If the vehicle longitudinal speed v _x is approximated by the vehicle speed v and the vehicle lateral speed v _y is obtained by integration of the vehicle lateral acceleration α, which is a detection value of the acceleration sensor 14, the slip can be obtained from the above equation (1). An approximate value of the angle β can be obtained.
In addition to this, there is also known a technique for estimating the slip angle β more accurately by an observer from signals such as the wheel speed Vw, the yaw rate γ, and the lateral acceleration α. Therefore, the slip angle β is calculated using such a method. May be obtained.

の演算により求めることができる。
以上に説明した通り、上記で挙げた自車1の運動状態を記述する運動物理量は全てセンサの検出信号を処理することによって具体的な値を算出することができる。 The front wheel steering angle δ is calculated from the steering angle θ _s detected by the steering angle sensor 15 and the gear ratio K of the steering system.

Can be obtained by the following calculation.
As described above, the physical physical quantities describing the motion state of the own vehicle 1 mentioned above can all be calculated by processing the detection signals of the sensors.

When the obstacle 32 is detected, as shown in FIG. 6, the coordinate position (x _p , y _p ) of the center point and the values of the width σ _y and the depth σ _x of the obstacle 32 are acquired by the camera. Calculated by processing image information.
The depth σ _x may be difficult to measure depending on the shooting direction. In this case, the depth σ _x is set to the same value as the width σ _{y for} convenience.
Of course, when the obstacle 32 is not detected, the physical quantity relating to the obstacle is not calculated.

Further, the positions of the left end and the right end of the road 31 obtained by detecting the road boundaries 31L, 31R by the cameras 11L, 11R are converted into values on the coordinate system, and Y = y _L and Y = y _R , respectively. To do.
As described above, in step S1 in FIG. 3, an arbitrary XY coordinate system is introduced, and information on the vehicle 1, the obstacle 32, and the road boundaries 31L and 31R is calculated as values on the XY coordinate system. To do.

In the next step S2, it is checked whether or not the current time is the update time of the target avoidance route.
If the target avoidance route update time is reached, the control proceeds to step S3. If the target avoidance route update time is not reached, the control proceeds to step S11.

In step S3, it is checked whether or not the obstacle 32 is detected. If the obstacle 32 is detected, the control proceeds to step S4. If the obstacle 32 is not detected, the control proceeds to step S11.
In step S4, the first risk potential is generated based on the detected information on the obstacle 32.
Since there are two types of obstacles 32 and road boundaries 31L and 31R that the vehicle 1 should avoid approaching in the scenes of FIGS. 5 and 6, these obstacles 32 and road boundaries 31L and 31R respectively. The overall risk potential is generated by taking the sum of the two terms.

The risk potential of the obstacle 32 can be expressed by a function whose value increases as the distance between the vehicle 1 and the obstacle 32 decreases.
Specifically, for example, a function of the following equation using a parameter w _P that defines the magnitude of the risk potential of the obstacle 32 can be used.

The risk potential of the road boundaries 31L and 31R is expressed by a function whose value increases as the distance between the vehicle 1 and the road boundaries 31L and 31R decreases.
Specifically, a parameter Δ _R that specifies the margin of approach to the road boundaries 31L and 31R (the larger the value, the more avoidance route that takes the approach margin with the road boundaries 31L and 31R is calculated), and the road boundary For example, a function of the following formula using the parameter w _R that defines the magnitude of the risk potential of 31L and 31R can be used.

と定義する。
第1リスクポテンシャルL₁をX―Y座標上にプロットした図を図5に示す。
図5における中央の山が障害物32に対応する関数L_Pによって形成されたリスクポテンシャルであり、両側の山が道路境界31L,31Rに対応する関数L_Rによって形成されたリスクポテンシャルである。 From the above two risk potentials L _P and L _R , the first risk potential L ₁ is

It is defined as
FIG. 5 shows a plot of the first risk potential L ₁ on the XY coordinates.
A risk potential central mountain is formed by the function L _P corresponding to the obstacle 32 in FIG. 5, a risk potential on both sides of the mountain is formed by road boundary 31L, function L _R corresponding to 31R.

In the next step S5 shown in FIG. 3, the obstacle avoidance direction of the target avoidance route generated based on the driver's operation is set.
Therefore, step S5 corresponds to the obstacle avoidance direction setting means in the present invention.
As shown in FIG. 7, the risk potential representing the obstacle 32 is a mountain shape having the peak at the obstacle detection position. Therefore, when the obstacle 32 exists near the center of the road 31, the right side of the obstacle 32 There is a region with a small risk potential on both the left sides, and there is a possibility that there is a route that can avoid the obstacle 32 in either direction.

When setting whether to generate a route that avoids the obstacle 32 in the left or right direction in step S5, for example, the steering angle θ _S is less than a predetermined threshold value and the driver is conspicuous steering If not, set the larger distance from the obstacle 32 to the road boundary 31L, 31R as the avoidance direction,
When the steering angle θ _S exceeds the above threshold and the driver is clearly steering, adopt a setting method such as selecting the steering direction as the avoidance direction. Can do.

In the next step S6, a target avoidance route candidate is set.
Therefore, step S6 corresponds to the target avoidance route candidate setting means in the present invention.
The target avoidance route is generally expressed as a time-series signal of the coordinate position (x, y) representing the own vehicle position. In this embodiment, a target value that more directly supports the driver's operation is generated. Therefore, a formulation for expressing the avoidance route by the time series change of the turning angle δ which is the operation amount of the own vehicle is performed.
In order to correlate the time-series change of the turning angle δ with the travel route of the host vehicle 1 on the road 31, a vehicle model that predicts the motion of the vehicle with the turning angle δ as an input is introduced.

ただし、ｍは車両質量、Iは車両ヨー慣性モーメント、l_fは車両重心から前輪軸までの距離、i_rは車両重心から後輪軸までの距離を表し、Y_f,Y_rはタイヤ横力をあらわす関数であり、それぞれ前輪すべり角β_f、後輪すべり角β_rの関数であると仮定している。 As a model that describes the motion of a vehicle, a two-wheel model that approximates the motion of a four-wheel vehicle by the motion of a two-wheel vehicle is generally well known.
Assuming that the vehicle speed v is constant, the two-wheel model is described by the following differential equation using the differential operator (d).

Where m is the vehicle mass, I is the vehicle yaw moment of inertia, l _f is the distance from the vehicle center of gravity to the front wheel axis, i _r is the distance from the vehicle center of gravity to the rear wheel axis, and Y _f and Y _r are the tire lateral forces. It is assumed that the functions are front wheel slip angle β _f and rear wheel slip angle β _r , respectively.

タイヤ横力関数Y_f,Y_rは、図8に示すような非線形関数で表現することができる。
以上の(6)〜(13)式をまとめると、状態ベクトルvxおよびその微分値dvxを用いた、前輪転舵角δを入力とする以下の微分方程式モデルが得られる。

ただし状態ベクトルvxは、vx＝（x y θ v β γ）と定義される。
車両が上記の(14)式に従って動くと仮定すると、入力である前輪転舵角δの時系列を決めれば、(14)式を積分することで状態ベクトルvxの時系列も決まることになる。 Note that the front wheel slip angle β _f and the rear wheel slip angle β _r can each be obtained by calculation of the following equations.

The tire lateral force functions Y _f and Y _r can be expressed by nonlinear functions as shown in FIG.
Summarizing the above formulas (6) to (13), the following differential equation model using the state vector vx and its differential value dvx and having the front wheel turning angle δ as an input can be obtained.

However, the state vector vx is defined as vx = (xy θ v β γ).
Assuming that the vehicle moves according to the above equation (14), if the time series of the input front wheel turning angle δ is determined, the time series of the state vector vx is also determined by integrating the equation (14).

と表すことができる。
ただし、t₀＝t t_i＝t＋(T/N)i,(i＝1,2,・・・N-1)である。
従って、指定された方向のリスクポテンシャルが小さな領域を通過するようなU(t)を算出することが走行経路算出の目的となる。 Since the state vector vx includes the coordinate position (x, y) of the own vehicle 1, the time series of vx includes information on the travel route.
That is, the avoidance route can be expressed by a time series of the steering angle δ.
For example, in this embodiment, when the T-second avoidance route is set to be calculated and the time interval of T seconds is equally divided into N and expressed as a signal obtained by sampling the value for each divided interval, the traveling at time t The route is generally

It can be expressed as.
However, t ₀ = tt _i = t + (T / N) i, (i = 1, 2,... N−1).
Therefore, the purpose of calculating the travel route is to calculate U (t) such that the risk potential in the specified direction passes through a small area.

Prior to the calculation of the route that matches the purpose, the content of the process in step S6 is to specifically set the candidate U (t).
Specifically, for example, when the travel route U ^* (t−Δt) is calculated at time t−Δt that is Δt before the current time t,
If Δt is sufficiently short, it can be considered that there is no significant change in the travel route. Therefore, a method of setting U ^* (t−Δt) as a candidate for U ^* (t) is conceivable.
If the travel route is not calculated at a time prior to the current time, it can be dealt with by setting U (t) = (0 0... 0).

In the next step S7, an avoidance route calculation process algorithm shown in the subroutine of FIG. 4 is executed.
In step S71 of FIG. 4, it is determined as follows to which side of the obstacle 32 the target avoidance route candidate set as described above in step S6 of FIG.
First, when the above equation (14) is integrated using U (t) obtained by the above equation (15), vx (t ₀ ), vx (t ₁ ),. (t _N-1 ) is obtained.

なお、x_P>x(t_N-1)となる場合には、t_P＝t_N-1 として上記(16)式の判定を行うこととする。 Here, since the x-axis of the coordinate system is taken along the traveling direction of the own vehicle, the x-coordinate component of the state vector can be regarded as monotonically increasing.
Thus, x _P is the x-coordinate of the obstacle _{32, x (t 0) <x} P ≦ x If (t _N-1) satisfies the condition of, x (t _P) = time satisfy x _P t _P is present, by approximating with a suitable interpolation technique linear interpolation or the like changes in the signal between the sample time, approximately calculate the time t _P in the vehicle of the y-coordinate value y (t _P) can do.
At this time, the obstacle avoidance direction of the target avoidance route candidate is determined as follows.

When x _P > x (t _N−1 ), the determination of the above equation (16) is performed with t _P = t _N−1 .

In the next step S72, the determination result based on the above equation (16) (the obstacle avoidance direction of the target avoidance route candidate) and the obstacle avoidance direction based on the driver's operation set in step S5 of FIG. If the two match, the control proceeds to step S73. If the two do not match, the control proceeds to step S76.
Therefore, step S72 corresponds to avoidance direction difference degree determination means in the present invention.

Step S73 corresponds to the first route evaluation means in the present invention. In this step S73, the first risk potential obtained in step S4 of FIG. 3 is read, and the avoidance route candidate correction calculation in the next step S74. Used for.
In step S74, a process for correcting the target avoidance route candidate so as to pass through a region having a smaller risk potential is performed.
Therefore, step S74 corresponds to the target avoidance route candidate correction means in the present invention.

ここで、Ψは時刻t＋Tにおける車両状態の望ましさを評価する評価式、Lは時刻tからt＋Tまでの間の各時刻における車両状態および操作量の望ましさを評価する評価式、τはtからt＋Tまで変化する積分変数である。 In order to formulate this correction process, an evaluation function of the following expression that numerically evaluates the target avoidance path candidate is introduced.

Here, Ψ is an evaluation formula that evaluates the desirability of the vehicle state at time t + T, L is an evaluation formula that evaluates the desirability of the vehicle state and the operation amount at each time between time t and t + T, and τ is from t It is an integral variable that changes up to t + T.

ただし、ｗ_Sは前輪舵角の大きさに対するペナルティの評価重みパラメータである。
従って、評価式Lは、

のように構成される。 In addition to incorporating the risk potential read in the previous step into the evaluation formula L, the size of the front wheel turning angle is also a risk to prevent the calculation of avoidance routes based on turning larger than necessary. Add the following evaluation term that imposes the same penalty as the potential.

Here, w _S is a penalty evaluation weight parameter for the magnitude of the front wheel steering angle.
Therefore, the evaluation formula L is

It is configured as follows.

のような評価式を導入する。
ただし、w_yは車両ヨー角のペナルティに対する評価重みを表すパラメータである。 Further, when the obstacle is avoided, the vehicle posture is deviated from the straight traveling state by the steering, so that an operation for resetting the vehicle posture after the obstacle avoidance is required.
In order to reflect the request for such an avoidance route, an evaluation term for evaluating the vehicle yaw angle θ at time t + T is introduced.
In this case, since the straight traveling state is defined as θ = 0, the following equation is imposed on the vehicle yaw angle at time t + T:

The following evaluation formula is introduced.
Here, w _y is a parameter representing the evaluation weight for the penalty of the vehicle yaw angle.

という不等式で制限する制約条件を最適制御問題に課すことにする。 When the evaluation function is defined as described above, the calculation of the steering angle time series can be formulated as an optimal control problem defined by the control object of the above equation (14) and the evaluation function of the equation (19). By using an algorithm that solves the control problem, a travel route with a small risk evaluation value can be calculated.
However, in order to prevent a very large front wheel rudder angle from being calculated as a command value, here, in addition to the penalty of equation (18), the range of values that can be taken as the command value for the front wheel rudder angle is set. ,

The constraint condition limited by the inequality is imposed on the optimal control problem.

When solving an optimal control problem, generally, an appropriate solution candidate is given, and a process of correcting the solution candidate is repeated.
In this case, a process of obtaining a new U (t) by applying an iterative calculation algorithm to the target avoidance path candidate U (t) stored in the storage area of the microprocessor 19 is performed.
Some specific algorithms for solving the optimal control problem are well known, and therefore detailed description thereof is omitted here.

In step S75, the result of the iterative calculation process in the previous step is checked to determine whether a condition for terminating the iterative calculation is satisfied.
Here, for example, it is possible to set a condition that the iterative operation is terminated when the number of iterations reaches a predetermined number or when the amount of change in the solution due to the iterative operation becomes smaller than a predetermined level. .
If the end condition is not satisfied, step S74 is repeatedly executed until the iterative calculation end condition is satisfied. If the iterative operation end condition is satisfied, the process exits the subroutine of FIG. Return to the routine and proceed to step 8.

Step S76 selected when it is determined in step S72 in FIG. 4 that the obstacle avoidance direction of the target avoidance route candidate does not match the obstacle avoidance direction based on the driver's operation set in step S5 in FIG. Then, by the same determination as in step S75, it is checked whether or not a condition for ending the iterative calculation is satisfied.
When the iterative calculation end condition is not satisfied, the control proceeds to step S77, and when the iterative calculation end condition is satisfied, the control proceeds to step S79.

In step S77, which is selected when the iterative calculation end condition is not satisfied, a second risk potential is generated.
This second risk potential is a risk potential for promoting that any desired target avoidance route candidate is corrected to satisfy the obstacle avoidance direction set based on the driver's operation.

といった関数で表されるリスクポテンシャルを設定することができる。
ただし、w₂はリスクポテンシャルの大きさを表すパラメータ、Δは自車と障害物が接触しないために最低限確保されなければならない距離の大きさを表す正のパラメータである。 As such risk potential, for example, when the obstacle avoidance direction is set to the left direction, a low risk evaluation value is assigned to the area of y> y _P that is the area on the left side of the obstacle with respect to the road traveling direction. Conversely, it is conceivable to set a risk potential that assigns a high risk evaluation value to a region where y <y _P. For example,

It is possible to set a risk potential expressed by a function such as
However, w ₂ is a parameter representing the magnitude of the risk potential, delta is a positive parameter representing the magnitude of the distance that must minimally be reserved for the vehicle and the obstacle is not in contact.

といった関数で表されるリスクポテンシャルを設定することができる。
図9は、上記した第2リスクポテンシャル関数の概形を例示するものである。 Conversely, if the obstacle avoidance direction is set to the right direction, a risk potential that assigns a high risk evaluation value to the area on the right side of the obstacle is required.

It is possible to set a risk potential expressed by a function such as
FIG. 9 illustrates an outline of the second risk potential function described above.

  In step S72, it is determined that the obstacle avoidance direction of the target avoidance route candidate does not match the obstacle avoidance direction based on the driver's operation, and it is determined in step S76 that the iterative calculation end condition is not satisfied. In this case, the reason for generating the second risk potential instead of the first risk potential in step S77 is as follows.

The avoidance path correction algorithm is generally configured to search for a solution whose evaluation value becomes small (improves) in the vicinity of a given solution.
Therefore, for example, if the target avoidance route candidate set as the initial solution is a route that passes the right side of the obstacle, if the solution is corrected so that the risk potential is reduced in the vicinity of the solution, the corrected route is also the obstacle There is a high possibility that the route passes through the right side of the object, and there is a concern that a desired avoidance route cannot be obtained when the passing direction set based on the driver's operation is the left side.

Therefore, when the obstacle avoidance direction of the target avoidance route candidate is different from the obstacle avoidance direction set based on the driver's operation, first, a target avoidance route candidate that matches both can be obtained. Is a predecessor,
For this reason, in step S77, the avoidance route is corrected using a simpler second risk potential that considers only the obstacle avoidance direction.

のような評価式であるのは言うまでもない。 In step S78, the target avoidance route candidate is corrected so as to pass through a region having a smaller risk potential by the same process as that performed in step S74.
Therefore, step S78 corresponds to the target avoidance route candidate correction means in the present invention.
However, in the evaluation formula used in the processing in step S78, the evaluation term of the first risk potential in the equation (19) is replaced with the evaluation term of the second risk potential.

Needless to say, the evaluation formula is as follows.

In step S79 that is selected when it is determined in step S76 that the iterative calculation end condition is satisfied, step S72 has not reached the avoidance direction coincidence even by the iterative calculation. A determination result that the target avoidance route satisfying the direction has not been obtained is confirmed, and an avoidance route calculation failure process such as issuing an alarm to the driver or canceling driving support based on the target avoidance route is executed.
Therefore, step S79 corresponds to avoidance route calculation error warning means in the present invention.

  After the execution of step S79, when it is determined in step S75 that the iterative calculation end condition is satisfied, that is, when the target avoidance path that satisfies the specified obstacle avoidance direction is obtained, the process exits the subroutine of FIG. Returning to the main routine of FIG. 3, control proceeds to step 8.

かかる転舵角指令値の時系列ベクトルがマイクロプロセッサ19内のメモリに書き込まれる。
過去の処理サイクルにおいて既に指令値ベクトルが生成・保持されていた場合であっても、過去の指令値ベクトルは破棄されて最新の指令値ベクトルに上書き、更新される。 Step S8 in FIG. 3 corresponds to the avoidance route determination means in the present invention. In this step S8, the avoidance route calculation process in step S7 ends normally, and the target avoidance route that satisfies the specified obstacle avoidance direction Is obtained, the solution obtained by the final iterative calculation is determined as the target avoidance path U ^* (t).
This target avoidance path U ^* (t) is obtained as a time series vector of the turning angle command value δ ^* as follows,

The time series vector of the turning angle command value is written in the memory in the microprocessor 19.
Even when the command value vector has already been generated and held in the past processing cycle, the past command value vector is discarded and overwritten with the latest command value vector.

となっていた場合には、制御目標値として目標転舵角θ_S ^*(t₁)が読み出された後、メモリ状態が、

となるようなシフト操作を行っておく。 In the next step S9, the command value corresponding to the current time is read from the command value vector written in the memory in the microprocessor 19.
In order to facilitate the correspondence with the current time, for example, a shift operation is performed on the memory holding the time series of command values in the microprocessor 19, and the command value stored at the start address of the memory is always read. Can be considered.
In that case, the memory state before executing step S9 is

When the target turning angle θ _S ^* (t ₁ ) is read as the control target value, the memory state is

Perform a shift operation such that

ここで、T_a ^*は補助操舵トルクの目標値を表す。補助操舵トルクと転舵アシストモータ17への電流i_Mとの間には、比例定数をK_Mとして、

という関係式が存在するため、上記(27)式から転舵アシストモータ17の電流指令値を算出し、この電流指令値をモータコントローラ18に出力して、処理が終了する。 In the next step S10, calculation / output of an auxiliary steering torque command value for driving the steering assist motor 17 is performed based on the read target turning angle so as to achieve this.
There are various methods for controlling the auxiliary steering torque. For example, the following control law is used to feed back the deviation between the steering angle command value (convertible from the target front wheel turning angle) and the actual steering angle. Can be used.

Here, T _a ^* represents the target value of the auxiliary steering torque. Between the auxiliary steering torque and the current i _M to the steering assist motor 17, the proportionality constant is K _M

Therefore, the current command value of the steering assist motor 17 is calculated from the above equation (27), this current command value is output to the motor controller 18, and the process ends.

In step S11, which is selected when it is determined in step S2 that it is not the update time of the target avoidance route or in step S3 that the obstacle 32 is not detected,
It is checked whether the time series of the target turning angle still remains in the memory of the microcomputer 19.
If the time series of the target turning angle still remains, the control proceeds to step S9 and step S10, and the support based on the time series of the remaining target turning angle is continued.
If it is determined in step S11 that the target turning angle time series does not remain, that is, if all the generated target turning angle time series have been realized, or the generation of the target avoidance route has been performed. If not, the process is terminated as it is.

A processing example of the avoidance route generation calculation described above will be schematically described below based on FIGS. 10 (a), (b), and (c).
FIG. 10 (a) shows the obstacle avoidance direction X1 and initial target avoidance route candidate X2 based on the driving operation at the start of avoidance route calculation, and a time-series initial solution of the target front wheel turning angle δ ^* .
Here, it is assumed that the avoidance route calculation is started for the first time at the time of FIG. 10 (a), and all target values are set to 0 as an initial solution, that is, the host vehicle 1 keeps the straight traveling state U (t) = Shows the status set to (0 0 .. 0).
Then, the detected obstacle 32 is detected at a position offset leftward from the front of the vehicle, the obstacle avoidance direction X1 set based on the driving operation is the left direction, and the obstacle avoidance direction of the avoidance route candidate X2 Is to the right.

Therefore, the obstacle avoidance direction X1 based on the driving operation and the obstacle avoidance direction of the avoidance route candidate X2 do not match, and the target avoidance route candidate correction calculation based on the second risk potential is shown in FIG. 10 (b). It will be executed as shown.
By this correction based on the second risk potential, the initial avoidance route candidate X2 is corrected to the avoidance route candidate X3 that avoids the obstacle 32 in the left direction as in the obstacle avoidance direction X1, as shown in FIG. 10 (b). In order to achieve this, a time series of the target front wheel turning angle δ ^* is obtained as shown in FIG.

By the way, by the above correction, the avoidance route candidate X3 has its obstacle avoidance direction coincident with the obstacle avoidance direction X1 based on the driver's operation, but this time it approaches the road boundary 31L. , Correction by the first risk potential is necessary,
The target avoidance route candidate correction calculation based on the first risk potential is executed as shown in FIG. 10 (c).
As a result of the correction based on the first risk potential, the above-mentioned candidate for avoidance route X3, as shown in FIG.10 (c), avoids the obstacle 32 in the left direction while reducing the approach to the road boundary 31L. The travel route X4 is corrected to obtain a time series of the target front wheel turning angle δ ^* for achieving this, as shown in FIG.

FIGS. 11 (a) and 11 (b) show the time series of the target front wheel turning angle and the avoidance path obtained when the obstacle avoidance direction is set to the left direction and the right direction, respectively.
If the avoidance route generation calculation is performed using only the first risk potential, the same travel route as when the obstacle avoidance direction is set to the right direction can be obtained, but the avoidance route to the left cannot be obtained. .
That is, the results shown in FIGS. 11 (a) and 11 (b) show the effectiveness of the present invention in generating a travel route in a direction-selective manner.

By the way, when setting the second risk potential in step S77 of FIG. 4, when the setting is performed using the equations (22) and (23), the second risk potential greatly deviates from the first risk potential.
In this case, the avoidance direction coincidence determination at Step S72 in FIG. 4 is switched from No to Yes, and the target avoidance path candidate correction calculation is changed from the correction based on the second risk potential at Step S77 to the first risk potential at Step S73. When switching to correction based on
As shown in FIG. 10, the time series of the target front wheel turning angle and the avoidance path change greatly, and the calculation efficiency deteriorates and the driver feels uncomfortable.

In order to avoid this problem, a function that is as close to the first risk potential as possible is set as the second risk potential, and the solution for switching the evaluation formula for avoidance path correction calculation from the second risk potential to the first risk potential is set. It is better to keep the deviation small.
Therefore, when setting the second risk potential in step S77 of FIG. 4, instead of setting the second risk potential using the above equations (22) and (23), the risk of the above equation (3) is used. The second risk potential is set by changing the potential.

Specifically, the detection position (x _P y _P ) of the obstacle 32 obtained from the sensor is changed to another virtual obstacle position, and the function applying the above equation (3) is used as the second risk potential. Set.
When setting the virtual obstacle position that constitutes the second risk potential, the avoidance direction of the target avoidance path candidate with respect to the virtual obstacle (the virtual obstacle avoidance direction of the target avoidance path candidate) is an obstacle based on the driver's operation. Set to a location that matches the object avoidance direction.

ここでty_Pは、前記(16)式の障害物回避方向条件が設定された障害物回避方向と一致する範囲内で、出来るだけy_Pに近い値を設定し、具体的には、

の式に従って当該設定を行う。 Here, since the y-axis is taken in a direction perpendicular to the traveling direction of the host vehicle 1, the obstacle avoidance direction of the target avoidance route candidate mainly depends on the y coordinate of the obstacle.
Therefore, the following risk potential in which the y coordinate of the obstacle is changed from y _P to ty _P is configured as the risk potential of the obstacle.

Here, ty _P is set to a value as close to y _P as possible within a range that matches the obstacle avoidance direction in which the obstacle avoidance direction condition of the equation (16) is set.

This setting is performed according to the following formula.

かかる第2リスクポテンシャルは、検出障害物ではなく上記仮想障害物に対して、目標回避経路候補が運転者の操作に基づく障害物回避方向と合致する方向へ向かうにつれて、リスク評価を低くするものとなる。 Note that the risk potential related to the road boundaries 31L and 31R uses the above-described equation (4) as it is, and therefore the second risk potential is configured as follows.

The second risk potential is that the risk evaluation is lowered as the target avoidance route candidate moves in a direction that matches the obstacle avoidance direction based on the driver's operation with respect to the virtual obstacle instead of the detected obstacle. Become.

As described above, step S78 after setting the second risk potential in step S77 performs the same processing as described above except that the second risk potential obtained by the above equation (32) is used.
Accordingly, also in this case, step S78 corresponds to the target avoidance route candidate correction means in the present invention.

The avoidance route calculation procedure when the second risk potential is set as described above will be described in detail below for the specific example of FIG.
FIG. 12 is a specific example under the same situation as in FIG. 10, and FIG. 12 (a) shows the driving operation reference at the start of the avoidance route calculation obtained by the processing in steps S1 to S6 in FIG. The obstacle avoidance direction X1 and the initial target avoidance route candidate X2 are shown.
That is, in FIG. 12A, the obstacle 32 is detected in step S3, the first risk potential corresponding to the detected obstacle 32 is generated in step S4, and the obstacle avoidance direction is determined based on the driver's operation in step S5. A situation is shown in which X1 is set to the left and the initial target avoidance route candidate is set as X2 in step S6.

FIG. 12B shows the process in step S7 following step S6, that is, the process in FIG.
However, FIG. 12B shows that the obstacle avoidance direction X1 based on the driving operation and the obstacle avoidance direction of the initial target avoidance route candidate X2 are different as shown in FIG. In order to proceed to step S76, the calculation result of step S71 → step S72 → step S76 → step S77 in FIG.

That is, step S77 installs the virtual obstacle 32a for generating the second risk potential based on the equation (31).
The position of the virtual obstacle 32a is such a position that the initial target avoidance route candidate X2 avoids the virtual obstacle 32a to the left in the same way as the obstacle avoidance direction X1 based on the driving operation, and this avoidance is sufficient. It is a position that is performed with a width Δ.
Therefore, step S77 corresponds to the virtual obstacle setting means in the present invention.

  FIG. 12 (c) shows the processing status in step S78 of FIG. 4, and generates a second risk potential as a second path evaluation means based on the set virtual obstacle 32a. Based on the optimization calculation, the target avoidance route candidate X2 is corrected to the target avoidance route candidate X3 for avoiding the left direction.

FIG. 12 (d) shows a step because the obstacle avoidance direction of the corrected target avoidance route candidate X3 corrected as shown in FIG. 12 (c) in step S78 of FIG. 4 has not yet reached the left direction avoidance determination criterion. The process in the case where S72 advances the control to step S76 to step S78 again is shown.
That is, step S77 installs a new virtual obstacle 32b for generating the second risk potential based on the equation (31), and regenerates the second risk potential based on the new virtual obstacle 32b. Based on the regenerated second risk potential, the corrected target avoidance route candidate X3 is corrected to the final target travel route X4 for avoiding leftward by optimization calculation.
Therefore, step S77 corresponds to the second route evaluation means in the present invention.

By repeating the generation calculation of the optimized target travel route as described above, the calculation results (the time series of the target front wheel rudder angle δ ^* and the target travel are finally shown in FIGS. 11 (a) and 11 (b)). Route).
11 (a) and 11 (b) further illustrate the time series of the target front wheel steering angle δ ^* and the target travel route with broken lines for reference when the obstacle avoidance direction is the right direction on the opposite side.

<Operational effects of the first embodiment>
According to the first embodiment described above, the obstacle avoidance direction set based on the driver's operation (step S5) and the obstacle avoidance direction of the target avoidance route candidate set for obstacle avoidance (step S6) Are different (step S72), the target avoidance route candidate is corrected so that the avoidance direction is directed to the obstacle avoidance direction set based on the driver's operation, and the avoidance route is determined (step S78).
It is possible to determine the avoidance route of the own vehicle 1 so that the obstacle avoidance direction surely matches the obstacle avoidance direction set based on the driver's operation, and therefore the own vehicle in a direction different from the driver's intention. The problem of giving the driver a sense of incongruity without evading 1 can be eliminated.

Moreover, according to the first embodiment, the positions of the virtual obstacles 32a and 32b to be set when correcting the target avoidance path candidate are the positions of the target avoidance path candidate X2 with respect to the virtual obstacle 32a as described above with reference to FIG. Since the avoidance direction matches the obstacle avoidance direction X1 set based on the driver's operation,
It is possible to reliably match the position of the virtual obstacle 32a with the avoidance direction based on the driver's operation, and it is possible to generate a target avoidance path that reflects the driver's intention to avoid the obstacle, and the above-described effects are further remarkable. Can be made.

In addition, the first risk potential lowers the risk assessment as the target avoidance route candidate for the detected obstacle moves in a direction that matches the obstacle avoidance direction based on the driver's operation.
The second risk potential is that the risk assessment becomes lower as the target avoidance route candidate moves in a direction that matches the obstacle avoidance direction based on the driver's operation for the above virtual obstacle instead of the detected obstacle. The following effects can be obtained.

Since the second risk potential is generated by shifting the position of the detected obstacle 32 to the position of the virtual obstacle 32a,
In the correction calculation of the avoidance route, when switching between the second risk potential and the first risk potential, the risk potential change margin can be suppressed to be small, and the avoidance route generation calculation can be efficiently advanced.

Further, in the first embodiment, even after the target avoidance route candidate correction is repeated a predetermined number of times, the avoidance direction for the detected obstacle of the target avoidance route candidate and the obstacle avoidance direction set based on the driver's operation are If they do not match, it is determined that generation of an appropriate avoidance route has failed and a warning is issued to the driver (steps S76 and S79).
In the case where there is no avoidance route in a suitable designated direction, it is possible to suppress a situation in which driver guidance or assistance is performed based on an inappropriate avoidance route.

<Second embodiment>
Next, a second embodiment of the present invention will be described with reference to FIGS.
Also in this embodiment, the system configuration is basically the same as described above with reference to FIGS. 1 and 2, and the calculation procedure by the microprocessor 19 is basically the same as described above with reference to FIGS.
In this embodiment, as shown in FIG. 13, when there are a plurality of obstacles such as the first obstacle 33 and the second obstacle 34 in front of the own vehicle 1, the obstacle avoidance route is generated. Is intended.
Therefore, in the following, with respect to the control program of FIGS. 3 and 4, only the portions to be changed in accordance with detection of a plurality of obstacles will be described.

In order to make the description easy to understand, the first obstacle 33 and the second obstacle 34 in front of the host vehicle 1 are traveling in the direction of the arrow on the straight road 31 as shown in FIG. The explanation will be expanded for the case where is detected.
In step S1 of FIG. 3, for each of the obstacles 33 and 34, position coordinates and size information of both are calculated.
Here, the position coordinates of the first obstacle 33 are (x _P ¹ , y _P ¹ ), the depth is σ _x ¹ , the width is σ _y ^1, and the position coordinates of the second obstacle 34 are (x _P ² , y _P ² ), the depth is σ _x ² and the width is σ _y ² .

In step S4 of FIG. 3, a risk potential is generated for each of the obstacles 33 and.
That is, L _P (x, y) shown in the equation (3) is as shown in the following equation.

In step S5 of FIG. 3, the obstacle avoidance direction is set. In the present embodiment, since there are two obstacles 33 and 34, the obstacle avoidance direction is also two for the obstacles 33 and 34, respectively. A total of four obstacle avoidance directions are possible.
In other words, the following combinations of avoidance directions for the first obstacle 33 on the front side in the running direction and the second obstacle 34 on the far side in the running direction include (left, left), (left, right), (right, left), ( Since there are four possible avoidance directions in all (right, right), the avoidance direction for generating the target avoidance route is selected from these four avoidance directions.

In step S71 in FIG. 4, it is determined whether the target avoidance route candidate avoids in the left or right direction of the obstacles 33 and 34. The above equation (16) is applied to each of the obstacles 33 and 34 to make the determination. .
The parameter Δ in equation (16) can be determined by setting a different Δ for each obstacle 33 and 34 when the widths σ _y ¹ and σ _y ² of the obstacles 33 and 34 are different.

  In the next step S72, it is checked whether or not the avoidance directions match for all obstacles 33 and 34, and control proceeds to step S73 only when the avoidance directions match for all obstacles 33 and 34. Control is advanced to step S76 unless complete coincidence of avoidance directions is obtained.

In step 77, the obstacle information whose avoidance direction does not match among the obstacles 33 and 34 is changed, and the second risk potential is constructed.
The construction of the second risk potential can be realized by applying the equation (31) for each applicable obstacle.
When both obstacles 33 and 34 are applicable, parameters representing two virtual obstacle positions ty _P ¹ and ty _P ² are generated.

In the traveling state shown in FIG. 13, the obstacle avoidance route generation result obtained by setting the virtual obstacle as described above is as shown in FIGS.
FIG. 14 shows the number of iterations on the horizontal axis for each obstacle avoidance direction, the vertical axis represents the virtual obstacle related to the first obstacle 33 in the Y-axis direction ty _P ¹ and the virtual related to the second obstacle 34. it is a plot of the Y-axis direction position ty _P ² of the obstacle.

In the setting for avoiding the first obstacle 33 in the right direction, as shown in FIGS. 14 (c) and 14 (d), as the iterative calculation progresses, the Y-axis direction position ty _{P of the} virtual obstacle related to the first obstacle 33 Y-axis direction position ty _P ² each virtual obstacle according to the ^first and second obstacle 34, the Y-axis direction detected position y in the Y-axis direction detected position y _P ¹ and the second obstacle 34 of the first obstacle 33 _P ² and the final matches.
This means that if the first obstacle 33 is set to avoid the right direction, an appropriate target avoidance route is found.
Therefore, in the driving situation as shown in FIG. 13, the first obstacle 33 is set to avoid the right direction, the virtual obstacle is set as shown in FIGS. 14 (c) and (d), and FIG. ) And (d) as shown in the target avoidance route.

However, in the setting of avoiding the first obstacle 33 in the left direction, as shown in FIGS. 14 (a) and 14 (b), the Y-axis direction positions ty _P ¹ and ty _P ² does not coincide with the corresponding detection positions y _P ¹ and y _P ² in the Y-axis direction.
In FIG. 14 (a), the Y-axis direction position ty _P ^{1 of the} virtual obstacle cannot coincide with the corresponding Y-axis direction detection position y _P ^{1 due} to the progress of the iterative calculation, and it is repeated in FIG. 14 (b). Even with the progress of the calculation, both the Y-axis direction positions ty _P ¹ and ty _P ² of the virtual obstacle cannot coincide with the corresponding Y-axis direction detection positions y _P ¹ and y _P ² .
This means that if the first obstacle 33 is set to avoid the left direction, the target avoidance route cannot be found.

Actually, when the target avoidance path is obtained based on the Y-axis direction positions ty _P ¹ and ty _P ² of the virtual obstacles as shown in FIGS. 14 (a) and 14 (b), the target avoidance paths are respectively shown in FIG. 15 (a). In both cases, the target avoidance route is a route that passes very close to the first obstacle 33 and cannot be the target route.
In such a case, even if the front wheel steering is supported based on the obtained target avoidance route, it is not always possible to avoid the first obstacle 33, so the avoidance route calculation is performed in step S79 of FIG. For example, a failure process is performed, for example, an alarm is issued to inform the driver of the fact, another obstacle avoidance direction is set, recalculation of avoidable routes is performed, and steering assist control is stopped. Then, a process such as switching to another support control that assumes contact with an obstacle is performed.

  The above is the procedure for calculating the target avoidance path when there are two obstacles 33 and 34, but when there are three or more obstacles, the target avoidance path is efficiently generated based on the same concept. Needless to say, you can.

<Operational effects of the second embodiment>
In the second embodiment, as is clear from the above, even if there are a plurality of obstacles, the calculation procedure is partially different from the first embodiment in accordance with the increase in the number of obstacles. The avoidance path can be generated basically in the same manner as described above, and the above-described operational effects of the first embodiment can be achieved as they are.

<Third embodiment>
Next, a third embodiment of the present invention will be described with reference to FIGS.
Also in this embodiment, the system configuration is basically the same as described above with reference to FIGS. 1 and 2, and the calculation procedure by the microprocessor 19 is basically the same as described above with reference to FIGS.
In this embodiment, as shown in FIG. 16, when there is an obstacle 35 that moves to cross the own vehicle traveling path from the left side to the right direction in front of the own vehicle 1, how to generate the obstacle avoidance route. It is intended.
Therefore, in the following, with respect to the control program of FIGS. 3 and 4, only portions to be changed along with detection of the moving obstacle 35 will be described.

In order to make the description easy to understand, as shown in FIG. 16, the host vehicle 1 is traveling on the straight road 31 in the direction of the arrow, and the obstacle 35 is moving on the left side in front as indicated by the arrow. The explanation will be expanded for the case where is detected.
In step S1 in FIG. 3, not only the position coordinates (x _P , y _P ) but also the moving speed v _P of the detected moving obstacle 35 are calculated.
Therefore, step S1 corresponds to the obstacle movement detecting means in the present invention.
However, in order to make the explanation easy to understand, it is assumed that the moving obstacle 35 crosses the straight road 31 in a direction perpendicular to the traveling direction of the host vehicle and the moving speed in the X-axis direction is zero.

の演算により求めるものとする。 When the first risk potential is generated in step S4 of FIG. 3, the risk potential is configured as a function that changes with the passage of time as the obstacle 35 moves.
First, the future movement locus of the obstacle 35 is estimated from the detection speed v _P of the obstacle 35.
Therefore, step S4 corresponds to the obstacle movement trajectory estimation means in the present invention.
Here, for the sake of simplicity, it is assumed that the obstacle 35 performs a constant linear movement, and the position cy _P (τ) of the obstacle 35 at the time τ (t ≦ τ ≦ t + T) is

It shall be obtained by the calculation of

従って第1リスクポテンシャルは、障害物移動軌跡に対して目標回避経路候補が、運転者の操作に基づき設定された障害物回避方向と合致する方向へ向かうにつれ、リスク評価を低くされたものとなる。 A risk potential L _P (x, y, τ) that changes according to time τ is obtained by calculation of the following equation based on the existence position cy _P (τ) that changes every moment of the moving obstacle 35.

Therefore, the first risk potential is that the risk evaluation is lowered as the target avoidance route candidate moves in the direction that matches the obstacle avoidance direction set based on the driver's operation with respect to the obstacle movement trajectory. .

を用いて障害物回避方向の判定を行う。 In consideration of avoidance direction coincidence in step S5 in FIG. 3, considering that the obstacle 35 is moving, instead of the above-described equation (16), the following equation:

Is used to determine the obstacle avoidance direction.

に基づき第2リスクポテンシャルを求める。
従って第2リスクポテンシャルは、障害物移動軌跡に対して目標回避経路候補が障害物移動方向と合致する方向へ向かうにつれてリスク評価を低くされたものとなる。 In generating the second risk potential performed in step S77 of FIG. 4, the moving speed v _P of the obstacle 35 is changed to tv _P , and the predicted obstacle position ty _P (t _P ) satisfies the equation (31). The second risk potential is generated by setting the value closest to the moving speed detection value v _P of the obstacle 35 to tv _P under the condition of performing. That is,

Based on the above, the second risk potential is obtained.
Accordingly, the second risk potential is such that the risk evaluation is lowered as the target avoidance route candidate moves in a direction that matches the obstacle movement direction with respect to the obstacle movement locus.

In the traveling state shown in FIG. 16, the obstacle avoidance route generation result obtained by setting the second risk potential as described above is as shown in FIG.
In FIG. 17 (a), the horizontal axis plots the number of iterations n, and the vertical axis plots the value tv _P (moving speed parameter) closest to the moving speed detection value v _P of the obstacle 35 described above. FIG. 17 (b) shows the initial avoidance route candidate X2 and the avoidance route when the movement speed parameter tv _P matches the movement speed detection value v _P.

When the movement speed parameter tv _P as shown in FIG. 17 (a) is not corrected, the initial avoidance route candidate X2 is a route that avoids the left side of the moving obstacle 35. Since the candidate X2 is corrected to the left side, a desired travel route cannot be obtained when the avoidance direction is set to the right direction.
However, if tv _P as shown in FIG. 17 (a) is used as the moving speed parameter, the obstacle avoidance direction of the initial target avoidance route candidate X2 can be kept to the right. Obtainable.

  The same applies to the case where the obstacle avoidance direction is set to the left direction, and even if the obstacle moves, the travel in the desired obstacle avoidance direction is performed by performing the travel route calculation using the second risk potential. The route can be calculated stably.

Also in this embodiment, when it is impossible to avoid an obstacle in the designated direction, the state where the moving speed parameter tv _P does not coincide with the detected moving speed v _P continues even if the iterative calculation is advanced.
In this case, as in the second embodiment, the steering assistance is discontinued after warning the driver that the certainty of obstacle avoidance cannot be supported.

<Operational effects of the third embodiment>
According to the third embodiment described above, the movement trajectory of the obstacle 35 is estimated based on the moving direction and the moving speed of the obstacle 35, and the first risk potential performs the risk evaluation based on the obstacle movement trajectory.
Even when the obstacle 35 moves, an appropriate obstacle avoidance route can be calculated in consideration of this movement.

In addition, the first risk potential is a risk assessment lowered as the target avoidance route candidate moves toward a direction that matches the obstacle avoidance direction set based on the driver's operation with respect to the obstacle movement trajectory.
Since the second risk potential is such that the risk evaluation is lowered as the target avoidance path candidate moves in a direction that matches the obstacle movement direction with respect to the obstacle movement locus, the following effects can be achieved.
In other words, in this case, not only the position of the detected obstacle 35 but also its speed is shifted to generate the risk potential, and even when the obstacle 35 is moving, the obstacle avoidance route correction calculation is performed. The risk potential change when switching from the 2 risk potential to the first risk potential can be suppressed to a small level, and the avoidance path generation calculation can be performed efficiently.

<Other examples>
In any of the above-described embodiments, the generation of the second risk potential is repeatedly performed every calculation cycle in step S77 of FIG.
Instead, it goes without saying that the second risk potential is generated at the beginning of FIG. 4 and the second risk potential obtained here is read in step S77 of FIG.

1 Vehicle volume (moving body)
2 Body
3L, 3R left and right front wheels
4L, 4R left and right rear wheels
5 Steering wheel
6 Steering column
7 Steering gear box
11L, 11R camera (obstacle detection means)
12L, 12R Wheel speed sensor
13 Yaw rate sensor
14 Accelerometer
15 Steering angle sensor
16 Steering torque sensor
17 Steering assist motor
18 Motor controller
19 Microprocessor
21 Sensor signal processor
22 Obstacle avoidance direction setting part (obstacle avoidance direction setting means)
23 Target avoidance route candidate setting section (target avoidance route candidate setting means)
24 Target avoidance route candidate memory
25 First risk potential generator (first path evaluation means)
26 Second risk potential generator (second path evaluation means)
27 Avoidance route correction unit (target avoidance route candidate correction means)
28 Command value output processing section
29 Avoidance route calculator
31 road
31L, 31R road boundary
32-35 obstacle

Claims (6)

Obstacle detection means for detecting the current position of the obstacle in front of the moving body driven by the driver;
Obstacle avoiding direction setting means for setting an avoiding direction necessary for the moving body to avoid the obstacle position detected by the obstacle detecting means in a direction crossing the traveling direction of the moving body based on the operation of the driver; ,
Target avoidance path candidate setting means for setting a target avoidance path candidate necessary for avoiding the obstacle position detected by the obstacle detection means in a direction crossing the moving body traveling direction;
Based on obstacle position information from the obstacle detection means, a first route evaluation means for calculating the degree of approach between the target avoidance route candidate and the detected obstacle;
Avoidance direction difference determination for determining a difference between the avoidance direction set by the obstacle avoidance direction setting unit and the avoidance direction of the target avoidance route candidate set by the target avoidance route candidate setting unit with respect to the detected obstacle Means,
Based on the determination result of the avoidance direction difference degree determination means, a virtual obstacle is set at a point where the target avoidance route candidate having a large difference degree is moved from the obstacle position in a direction crossing the moving body traveling direction. Obstacle setting means,
A second route evaluation means for calculating the degree of approach between the virtual obstacle and the target avoidance route candidate;
Target avoidance route candidate correction means for correcting the target avoidance route candidate based on the evaluation value by the first route evaluation means or the second route evaluation means;
Avoidance route determination means for determining an avoidance route based on the corrected target avoidance route candidate and the avoidance direction set by the obstacle avoidance direction setting means;
A moving body travel route generating apparatus comprising: In the moving body travel route generation device according to claim 1,
The virtual obstacle setting means is at a position where the avoidance direction of the target avoidance route candidate with respect to the virtual obstacle coincides with the obstacle avoidance direction based on the driver's operation set by the obstacle avoidance direction setting means. A moving body travel route generation device for setting a virtual obstacle detection point. In the moving body travel route generating device according to claim 1 or 2,
The first route evaluation means sets the target avoidance route candidate set by the target avoidance route candidate setting means for the obstacle position detected by the obstacle detection means by the obstacle avoidance direction setting means. The risk assessment is lowered as it goes in the direction that matches the obstacle avoidance direction
The second route evaluation unit is configured such that the target avoidance route candidate set by the target avoidance route candidate setting unit is the obstacle avoidance direction setting unit for the virtual obstacle set by the virtual obstacle setting unit. A moving body travel route generation device characterized by lowering a risk evaluation toward a direction that matches a set obstacle avoidance direction. In the moving body travel route generating device according to any one of claims 1 to 3,
Obstacle movement detecting means for detecting the moving direction and moving speed of the obstacle detected by the obstacle detecting means;
Obstacle moving trajectory estimation means for estimating the moving trajectory of the obstacle according to the detected moving direction and moving speed of the obstacle,
The mobile body travel route generation device, wherein the first route evaluation means performs risk evaluation according to a movement locus of the obstacle. In the mobile body travel route generation device according to claim 4,
The first route evaluation means sets the obstacle avoidance direction setting for the obstacle avoidance trajectory calculated by the obstacle movement trajectory estimation means, the target avoidance path candidate set by the target avoidance path candidate setting means. The risk assessment is lowered as it goes in the direction that matches the obstacle avoidance direction set by the means,
The second route evaluation unit is configured such that a target avoidance route candidate set by the target avoidance route candidate setting unit with respect to the obstacle movement locus estimated by the obstacle movement locus estimation unit is the obstacle movement detection unit. A moving body travel route generation device characterized in that the risk evaluation is lowered as it goes in a direction that matches the obstacle movement direction detected in step (b). In the mobile body travel route generation device according to any one of claims 1 to 5,
Even after the target avoidance path candidate correction means repeats the correction of the target avoidance path candidate a predetermined number of times, the avoidance direction for the detected obstacle of the target avoidance path candidate and the avoidance direction set by the obstacle avoidance direction setting means A mobile travel route generation device comprising an avoidance route calculation error warning means for determining that the generation of an appropriate avoidance route has failed if the two do not match, and issuing a warning to the driver.

Images