JP2019178971A - Device, method, and program for generating environmental map - Google Patents

Abstract

To provide an improved device, method, and program for generating an environmental map which can reduce the probability that a sensor device fails to detect or detects wrong, by a sensor fusion method.SOLUTION: The device for generating an environmental map includes: a first probability calculation circuit for calculating a first occupation probability for a first position according to the measured value of an object in the first position in a first region by a first sensor device and the feature of the first position identified by the second sensor device; and a map generation circuit for generating a map showing a second occupation probability that an object exists in each position of a second region on the basis of the first occupation probability.SELECTED DRAWING: Figure 1

Description

  The present disclosure relates to an environment map generation device, an environment map generation method, and an environment map generation program.

  In recent years, with the development of driving assistance technology such as automatic driving technology for automobiles, demands for various in-vehicle sensor devices such as an in-vehicle camera, an in-vehicle sonar, and an in-vehicle radar for grasping the situation where the automobile is placed are increasing. There are various types of in-vehicle sensor devices, and the operation principle and operation characteristics differ depending on the type. Therefore, it is difficult to cope with various situations where a vehicle is placed with a single in-vehicle sensor device. Therefore, in order to deal with various situations in which the vehicle is placed, a sensor fusion method has been proposed that fuses the measurement results according to the circumstances that the plurality of in-vehicle sensor devices are good at.

JP 2012-048642 A

  One aspect of the present disclosure contributes to the provision of an improved environment map generation device, an environment map generation method, and an environment map generation program that reduce the probability of detection omission or false detection of a sensor device in a sensor fusion technique.

  An environment map generation device according to an aspect of the present disclosure includes a measurement value of a first position in a first region by a first sensor device and a feature of the first position identified by a second sensor device. And a first probability calculation circuit for calculating a first occupation probability that an object at the first position exists, and for each position in the second region based on the first occupation probability And a map generation circuit that generates a map indicating the second occupation probability that the object is present.

  The environmental map generation method according to an aspect of the present disclosure is based on the measurement value of the first position in the first region by the first sensor device and the feature of the first position identified by the second sensor device. In response, a first occupancy probability that an object at the first position exists is calculated, and based on the first occupancy probability, a second object exists for each position in the second region. The structure which produces | generates the map which shows an occupation probability is taken.

  An environment map generation program according to an aspect of the present disclosure is a computer in which a measured value of a first position in a first region by a first sensor device and the first position identified by a second sensor device The first occupancy probability that the object at the first position exists is calculated according to the characteristics of the object, and the object exists at each position in the second region based on the first occupancy probability. A configuration is adopted in which a map showing the second occupation probability is generated.

  Note that these comprehensive or specific aspects may be realized by a system, method, integrated circuit, computer program, or recording medium. Any of the system, apparatus, method, integrated circuit, computer program, and recording medium may be used. It may be realized by various combinations.

  According to one aspect of the present disclosure, in the sensor fusion technique, it contributes to the provision of an improved environment map generation device, an environment map generation method, and an environment map generation program that reduce the probability of detection failure or false detection of the sensor device. .

  Further advantages and effects in one aspect of the present disclosure will become apparent from the specification and drawings. Such advantages and / or effects are each provided by the features described in some embodiments and the specification and drawings, but not necessarily all in order to obtain one or more identical features. There is no need.

It is a block diagram of an environmental map generation system concerning this indication. It is a flowchart which shows operation | movement of the environmental map production | generation apparatus which concerns on this indication. It is a figure which shows an example of the hardware constitutions of a computer. It is explanatory drawing of the occupation probability of the area estimated based on the sonar observation.

[Background to the Disclosure]
First, an environmental map used in the sensor fusion method will be described.

<Environmental map>
The OGM (Occupancy Grid Map) is a map of the surrounding area observed by the host vehicle divided into grids and divided into grids (hereinafter simply referred to as "compartments"). It is a map obtained by mapping the probability that the section is occupied by an object to be avoided by the vehicle.

For event X, the position of the vehicle at discrete times _{1 to t} is x _{1: t} (= (x ₁ , x ₂ ,..., X _t )), and the sensor measurement values at discrete times 1 to _t −1 are Z _{When 1: t−1} (= (Z ₁ , Z ₂ ,..., Z _t−1 )), it is assumed that the sensor measurement value at the discrete time t is Z _t . Event A is defined by an object to be avoided when the position of the vehicle at discrete times _{1 to t} is x _{1: t} and the sensor measurement value at discrete times _{1 to t-1} is Z _{1: t-1.} It is assumed that m is occupied. In the present disclosure, the sensor measurement value may be either a scalar value or a vector value.

で表されるベイズの定理を用いると、次の式（２）：

が得られる。ここで、回避すべき物体によって区画ｍが占有される事象を、同じ記号ｍ（事象ｍ）で表す。 The following formula (1):

Using Bayes' theorem expressed by the following equation (2):

Is obtained. Here, an event in which the section m is occupied by an object to be avoided is represented by the same symbol m (event m).

が得られる。上述の式（２）に式（３）を代入すると、次の式（４）：

が得られる。事象ｍの余事象についても、同様にして次の式（５）：

が得られる。上述の式（４）及び式（５）より、次の式（６）：

が得られる。 Assuming that the random variable Z _t is independent of the random variables x _{1: t−1} and Z _{1: t−1} , the following equation (3):

Is obtained. Substituting equation (3) into equation (2) above, the following equation (4):

Is obtained. The following equation (5) is similarly applied to the remaining event of event m:

Is obtained. From the above equations (4) and (5), the following equation (6):

Is obtained.

で定義する。上述の式（７）をＰ（ｘ）について解くことにより、確率Ｐ（ｘ）とオッズＯｄｄｓ（ｘ）とは、相互に変換可能であることがわかる。上述の式（６）及び式（７）より、次の式（８）：

が得られる。ここで、センサ測定値などの情報がない時、区画ｍの事前占有確率は０．５とし、即ち、Ｐ（ｍ）＝０．５とおくと、ｌｏｇＯｄｄｓ（ｍ）＝０となる。これを上述の式（８）に代入すると、次の式（９）：

が得られる。 Here, for the event x, odds Odds (x) is expressed by the following equation (7):

Define in. It is understood that the probability P (x) and the odds Odds (x) can be converted into each other by solving the above equation (7) for P (x). From the above equations (6) and (7), the following equation (8):

Is obtained. Here, when there is no information such as sensor measurement values, the prior occupation probability of the section m is 0.5, that is, if P (m) = 0.5, logOdds (m) = 0. Substituting this into equation (8) above, the following equation (9):

Is obtained.

Hereinafter, the probability that the section m is occupied by the object to be avoided at the discrete time t is referred to as the occupation probability of the section m. The above equation (9), occupancy probability _P compartment m at discrete time _{t (m | x 1: t} , Z 1: t) is occupancy probability P compartment m at discrete time _t-1 (m | _x 1 _{: t-1, Z 1:} t-1) = P (m | x 1: t, Z 1: t-1) and the discrete time _x-position and the sensor measurements of the vehicle at t, respectively _t and _{Z t} It is shown that it is obtained from the occupation probability P (m | x _t , Z _t ) of the section m in the case of. That is, the environment map at the discrete time t is the environment map at the discrete time t−1 and the occupation probability P of the section m under the condition that the position of the vehicle and the sensor measurement values at the discrete time t are x _t and Z _t , respectively. (M | x _t , Z _t ) In other words, the environment map at the discrete time t is obtained from the occupation probabilities P (m | x _i , Z _i ) and i = 1 to t of the section m based on the sensor measurement values from the discrete times 1 to t.

<Occupancy probability based on sensor measurement value>
For example, consider the case where the sensor measurement Z _t is the sonar measurement Z _t ^uss . The sonar emits a sound wave signal toward the observation area, detects a reflected wave, and outputs a measured value Z _t ^uss . The measurement value Z _t ^uss includes the position information of the object in the observation region and the intensity information of the reflected wave. Based on the measured value Z _t ^uss , the occupation probability P (m | x _t , Z _t ^uss ) of the section m is estimated. Occupancy probability _P compartment ^{_{m (m | x t, Z}} t uss) is estimated based on a value obtained in advance through experiments. The estimated occupation probability P (m | x _t , Z _t ^uss ) of the section m is usually a value reflecting the intensity of the detected reflected wave. The greater the intensity of the reflected wave, the greater the occupation probability P (m | x _t , Z _t ^uss ) of the estimated section m.

FIG. 4 is an explanatory diagram of the occupation probability P (m | x _t , Z _t ^uss ) of the section m estimated based on the observed value Z _t ^uss of the sonar S.

ここで、ａ，ｂは定数の係数であり、Ｚ_ｔ ^ｕｓｓ［１］はソナーＳが測定値した反射波強度成分であって、測定された距離で正規化された反射波強度成分である。 Usually, the measured value Z _t ^uss of the reflected wave by the sonar S, include detection error of the azimuth and distance. In the explanatory view shown in FIG. 4, the sound wave signal emitted by the sonar S is reflected by the object T, for example. The sonar S detects a reflected wave from the object T. The occupancy probability P (m | x _t , Z _t ^uss ) of the section m of the grid map G reflects the direction and distance detection errors. For example, when the section m is included in the sector region R1 determined by the position information component of the sonar measurement value and the error range, the estimated occupation probability P (m | x _t , Z _t ^uss ) of the section m is P (However, 0.5 <P ≦ 1). Further, when the section m is included in the sector area R2, the estimated occupation probability P (m | x _t , Z _t ^uss ) of the section m is 1-P. When the section m is included in another region (that is, a section outside the sonar measurement range), the estimated occupation probability P (m | x _t , Z _t ^uss ) of the section m is 0.5. Note that the value of P takes a larger value as the reflection intensity is higher. For example, the value of P is calculated by the following formula.

Here, a and b are constant coefficients, and Z _t ^uss [1] is a reflected wave intensity component measured by the sonar S, and is a reflected wave intensity component normalized by the measured distance.

Consider a case where the sonar S is used to detect an object T that the vehicle should avoid. Examples of the object T to be avoided by the vehicle include a pedestrian on a road and a shopping cart left in a parking lot. However, the skin and clothes of pedestrians are difficult to reflect sound wave signals, and the intensity of reflected waves detected by the sonar S is also weakened. Further, a shopping cart is often composed of a net and has a small area for reflecting a sound wave signal, so that the intensity of the reflected wave detected by the sonar S becomes weak. Therefore, the occupation probability P (m | Z _t ^uss ) of the estimated section m becomes small, and the sonar S may cause a detection failure of the pedestrian and the shopping cart.

On the other hand, as the object T that the vehicle should not avoid, for example, it is snowfall. Even if there is snow in the direction of travel of the vehicle, usually the vehicle should not avoid snow. However, when the sonar S detects a reflected wave having a high intensity due to, for example, snow adhering to the sonar S, the occupation probability P (m | Z _t ^uss ) of the estimated section m increases, and the sonar S There is a possibility of false detection that the object should avoid snowfall. On the other hand, due to the nature of snow to absorb sound, the surface of an object covered with snow may have a weaker reflected wave than when no snow is accumulated, and sonar S may cause detection omission of an object to be avoided. There is sex.

Therefore, in order to avoid omissions and false detections, the sensor fusion technique is used to identify the type of object to be measured by the sonar using a sensor different from the sonar, and the estimated section using the identification result. It is conceivable to correct the occupation probability P (m | Z _t ^uss ) of m.

  The driving environment recognition device described in Patent Document 1 creates an occupancy grid map (environment map) for each sensor such as a laser radar, a communication device, and a map database, and calculates an occupancy probability of an obstacle that causes the vehicle to travel. (Paragraph [0028], FIG. 2). However, even if an occupancy grid map is generated for each sensor and fusion calculation is performed, it is difficult to calculate an appropriate occupancy probability for each type of detected object.

  Accordingly, the present disclosure has been reached in order to calculate an appropriate occupation probability for each type of detected object.

[Embodiment 1]
Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings. In the embodiment, the same components are denoted by the same reference numerals, and the description thereof will be omitted because it is duplicated.

  The embodiment described below is an example, and the present disclosure is not limited to the following embodiment.

  FIG. 1 is a block diagram of an environment map generation system 1 according to the present disclosure.

  The environment map generation system 1 includes a vehicle positioning sensor 110, a sonar 120, a radar 130, an imaging device 140, an object identification device (second sensor device) 150, and an environment map generation device 200.

  The vehicle positioning sensor 110 measures the position and orientation of the host vehicle and outputs positioning information. The vehicle positioning sensor 110 is, for example, a GPS terminal device, a Wi-Fi terminal device, or an ETC 2.0 vehicle-mounted device.

The sonar 120 emits a sound wave toward the observation area and receives a reflected wave reflected by the object. Next, the sonar 120 detects the position, size, or orientation of the object based on the received reflected wave, and outputs a measurement value Z _t ^uss for each position in the observation region.

The radar 130 emits a radar wave toward the observation area and receives the reflected wave reflected by the object. Next, the radar 130 detects the position, size, or direction of the object based on the received reflected wave, and outputs a measurement value Z _t ^radar for each position in the observation region. The radar 130 is, for example, a pulse compression radar or a continuous wave radar that uses millimeter waves or microwaves.

  The imaging device 140 images the imaging area. The imaging region is outside the vehicle, for example, the front of the vehicle, the side of the vehicle, or the rear of the vehicle. The imaging device 140 is, for example, a vehicle-mounted monocular camera or a vehicle-mounted stereo camera.

  The sonar 120, the radar 130, and the imaging device 140 function as a sensor device (first sensor device) that measures each position (first position) in the observation region or the imaging region (first region). .

  The object identification device 150 identifies the type of the object from the image captured by the imaging device 140 and outputs the identification result. The types of objects are, for example, pedestrians, utility poles, signboards, shopping carts, snowfall, and manholes. For example, when the imaging device 140 is an in-vehicle stereo camera, the object identification device 150 may identify the position of the object in addition to the type of the object. The object identification device 150 functions as a sensor device (second sensor device) that identifies the feature of the first position included in the image.

  The environment map generation device 200 includes a vehicle position estimation circuit 210, a first probability calculation circuit 220, a second probability calculation circuit 230, a third probability calculation circuit 240, and an environment map grid occupation probability calculation circuit 250. And an environmental map output circuit 260.

The vehicle position estimation circuit 210 estimates the position _xt and the direction of the host vehicle. For example, the vehicle position estimation circuit 210 inputs positioning information from the vehicle positioning sensor 110, and estimates the position _xt and direction of the host vehicle based on the positioning information.

The first probability calculation circuit 220 receives the measurement value Z _t ^uss from the sonar 120 and the identification result Z _t ^cam from the object identification device 150. Then, the first probability calculation circuit 220, the measured value _Z ^{t uss} and identification results based on _Z ^{t cam,} calculating the occupied odds _{O 1} compartment m. Details of the contents of the calculation will be described later with reference to FIG.

The second probability calculation circuit 230 receives the measurement value Z _t ^radar from the radar 130 and the identification result Z _t ^cam from the object identification device 150. Next, the second probability calculation circuit 230 calculates the occupation odds O ₂ of the partition m based on the measurement value Z _t ^radar and the identification result Z _t ^cam . Details of the contents of the calculation will be described later with reference to FIG.

The third probability calculation circuit 240 receives the captured image Z _t ^{cam ′} from the imaging device 140 and the identification result Z _t ^cam from the object identification device 150. Next, the third probability calculation circuit 240 calculates the occupation odds O ₃ of the section m based on the captured image Z _t ^{cam ′} and the identification result Z _t ^cam . Details of the contents of the calculation will be described later with reference to FIG.

The environment map grid occupation probability calculation circuit 250 converts the division m of the occupation odds O _{1 to} O ₃ from _the sensor coordinate system to the coordinate system of the own vehicle, and subsequently from the coordinate system of the own vehicle to the environment map coordinate system. Then, the environment map grid occupancy probability calculation circuit 250, based on the occupancy odds _O 1 ~ O _3, fused occupancy probability _P for each partition m grid map _{(m | x 1: t,} Z 1: t) a Update. Details of the contents of the calculation will be described later with reference to FIG.

Environment map output circuit 260, updated fused occupancy probability _{P (m | x 1: t} , Z 1: t) based on each section of the grid mapped to (position) m, compartment m is occupied Update and output an environmental map showing the occupancy probability. The environment map grid occupancy probability calculation circuit 250 and the environment map output circuit 260 have the occupancy probabilities calculated by at least one of the first probability calculation circuit 220, the second probability calculation circuit 230, and the third probability calculation circuit 240. Based on the map (environment map) indicating the fusion occupation probability (second occupation probability) P (m | x _{1: t} , Z _{1: t} ) for each section m in the grid map (second region). ) To generate a map generation circuit.

  FIG. 2 is a flowchart illustrating the operation of the environment map generation device 200 according to the present disclosure.

For example, as an initial value, P (m | x ₀ , Z ₀ ) = P (m | x ₀ ) = 0.5 is set for all sections m of the grid map. Next, by executing steps S110 to S170 of FIG. 2 at a predetermined time interval, the fusion occupation probability P (m | x _{1: t} , Z _{1: t} ) is updated, and the environment map is also updated.

In step S110, the vehicle position estimation circuit 210 estimates the position and orientation of the host vehicle. For example, the vehicle position estimation circuit 210 inputs positioning information from the vehicle positioning sensor 110, and estimates the position _xt and direction of the host vehicle based on the positioning information.

In step S120, the first probability calculation circuit 220, the second probability calculation circuit 230, and the third probability calculation circuit 240 receive the sensor measurement values. For example, the first probability calculation circuit 220 receives the measurement value Z _t ^uss from the sonar 120. The second probability calculation circuit 230 receives the measurement value Z _t ^radar from the radar 130. The third probability calculation circuit 240 receives the captured image Z _t ^{cam ′} from the imaging device 140.

In step S <b> 130, the first probability calculation circuit 220, the second probability calculation circuit 230, and the third probability calculation circuit 240 receive the object identification result Z _t ^cam included in the captured image from the object identification device 150. The identification result Z _t ^cam includes, for example, the type of the object in the section m.

In step S140, the first probability calculation circuit 220, the second probability calculation circuit 230, and the third probability calculation circuit 240, for example, occupy odds O _{1 to} O _{3 in} a format based on the measurement value and the identification result. Calculate the occupation probability.

を用いて占有オッズＯ_１の形式で占有確率Ｐ（ｍ｜ｘ_ｔ，Ｚ_ｔ ^ｕｓｓ，Ｚ_ｔ ^ｃａｍ）（第１の占有確率）を計算する。占有オッズＯ_１は、ソナー１２０の測定値Ｚ_ｔ ^ｕｓｓに加えて、物体識別装置１５０による識別結果Ｚ_ｔ ^ｃａｍに依存する。 For example, the first probability calculation circuit 220 has the following equation (10):

Is used to calculate the occupation probability P (m | x _t , Z _t ^uss , Z _t ^cam ) (first occupation probability) in the form of occupation odds O ₁ . Occupied odds O ₁ depend on the identification result Z _t ^cam by the object identification device 150 in addition to the measured value Z _t ^uss of the sonar 120.

を用いて占有オッズＯ_１を計算する。ここで、ｆ_１は、次の式（１２）：

で定められる関数であり、αは、ある正の数である。また、Ｒ_αは、区画ｍに歩行者が存在する場合に、ｌｏｇＯｄｄｓ（ｍ｜ｘ_ｔ，Ｚ_ｔ ^ｕｓｓ）がとる値の範囲である。即ち、上述の式（１１）を用いて計算された占有オッズＯ_１は、区画ｍに歩行者が存在する場合に正方向に補正され、ソナーによる歩行者の検出が容易になるように、ソナーが検出した弱い反射が補正される。 In one example, if the object identification device 150 when there is a pedestrian in a compartment m has identified, for example, an output Z _t ^cam of object identification apparatus 150, if represents the probability that the pedestrian exists, the first probability computing The circuit 220 has the following equation (11):

Is used to calculate the occupied odds O ₁ . Where f ₁ is the following equation (12):

Where α is a positive number. R _α is a range of values taken by logOdds (m | x _t , Z _t ^uss ) when a pedestrian is present in the section m. That is, the occupation odds O ₁ calculated using the above equation (11) is corrected in the positive direction when there is a pedestrian in the section m, so that the sonar can be easily detected by the sonar. The weak reflection detected by is corrected.

を用いて占有オッズＯ_１を計算する。ここで、ｆ_２は、次の式（１４）：

で定められる関数であり、βは、ある正の数である。また、Ｒ_βは、区画ｍに降雪が存在する場合に、ｌｏｇＯｄｄｓ（ｍ｜ｘ_ｔ，Ｚ_ｔ ^ｕｓｓ）がとる値の範囲である。即ち、上述の式（１４）を用いて計算された占有オッズＯ_１は、区画ｍに降雪が存在する場合に負方向に補正され、ソナーが検出した降雪による反射の影響が除去されるように補正される。 In another example, when the object identification device 150 identifies that there is snowfall in the section m, if the output Z _t ^cam of the object identification device 150 represents the probability of snowfall, the first probability calculation The circuit 220 has the following equation (13):

Is used to calculate the occupied odds O ₁ . Here, _{f 2,} the following equation (14):

Where β is a certain positive number. Further, the R _beta, when there is snow in the compartment m, logOdds | in the range of ^{_{(m x t, Z t uss}} ) takes values. That is, the occupancy odds O ₁ calculated using the above equation (14) is corrected in the negative direction when there is snow in the section m, so that the influence of the reflection due to snow detected by the sonar is removed. It is corrected.

を用いて占有オッズＯ_２の形式で占有確率Ｐ（ｍ｜ｘ_ｔ，Ｚ_ｔ ^{ｒａｄａｒ}，Ｚ_ｔ ^ｃａｍ）を計算する。占有オッズＯ_２は、レーダ１３０の測定値Ｚ_ｔ ^{ｒａｄａｒ}に加えて、物体識別装置１５０による識別結果Ｚ_ｔ ^ｃａｍに依存する。 For example, the second probability calculation circuit 230 has the following formula (15):

Is used to calculate the occupation probability P (m | x _t , Z _t ^radar , Z _t ^cam ) in the form of occupation odds O ₂ . Occupancy odds O ₂ depends on the identification result Z _t ^cam by the object identification device 150 in addition to the measurement value Z _t ^radar of the radar 130.

を用いて占有オッズＯ_２を計算する。ここで、ｆ_３は、次の式（１７）：

で定められる関数であり、γは、ある正の数である。また、Ｒ_γは、区画ｍに買い物用のカートが存在する場合に、ｌｏｇＯｄｄｓ（ｍ｜ｘ_ｔ，Ｚ_ｔ ^{ｒａｄａｒ}）がとる値の範囲である。即ち、上述の式（１６）を用いて計算された占有オッズＯ_２は、区画ｍに買い物用のカートが存在する場合に正方向に補正され、レーダによる買い物用のカートの検出が容易になるように、レーダが検出した弱い反射が補正される。 In one example, if the object identification device 150 identifies that there is a shopping cart in the section m, the output Z _t ^cam of the object identification device 150 represents the probability that the shopping cart exists, The probability calculation circuit 230 has the following formula (16):

Is used to calculate the occupancy odds O ₂ . Here, _{f 3,} the following equation (17):

Γ is a certain positive number. R _γ is a range of values taken by logOdds (m | x _t , Z _t ^radar ) when a shopping cart exists in the block m. That is, the occupation odds O ₂ calculated using the above equation (16) is corrected in the forward direction when a shopping cart exists in the section m, and the shopping cart can be easily detected by the radar. Thus, the weak reflection detected by the radar is corrected.

を用いて占有オッズＯ_２を計算する。ここで、ここで、ｆ_４は、次の式（１９）：

で定められる関数であり、δは、ある正の数である。また、Ｒ_δは、区画ｍにマンホールが存在する場合にｌｏｇＯｄｄｓ（ｍ｜ｘ_ｔ，Ｚ_ｔ ^{ｒａｄａｒ}）がとる値の範囲である。即ち、上述の式（１８）を用いて計算された占有オッズＯ_２は、区画ｍにマンホールが存在する場合に負方向に補正され、レーダが検出したマンホールによる強い反射波の占有確率への影響が除去されるように補正される。 In another example, when the object identification device 150 identifies that a manhole exists in the section m, the output Z _t ^cam of the object identification device 150 represents the probability that a manhole exists in the section m. The probability calculation circuit 230 of the following equation (18):

Calculating the occupied odds O ₂ used. Where f ₄ is the following equation (19):

Where δ is a certain positive number. R _δ is a range of values taken by logOdds (m | x _t , Z _t ^radar ) when a manhole exists in the partition m. That is, the occupation odds O ₂ calculated using the above equation (18) is corrected in the negative direction when a manhole exists in the section m, and the influence on the occupation probability of a strong reflected wave by the manhole detected by the radar. Is corrected to be removed.

を用いて占有オッズＯ_３の形式でＰ（ｍ｜ｘ_ｔ，Ｚ_ｔ ^ｃａｍ’，Ｚ_ｔ ^ｃａｍ）を計算する。占有オッズＯ_３は、撮像装置１４０による撮像画像Ｚ_ｔ ^ｃａｍ’に加えて、物体識別装置１５０による識別結果Ｚ_ｔ ^ｃａｍに依存する。 For example, the third probability calculation circuit 230 has the following formula (20):

Is used to calculate P (m | x _t , Z _t ^{cam ′} , Z _t ^cam ) in the form of occupation odds O ₃ . The occupation odds O ₃ depends on the identification result Z _t ^cam by the object identification device 150 in addition to the captured image Z _t ^{cam ′} by the imaging device 140.

In step S150, the environment map grid occupation probability calculation circuit 250 converts the position measured by each sensor device from the sensor coordinate system to the coordinate system of the host vehicle, and subsequently from the coordinate system of the host vehicle to the map coordinates. For example, the environment map grid occupation probability calculation circuit 250 uses the installation information (position and orientation in the vehicle) of each sensor device as necessary in the sensor coordinate system of the section m of the occupation odds O _{1 to} O ₃ . The coordinates are converted into the coordinate system of the host vehicle. Next, the environment map grid occupancy probability calculation circuit 250 converts the coordinates in the coordinate system of the host vehicle of the section m into map coordinates based on the position _xt and the direction of the host vehicle that change according to the discrete time t.

を用いて、環境地図（グリッドマップ）の区画ｍの融合占有確率（第２の占有確率）Ｐ（ｍ｜ｘ_１：ｔ，Ｚ_１：ｔ）を更新する。ここで、ｌｏｇＯｄｄｓ（ｍ｜ｘ_ｔ，Ｚ_ｔ）＝ｌｏｇＯ_１＋ｌｏｇＯ_２＋ｌｏｇＯ_３である。 In step S160, the environment map grid occupation probability calculation circuit 250 calculates the following equation (21):

Using fusion occupancy probability compartment m environment map (grid map) (second occupancy _{probability) P (m | x 1:} t, Z 1: t) Update. Here, logOdds (m | x _t , Z _t ) = logO ₁ + logO ₂ + logO ₃ .

In step S170, the environmental map output circuit 260 outputs an environmental map in which the fusion occupation probability P (m | x _{1: t} , Z _{1: t} ) is updated for each section m. The information of each section of the output environmental map may be expressed as an occupancy probability converted from log odds, or the occupancy probability is rounded up by a threshold value, and binary indicating whether or not an object exists in each section, for example, section m 1 is output if the occupancy probability is greater than or equal to the threshold, and 0 is output if it is less than or equal to the threshold. Steps S110 to S170 are repeated at predetermined time intervals.

  FIG. 3 is a diagram illustrating an example of a hardware configuration of a computer.

  The function of each part in each embodiment and each modification described above is realized by a program executed by the computer 2100.

  As shown in FIG. 3, the computer 2100 includes the above-described various sensor devices, an input device 2101 such as an input button and a touch pad, an output device 2102 such as a display and a speaker, a CPU (Central Processing Unit) 2103, a ROM (Read Only Memory). ) 2104 and RAM (Random Access Memory) 2105. The computer 2100 includes a hard disk device, a storage device 2106 such as an SSD (Solid State Drive), a reading device 2107 that reads information such as a DVD-ROM (Digital Versatile Disk Read Only Memory) and a USB (Universal Serial Bus), and a network. A transmission / reception device 2108 that performs communication via the network is provided. Each unit described above is connected by a bus 2109. Various sensor devices may be connected to the computer 2100 through the USB.

  Then, the reading device 2107 reads the program from a recording medium on which a program for realizing the functions of the above-described units is recorded, and stores the program in the storage device 2106. Alternatively, the transmission / reception device 2108 communicates with the server device connected to the network, and causes the storage device 2106 to store a program for realizing the function of each unit downloaded from the server device.

  Then, the CPU 2103 copies the program stored in the storage device 2106 to the RAM 2105, and sequentially reads out and executes the instructions included in the program from the RAM 2105, thereby realizing the functions of the above-described units. When executing the program, the RAM 2105 or the storage device 2106 stores information obtained by the various processes described in each embodiment, and appropriately uses the information.

  According to the first embodiment of the present disclosure, the occupation probability calculated based on the measurement value of the other sensor device is corrected according to the type of the object measured by the sensor device. Therefore, for example, it is possible to adjust the occupation probability based on the measurement value of the other sensor device by focusing on the object that the other sensor device is not good at measurement, resulting from the deterioration of the accuracy of the measurement value of the other sensor device. Therefore, it is possible to reduce detection omission of an object to be avoided or false detection of an object that should not be avoided.

[Other embodiments]
In the first embodiment, a vehicle positioning sensor 110, a sonar 120, a radar 130, an imaging device 140, and an object identification device 150 are used as sensor devices. In addition to or in place of these, embodiments using other sensor devices such as lidar, illuminance sensor, vehicle speed sensor, and gyroscope are also conceivable.

In the first embodiment, as shown in Expression (10), Expression (15), and Expression (20), the value of the occupied odds O ₁ becomes the occupied odds in the measured value Z _t ^uss and the identification result Z _t ^cam. The value of O ₂ depends on the measured value Z _t ^radar and the identification result Z _t ^cam , and the value of the occupied odds O ₃ depends on the captured image Z _t ^{cam ′} and the identification result Z _t ^cam . In implementation, it may be corrected by using other sensor measurement information in addition. For example, at night, the image of the imaging device 140 becomes unclear compared to the daytime, and the identification accuracy of the object identification device 150 also deteriorates. Thus, for example, if the illuminance sensor detects ambient light intensity and the ambient light intensity is less than a certain threshold, the value of the occupancy odds O ₃ depends on the measured values Z _t ^uss and Z _t ^radar. Forms are also conceivable. Furthermore, when a highly reliable sensor device is known in the situation where the vehicle is placed, the occupation odds based on the measured value of the sensor device with low reliability are corrected based on the measured value of the sensor device with high reliability. An embodiment is also conceivable.

  In the environmental map generation system 1 shown in FIG. 1, an object identification device 150 that identifies the type of an object is provided as a separate body from the environmental map generation device 200. Instead of this, an embodiment in which the object identification device 150 is integrated into, for example, the third probability calculation circuit 240 of the environment map generation device 200 is also conceivable.

  While various embodiments have been described above with reference to the drawings, it goes without saying that the present disclosure is not limited to such examples. It will be apparent to those skilled in the art that various changes and modifications can be made within the scope of the claims, and these are naturally within the technical scope of the present disclosure. Understood. In addition, the constituent elements in the above embodiments may be arbitrarily combined within the scope not departing from the spirit of the disclosure.

[Summary of this disclosure]
The environmental map generation device according to the present disclosure is configured to determine the first position in the first region by the first sensor device and the feature of the first position identified by the second sensor device. A first probability calculation circuit for calculating a first occupation probability that an object at the first position exists, and an object exists for each position in the second region based on the first occupation probability. A map generation circuit for generating a map indicating the second occupation probability.

  In the environment map generation device of the present disclosure, the second sensor device identifies the feature based on an image captured by the imaging device.

  In the environment map generation device of the present disclosure, the feature is a type of an object that occupies the first position.

  In the environmental map generation device of the present disclosure, the type includes a pedestrian, a shopping cart, a manhole, or snowfall.

  In the environmental map generation device of the present disclosure, the first sensor device is a radar or a sonar.

  In the environment map generation device of the present disclosure, when the type is a pedestrian or a shopping cart, the first probability calculation circuit increases the occupation probability of the first position.

  In the environment map generation device of the present disclosure, when the type is snowfall or manhole, the first probability calculation circuit reduces the occupation probability of the first position.

  In the environment map generation device of the present disclosure, the map is a grid map.

  The environmental map generation method according to the present disclosure is based on the measurement value of the first position in the first region by the first sensor device and the feature of the first position identified by the second sensor device. A first occupancy probability that an object exists at the first position is calculated, and a second occupancy probability that an object exists for each position in the second region is indicated based on the first occupancy probability Generate a map.

  The environment map generation program according to the present disclosure causes a computer to respond to the measurement value of the first position in the first region by the first sensor device and the feature of the first position identified by the second sensor device. Calculating a first occupancy probability that an object exists at the first position and, based on the first occupancy probability, a second occupancy where an object exists for each position in the second region Generate a map showing the probability.

  The present disclosure is suitable for driving assistance technology using an environmental map.

DESCRIPTION OF SYMBOLS 1 Environment map generation system 110 Vehicle positioning sensor 120 Sonar 130 Radar 140 Imaging device 150 Object identification device 200 Environment map generation device 210 Vehicle position estimation circuit 220 1st probability calculation circuit 230 2nd probability calculation circuit 240 3rd probability calculation Circuit 250 Environmental map grid occupation probability calculation circuit 260 Environmental map output circuit

Claims (10)

There is an object at the first position according to the measured value of the first position in the first region by the first sensor device and the characteristics of the first position identified by the second sensor device. A first probability calculation circuit for calculating a first occupation probability;
A map generation circuit that generates a map indicating a second occupation probability that an object exists for each position in the second region based on the first occupation probability;
Comprising
Environmental map generator.   The environment map generation device according to claim 1, wherein the second sensor device identifies the feature based on an image captured by the imaging device. The feature is a type of an object that occupies the first position.
The environmental map production | generation apparatus of Claim 1 or 2. The type includes pedestrians, shopping carts, manholes, or snowfall,
The environmental map production | generation apparatus of Claim 3. The first sensor device is a radar or a sonar;
The environmental map production | generation apparatus of Claim 4. When the type is a pedestrian or a shopping cart, the first probability calculation circuit increases the occupation probability of the first position.
The environmental map production | generation apparatus of Claim 5. When the type is manhole or snowfall, the first probability calculation circuit decreases the occupation probability of the first position.
The environmental map production | generation apparatus as described in any one of Claim 4 to 6. The map is a grid map;
The environmental map production | generation apparatus as described in any one of Claim 1 to 7. There is an object at the first position according to the measured value of the first position in the first region by the first sensor device and the characteristics of the first position identified by the second sensor device. Calculate a first occupancy probability;
Based on the first occupancy probability, a map indicating a second occupancy probability that an object exists for each position in the second region is generated.
Environmental map generation method. On the computer,
There is an object at the first position according to the measured value of the first position in the first region by the first sensor device and the characteristics of the first position identified by the second sensor device. Let the first occupancy probability be calculated,
Based on the first occupancy probability, a map indicating a second occupancy probability that an object exists for each position in the second region is generated.
Environmental map generation program.

Images