JP4978099B2 - Self-position estimation device - Google Patents

Description

  The present invention relates to a self-position estimation apparatus that estimates a self-position.

  Conventionally, as a technique for estimating a self-position of a moving body that is mounted on a moving body that moves on a traveling path, there are those shown in Patent Document 1 and Patent Document 2 below. A self-position identification device for a moving body described in Patent Document 1 includes a sign, a traffic light, a power pole, and the like installed on the side of a road (road) as a landmark, an imaging unit that images the landmark, an actual landmark Map data storage means for holding map data in which position and type information is described in advance. In this mobile body self-position identification device, a landmark image is extracted from an image picked up by an image pickup means to recognize the position and type of the landmark, and the position and type of the recognized landmark and map data storage are stored. The self-position of the moving body is identified by collating with the map data stored in the means.

In addition, in the unmanned vehicle guidance system described in Patent Document 2, a landmark placed in advance is photographed by a pair of cameras, and the position of the unmanned vehicle (moving body) with respect to the photographed landmark is obtained and photographed. By comparing the landmark with the reference pattern, the position of the landmark in the absolute coordinate system is determined, and the position of the unmanned vehicle is determined. In this unmanned vehicle guidance system, each landmark is provided with a unique pattern in order to identify a plurality of landmarks.
JP-A-8-247775 JP-A-9-218712

  However, in the prior art described in the above-mentioned Patent Document 1, since the self-position of the moving body is determined by using a fixed object such as a road sign as a landmark, an environment with few common fixed objects such as a road sign (for example, indoors) ), It is difficult to estimate the self-position of the moving object.

  Further, in the prior art described in Patent Document 2, the self-position of the moving body is determined based on the landmark provided with the unique pattern, and therefore it is necessary to previously arrange the landmark at a predetermined position. . In addition, in order to obtain an accurate position, it is necessary to install many landmarks. However, it has been difficult to install many fixed landmarks in an actual environment (for example, indoors) where people live.

  The present invention has been made to solve such a problem, and provides a self-position estimation device capable of estimating the self-position even when a landmark that may move is used. With the goal.

  A self-position estimation apparatus according to the present invention is a self-position estimation apparatus that estimates a self-position by referring to landmark arrangement information data, and includes a collation means that collates a landmark with the landmark arrangement information data, and a collation means. Self-position estimation means for estimating the self-position using the landmarks that can be collated, and the self-position estimation means increases the influence of the landmark with higher reliability than the landmark with low reliability, It is characterized by estimating.

  Such a self-position estimation apparatus includes collation means for collating the landmark with the landmark arrangement information data, and self-position estimation means for estimating the self-position using the landmark that has been collated. The self-position estimating means increases the influence degree of the landmark having the higher reliability than the landmark having the lower reliability when estimating the self-position. Thereby, since the position estimation is performed by reducing the influence of a landmark having low reliability, even an object that is likely to move can be used as a landmark with low reliability. As a result, it is not necessary to previously install a fixed object as a landmark, and the self position can be estimated even in an environment where there are few fixed objects as landmarks. This is particularly effective in estimating the self-location in a living environment where the environment is likely to change. In addition, since the self-position estimation is performed by increasing the influence degree of the landmark having high reliability, the position accuracy can be improved. Further, in the conventional technology, when a moved object is detected, the moved object is excluded without being adopted as a landmark. For this reason, the landmarks used for position estimation are reduced, and the position accuracy may be lowered. However, even if the object is moved, the number of landmarks used for position estimation can be sufficiently secured by using the landmarks with low reliability. As a result, it is possible to prevent a decrease in position accuracy. Note that the “landmark” here is a mark in position estimation, and includes, for example, the object itself existing in the real environment or an arbitrary point of the object. The “landmark” may be a movable object, an arbitrary point of the movable object, a fixed object, or an arbitrary point of the fixed object. The “landmark reliability” here is an index indicating whether the landmark is an effective landmark when estimating the self-position.

  Here, it is preferable that the landmark reliability with a low possibility of movement is evaluated higher than the landmark with a high possibility of movement. As a result, the influence of the landmark having a low possibility of movement is increased and the influence of the landmark having a high possibility of movement is reduced to estimate the self position, so that the position accuracy can be improved.

  In addition, it is preferable that a detection unit for detecting the landmark is further provided, and the landmark reliability that is easy to detect by the detection unit is evaluated higher than the landmark that is difficult to detect by the detection unit. Depending on the characteristics of the detection means, the color, pattern, shape, etc. of the landmark, the same landmark may be detected stably or may not be detected stably. Since the influence of the landmark that can be detected stably is increased and the influence of the landmark that cannot be detected stably is reduced to estimate the self-position, it is possible to improve the position accuracy.

  In addition, a plurality of types of detection means with different position detection accuracy are provided, and the landmark reliability is higher than the landmark detected by the detection means with lower position detection accuracy than the landmark detected by the detection means with higher position detection accuracy. It is preferable that is highly evaluated. For example, when a plurality of types of detection means such as a camera, a laser range finder, and an RFID tag reader are used in combination, the landmark position detection accuracy differs depending on the type of detection means. Since the influence of the landmark detected by the detection means with high position detection accuracy is increased and the influence of the landmark detected by the detection means with low position detection accuracy is reduced to estimate the self position, the position accuracy is improved. You can plan.

  In addition, the landmark reliability is preferably evaluated more highly for a landmark that has not moved than for a landmark that has moved. Thereby, reliability can be changed according to the track record which the landmark moved. In addition, since the influence of the landmark that has not moved can be increased and the influence of the landmark that has moved can be reduced to estimate the self-position, the position accuracy can be improved.

  The reliability of the landmark is preferably evaluated according to the type to which the landmark belongs. Thereby, the reliability of the same kind of landmark can be applied to the landmark detected for the first time according to the kind to which the landmark belongs. For example, when the detected object is a cup, the reliability of another cup detected in the past can be taken over.

  The reliability of the landmark is preferably evaluated according to the place where the landmark exists. For example, if a plurality of landmarks existing in a specific area (location) are frequently moved, it can be determined that the landmarks in this area are easy to move. Reduce the degree. Thereby, the influence of the landmark which exists in this area | region can be made small, and a self-position can be estimated. As a result, position accuracy can be improved. In addition, when a plurality of landmarks existing in a specific area do not move for a certain period of time, it is possible to determine that the landmarks in this area are difficult to move, so the reliability of the landmark information is increased. . Thereby, the influence of the landmark which exists in this area | region can be enlarged, and a self-position can be estimated. As a result, position accuracy can be improved.

  Moreover, it is preferable to further include an evaluation means for evaluating the reliability of the landmark. As a result, the reliability of landmarks can be evaluated in accordance with changes in the actual environment, so that self-location can be estimated using landmarks that correspond to changes in the actual environment, and location accuracy can be improved. Can be achieved.

  In addition, it is preferable that a reliability information storage unit that stores in advance information on the reliability of the landmark is provided. By using information on the reliability of landmarks stored in advance, the processing speed for position estimation can be improved.

  According to the self-position estimation device of the present invention, not only a fixed object but also an object that may move can be used as a landmark to estimate the self-position and improve the position estimation accuracy. Can be planned.

  Hereinafter, a preferred embodiment of a self-position estimation apparatus according to the present invention will be described with reference to the drawings. In the description of the drawings, the same or corresponding elements are denoted by the same reference numerals, and redundant description is omitted.

  The self-position estimation apparatus of the present invention is mounted on a moving body such as a robot that can move in a human living environment, for example, and estimates the self-position of the moving body. This self-position estimation device detects an object, collates a landmark with map data (landmark arrangement information data) using the detected object as a landmark, and moves based on the landmark that can be collated. Estimate the body's self-position. The map data includes information on the absolute position and characteristics of the landmark. The absolute position of the landmark is based on an arbitrary point (for example, a corner of the room) (0, 0, 0). This is the position. As a first embodiment, a self-position estimation apparatus that does not include a map creation function and a map update function (details will be described later), and a self-position estimation apparatus that includes a map creation function and a map update function will be described as a second embodiment. The “landmark” here is a mark in position estimation, and includes an object existing in the real environment and an arbitrary point of this object. The “landmark” may be a movable object and an arbitrary point of the movable object, or may be a fixed object and an arbitrary point of the fixed object.

  First, the self-position estimation apparatus according to the first embodiment will be described with reference to FIGS. FIG. 1 is a system configuration diagram showing a self-position estimation apparatus according to the first embodiment of the present invention. A self-position estimation apparatus 1 shown in FIG. 1 includes a plurality of landmark detection sensors 2 that detect an object used as a landmark, and a position estimation electronic control unit (hereinafter referred to as a position estimation ECU) that executes position estimation processing. 3).

  The landmark detection sensor 2 detects data (for example, image data) for extracting a landmark from the real environment. The self-position estimation apparatus 1 includes a stereo camera (hereinafter referred to as a camera), a laser range finder, an RFID tag reader, and the like as a plurality of landmark detection sensors 2. The detection means of the present invention includes a landmark detection sensor 2 and a landmark extraction unit 6 described later. The landmark detection sensor 2 detects “data for extracting landmarks” and detects the landmark extraction unit 6. To extract landmarks.

  “Data for extracting landmarks” detected by the landmark detection sensor 2 includes information (features) on the features of the landmarks, and positions of the landmarks (relative positions of the landmarks and the moving body). It is information about. Note that “information on the feature of the landmark (feature amount)” is information that can identify the type of the landmark. For example, when an image is picked up by a camera, the outline, pattern, etc. of the object become “information on the feature of the landmark”. The type of landmark is identified based on the outline, the design, and the like.

  The position estimation ECU 3 includes a CPU 4 that performs arithmetic processing, a ROM and a RAM that serve as a storage unit 5, an input signal circuit, an output signal circuit, a power supply circuit, and the like, and is electrically connected to the plurality of landmark detection sensors 2. Yes. The position estimation ECU 3 estimates the position of the moving body with reference to the map data stored in the storage unit 5.

  The storage unit 5 of the position estimation ECU 3 stores a program for operating the CPU 4 and the like. The storage unit 5 of the position estimation ECU 3 stores map data including information on landmarks in advance. The storage unit 5 of the position estimation ECU 3 functions as reliability information storage means of the present invention, and stores in advance the evaluation value of the landmark (landmark information reliability).

  Next, landmark evaluation values will be described. The “landmark evaluation value” here is an index indicating whether the landmark is an effective landmark when estimating the position. For example, a landmark that is difficult to move is a highly reliable landmark. It becomes. In the present embodiment, the landmark is evaluated based on the “landmark evaluation criterion”, and the rank of the evaluation value is classified into 0.0 to 1.0. The evaluation value of the landmark with the highest reliability is “1.0”, and the evaluation value of the landmark with the lowest reliability is “0.0”. In the self-position estimation apparatus 1, landmark evaluation values are set according to a plurality of evaluation criteria (“easy to move due to external factors”, “stability due to sensor characteristics”).

  Next, landmark evaluation based on “ease of movement due to external factors” will be described. In the evaluation of landmarks based on “ease of movement due to external factors”, whether or not landmarks are easily moved artificially is used as an evaluation criterion. Landmark evaluation values are set according to “evaluation by landmark location” and “evaluation by spatial arrangement of landmarks”.

In “evaluation by landmark height”, an evaluation value W ₁ is set according to the height at which the landmark is detected. Things that exist in high places are unlikely to move because human hands are difficult to reach, and those that are in low places are more likely to move because landmarks are difficult to reach. . Therefore, when the detected position of the landmark is high, high and sets an evaluation value W ₁ of the landmark when the detected position of the landmark is low, setting a low evaluation value W ₁ of the landmark.

FIG. 2 is a schematic diagram showing an indoor environment, and shows an example of a landmark for setting an evaluation value based on “evaluation by landmark height”. In the indoor environment shown in FIG. 2, a shelf 12 and a locker 13 are arranged on the floor 11. Here, the landmark 14 on the floor surface 11, the landmark 15 installed on the shelf 12, and the landmark 16 installed on the outer wall of the locker 13 are detected. As the evaluation value W ₁ which is set based on the "evaluation by the height of the landmark" evaluation value W ₁ of the landmark 14 on the floor surface 11 is "0.1", the landmark 15 placed on a shelf 12 the evaluation value _{W 1} of the "0.4", the evaluation value _{W 1} of the landmark 16 installed on the outer wall of the locker "0.9". Accordingly, it is possible to estimate the position by increasing the influence of the landmark in the high place and reducing the influence of the landmark in the low place.

In the "evaluation by the location of the landmark", depending on where the landmark has been detected, to set the evaluation value W _2. That is, according to the movement ease Landmarks within any area, it sets an evaluation value W ₂ for each location (area). For example, those present in the shelves frequently used is liable to move, set the evaluation value W ₂ landmarks present in the shelf lower, those present in the shelves rarely use, to move hard for, setting a high evaluation value W ₂ of the landmark. Moreover, those present in the floor surface, and is easy to artificially moved, setting a low evaluation value W ₂ landmarks.

FIG. 3 is a schematic diagram showing an indoor environment, and shows an example of a landmark for setting an evaluation value based on “evaluation by landmark location”. As shown in FIG. 3, the landmark 17 on the floor surface 11, the landmark 18 installed on the middle stage 12a of the shelf 12, and the landmark 19 installed on the upper stage 12b of the shelf 12 are detected. The middle stage 12 a of the shelf 12 is a place area that is less frequently used than the upper stage 12 b of the shelf 12. As an evaluation value W ₂ that is set based on the "evaluation by location of landmark" evaluation value W ₂ landmark 17 on the floor surface 11 is "0.1", installed in the middle 12a of the shelf 12 Land evaluation value _{W 2} of the mark 18 is "0.8", the evaluation value _{W 2} landmarks 19 installed on the upper 12b of the shelf 12 is "0.4". Thus, according to the degree of location of environmental change landmark is detected, it is possible to change the evaluation value W _2.

In "Evaluation of the spatial arrangement of the landmark", depending on the spatial arrangement of landmarks is detected, it sets an evaluation value W _3. In this case, it is preferable to measure the spatial arrangement of the landmarks based on data detected by the three-dimensional space restoration sensor (for example, a stereo camera or a laser range finder) as the landmark detection sensor 2. Those that are in contact with the vertical surface are often fixed to the vertical surface, so the landmark is unlikely to move. Those that are in contact with the horizontal surface are on the horizontal surface (for example, the floor surface or the upper surface of the desk). Since the landmarks are often arranged as they are, there is a high possibility that the landmark will move. Therefore, when the landmark is in contact with the vertical plane, set high evaluation value W ₃ landmarks, when the landmark is in contact with the horizontal plane, setting a low evaluation value W ₃ landmarks .

FIG. 4 is a schematic diagram showing an indoor environment, and shows an example of a landmark for setting an evaluation value based on “evaluation by spatial arrangement of landmarks”. As shown in FIG. 4, the landmark 20 on the floor surface 11 (horizontal plane), the landmark 21 existing on the top plate 12 c (horizontal plane) of the shelf 12, and the land installed on the outer wall 13 a (vertical plane) of the locker 13. A mark 22 is detected. As an evaluation value W ₃ that is set based on the "evaluation by spatial arrangement of landmarks" evaluation value W ₃ landmarks 20 on the floor surface 11 is "0.2", the top plate 12c of the shelf 12 evaluation value _{W 3} landmarks 21 that exists is "0.2", the evaluation value _{W 3} landmarks 22 installed on the outer wall 13a of the rocker 13 is "0.8". Thus, based on the spatial arrangement of the landmarks, it is possible to change the evaluation value W _3.

  Next, landmark evaluation based on “stability by sensor characteristics” will be described. In the evaluation of the landmark based on “stability by sensor characteristics”, whether or not the same landmark can be detected stably even if detected multiple times by the sensor characteristics of the landmark detection sensor 2 is used as an evaluation criterion. Specifically, landmark evaluation values are set in accordance with “evaluation by landmark characteristics” and “evaluation by landmark detection sensor”.

In “evaluation based on landmark characteristics”, the evaluation value W ₄ is set according to the ease of detection by the landmark detection sensor 2 (detection stability). For example, a corner portion in an image acquired by a camera may or may not be detected as a landmark depending on the contrast between the corner portion and its surroundings. A sharp edge with high contrast may be detected stably, and a blurred edge with low contrast may not be detected stably. Further, depending on the color, shape, pattern, and the like of the landmark, the landmark detection sensor 2 may detect it stably or may not detect it stably. Therefore, the landmark detection sensor 2, landmarks are detected stably is set higher the evaluation value W _4, landmark can not be detected stably, the setting a low evaluation value W ₄ .

In "Evaluation of landmark detection sensor", according to the position detection accuracy of the landmark by landmark detection sensor 2, sets an evaluation value W _5. When a plurality of types of landmark detection sensors 2 are provided, the position detection accuracy differs depending on the landmark detection sensor 2. Therefore, by setting a high evaluation value W ₅ landmark position detection accuracy is detected by the high landmark detection sensor 2, the evaluation value W ₅ landmark position detection accuracy is detected by the lower landmark detection sensor 2 Set low.

（ただし、Ｗ_ｆ：最終評価値（０．０〜１．０）、Ｗ_α：各評価基準での評価値、Ａ_α：各評価基準の重み、Ｎ：評価基準数）
なお、「各評価基準の重みＡ_α」は、実験データや過去の実績等によって求められる数値であり、評価基準の重要度（優先度）に応じて、設定される。 For each of the plurality of evaluation values set according to these evaluation criteria, an average value is calculated by the following formula (1) for each landmark, and position estimation is performed using the average (final evaluation value W _f ) of the evaluation values. Do.

(However, _{W f:} final evaluation value (0.0 to 1.0), W _alpha: evaluation value at each evaluation criteria, A _alpha: weighting of each criteria, N: Based Number)
The “weight A _α of each evaluation criterion” is a numerical value obtained from experimental data, past results, etc., and is set according to the importance (priority) of the evaluation criterion.

The absolute position (x, y, z) of the landmark, the feature amount, and the evaluation values W _{1 to} W ₅ , W _f are stored in the storage unit 5 as map data.

  The CPU 4 of the position estimation ECU 3 functions as a landmark extraction unit 6, a map collation unit (collation unit) 7, and a position estimation unit (self-position estimation unit) 8 by executing a program stored in the storage unit 5. .

  The landmark extraction unit 6 extracts a landmark from each data (“data for extracting a landmark”) acquired by the landmark detection sensor 2. The landmarks to be extracted are corners, edges, characteristic shapes and designs of the object. FIG. 5 is a schematic diagram illustrating an example of an image captured by the camera. As shown in FIG. 5, for example, landmarks are extracted based on image data captured by a camera. As the landmark, for example, a corner portion of a shelf, a corner portion of a book, and the like are extracted.

  The map collating unit 7 collates the landmark extracted by the landmark extracting unit 6 with the map data stored in the storage unit 5. FIG. 6 is a schematic diagram illustrating an example of map data (landmark positions) stored in the storage unit. More specifically, the feature quantity of each landmark extracted is compared with the feature quantity of each landmark of the map data, and a search is made for a match between the feature quantities. In the case of the example shown in FIGS. 5 and 6, the landmarks 23, 24, 25, and 26 shown in FIG. 5 correspond to the landmarks 27, 28, 29, and 30 shown in FIG.

  In the position estimation unit 8, a position estimation process for estimating the self-position of the moving object based on information on the landmarks that can be collated by the map collation unit 7 (detected data, map data stored in the storage unit 5). I do. When a landmark having an evaluation value equal to or greater than the threshold is sufficiently extracted, position estimation is performed based on the landmark having an evaluation value equal to or greater than the threshold. On the other hand, when a landmark having an evaluation value equal to or greater than the threshold is not sufficiently extracted, the position is estimated by increasing the degree of influence of the landmark having a high evaluation value.

（ただし、β：ランドマークナンバー、Ｗ_ｆβ：最終評価値（０．０〜１．０）、ｍ_β：検出したランドマークの位置（ｘ，ｙ）、Ｍ_β：地図データとして記憶されたランドマークの３次元位置（Ｘ，Ｙ，Ｚ）、Ｋ：カメラの内部パラメータ（カメラをキャリブレーションしておくことで既知となる定数）、Ｒ：回転成分（ロール、ヨー、ピッチ）、ｔ：平行移動成分（Ｘ，Ｙ，Ｚ））
この数式（２）を用いて、位置推定を行うことで、位置推定精度の向上を図ることができる。また、ランドマークとしての固定物を設置していない環境であっても、位置推定を行うことができる。 For example, when position estimation is performed using image data captured by a camera and landmark three-dimensional position information, the following equation (2) “weighted evaluation function P” is used. By using the following equation (2), the rotation component R (roll, yaw, pitch) and the translation component t (X, Y, Z) are obtained so that the weighted evaluation function P is minimized. Estimate the position of the body.

(Where β: landmark number, W _fβ : final evaluation value (0.0 to 1.0), m _β : detected landmark position (x, y), M _β : land stored as map data. Mark three-dimensional position (X, Y, Z), K: camera internal parameters (constants known by calibrating the camera), R: rotation components (roll, yaw, pitch), t: parallel Moving component (X, Y, Z))
By performing position estimation using this mathematical formula (2), it is possible to improve position estimation accuracy. Further, position estimation can be performed even in an environment where no fixed object as a landmark is installed.

  Next, the basic operation of the self-position estimation apparatus 1 according to the first embodiment will be described with reference to the drawings. Each landmark detection sensor 2 collects information around the moving body and transfers it to the position estimation ECU 3. For example, the camera is used to image the surroundings of the moving body, and the image data is transferred to the position estimation ECU 3.

  Next, the position estimation process executed by the position estimation ECU 3 will be described with reference to the flowchart of FIG. FIG. 7 is a flowchart showing an operation procedure of position estimation processing executed by the position estimation ECU. First, the position estimation ECU 3 inputs detection data from each landmark detection sensor 2 (S1). Next, the position estimation ECU 3 performs landmark extraction processing to extract landmarks (S2).

The position estimation ECU 3 extracts a landmark and then performs a map matching process (S3). Next, using the landmark information that has been verified, self-position estimation of the moving object is performed (S4). Here, a self-position estimation by changing the degree of influence with respect to the position estimation according to the evaluation value W _f for each landmark. The position estimation ECU 3 estimates the self position of the moving body, and then outputs the result to, for example, a control unit that controls the moving body (S5) and ends the position estimation process.

  In such a self-position estimation apparatus 1, the object detected by the landmark detection sensor 2 is used as a landmark, and the position estimation is performed while reducing the influence of a landmark having a low evaluation value. Even an object can be used as a landmark. As a result, it is not necessary to previously install a fixed object as a landmark, and the self position can be estimated even in an environment where there are few common fixed objects as landmarks. This is particularly effective in estimating the self-location in a living environment where the environment is likely to change. Further, since the self-position estimation is performed by increasing the influence degree of the landmark having a high evaluation value, it is possible to improve the position accuracy. Further, in the conventional technology, when a moved object is detected, the moved object is excluded without being adopted as a landmark. For this reason, the landmarks used for position estimation are reduced, and the position accuracy may be lowered. However, in the self-position estimation apparatus 1, even a moving object can be used as a landmark with low reliability, so that the number of landmarks used for position estimation can be sufficiently secured. As a result, it is possible to prevent a decrease in position accuracy.

Further, the evaluation value W ₁ based on “evaluation by the height of the landmark” is evaluated according to the height at which the landmark is detected. When the detected position of the landmark is high, it is determined that the landmark is not easily moved, evaluation value W ₁ of the landmark is set high. When the detected position of the landmark is low, it is determined that the likely landmark is moved, the evaluation value W ₁ of the landmark is set low. As a result, the influence of the landmark that is unlikely to move can be increased and the influence of the landmark that is likely to move can be reduced to estimate the self-position, thereby improving the position accuracy.

Further, by performing the self-position estimation using the evaluation value W ₂ based on the "evaluation by location of landmark", to increase the influence of the low landmarks likely to move, landmark likely to move The influence of can be reduced. Thereby, the position accuracy can be improved.

Further, by performing the self-position estimation using the evaluation value W ₃ based on the "evaluation by spatial arrangement of landmarks" to increase the effect of low landmarks likely to move, the possibility of moving It is possible to estimate the self-position by reducing the influence of high landmarks. Thereby, the position accuracy can be improved.

Also, by using the evaluation value W ₄ based on “evaluation based on the characteristics of the landmark”, the influence of the landmark that can be detected stably is increased by estimating the self-position, and the landmark that cannot be detected stably is detected. The self-position of the moving object can be estimated by reducing the influence. Thereby, the position accuracy can be improved.

Further, by using the evaluation value W ₅ based on “evaluation by the landmark detection sensor”, the self-position is estimated, thereby increasing the influence of the landmark detected by the landmark detection sensor 2 with high position detection accuracy. Thus, the influence of the landmark detected by the landmark detection sensor 2 with low position detection accuracy can be reduced, and the self position of the moving object can be estimated. Thereby, the position accuracy can be improved.

  In addition, since the self-position estimation apparatus 1 includes the storage unit 5 in which the landmark evaluation values W1 to W5 are stored in advance, the processing speed in the position estimation ECU 3 can be improved.

  Next, a self-position estimation apparatus according to a second embodiment of the present invention will be described with reference to the drawings. FIG. 8 is a system configuration diagram showing a self-position estimation apparatus according to the second embodiment of the present invention. The self-position estimation device 41 of the second embodiment is different from the self-position estimation device 1 of the first embodiment in that, in place of the position estimation ECU 3 having a position estimation function, a map creation function, This is a point that includes a position estimation ECU 43 having a map update function and a position measurement unit 42.

  The position estimation ECU 43 can estimate the self-position as in the first embodiment. Further, the position estimation ECU 43 can perform map creation and map update.

  The self-position estimation device 41 includes a position measurement unit 42 that measures the position of a landmark to be registered as map data at the time of map creation and map update. The position measuring unit 42 is configured by a GPS, a surveying instrument, or the like. Information on the measured position of the landmark is input to the map creation unit 48 and stored in the storage unit 5. Note that the position measurement unit 42 may not be provided, and information regarding the position of the landmark may be manually input, or the landmark detection sensor 2 may be used.

  Next, the map creation function will be described. The CPU 44 of the position estimation ECU 43 functions as a landmark extraction unit 6, an evaluation value setting unit (evaluation means) 47, and a map creation unit 48 by executing a program stored in the storage unit 5 at the time of map creation. . The map creation is executed to create initial data when the self-position estimation apparatus 1 is activated.

The evaluation value setting unit 47 sets evaluation values W _{1 to} W ₅ for the landmarks extracted by the landmark extraction unit 6. The evaluation value setting unit 47 uses a plurality of evaluation criteria “easy to move due to external factors”, “sensor characteristics” for each landmark extracted by the landmark extraction unit 6 by the same method as in the first embodiment. Landmark evaluation values W _{1 to} W ₅ are set according to “stability by”. Further, the evaluation value setting unit 47, based on the evaluation value _W 1 to _W-5, and calculates the final evaluation value _{W f.}

In the map creation unit 48, for each landmark, data including information regarding the evaluation values W _{1 to} W ₅ and the final evaluation value W _f set by the evaluation value setting unit 47 and the position and characteristics of the extracted landmarks is obtained as map data. Is stored in the storage unit 5 as follows.

  Next, the map update function will be described. The CPU 44 of the position estimation ECU 43 updates the map when the movement of the landmark is detected and when a new landmark is detected during the position estimation. The CPU 44 of the position estimation ECU 43 functions as an evaluation value setting unit 47 and a map creation unit 48 by executing a program stored in the storage unit 5 when updating the map.

When the map collation unit 7 detects the movement of the landmark, the evaluation value setting unit 47 resets the evaluation values W _{1 to} W ₅ and the final evaluation value W _f . Further, the evaluation value setting unit 47 resets the landmark evaluation value W in accordance with “evaluation based on the movement result of the landmark”. The map matching unit 7 lowers the evaluation value W of the landmark determined to have moved, while increasing the evaluation value W of the landmark determined to have not moved for a certain period of time.

  FIG. 9 is a schematic diagram showing an indoor environment, and shows an example of a landmark for setting an evaluation value based on “evaluation based on the movement result of the landmark”. As shown in FIG. 9, for example, when it is determined that the landmark 51 existing on the outer wall of the rocker 13 has moved from the position A indicated by the broken line to the position B indicated by the solid line, the evaluation value of the landmark 51 is set to “1. Decrease from “0” to “0.8”.

Further, the evaluation value setting unit 47, resets evaluation value for each region to be applied in the "Evaluation of the location of the landmark" (the area correction value) W _2. If the landmark in the region is more moving, while lowering the evaluation value W ₂ landmarks existing in the same area, if the landmark in the region has not moved a certain time are the same increase the evaluation value W ₂ of the landmark that exist in the area. Thereby, the movement result of the landmark in the area is reflected in the setting of the evaluation value of the other landmark in the same area.

FIG. 10 is a schematic diagram showing an indoor environment, and shows an example of a landmark for setting an evaluation value based on “evaluation based on the location of the landmark”. As shown in FIG. 10, when a plurality of landmarks 52 to 55 in the upper 12b of the shelf 12 (location area) is moving, the evaluation value _{W 2} of the landmark 52 to 55 from the "1.0" Lower to “0.6”.

When a new landmark is detected, the evaluation value setting unit 47 sets the evaluation values W _{1 to} W ₅ . Therefore, the evaluation value setting unit 47 sets the evaluation value of the landmark according to “evaluation according to the type to which the landmark belongs”. Specifically, it is possible to determine the type to which the newly detected landmark belongs (whether or not it has the same feature amount) and take over the evaluation values W _{1 to} W ₅ of the same type of landmark. Since the objects of the same type are likely to exhibit similar properties such as the presence / absence of movement and the place where they are placed, the evaluation values W _{1 to} W ₅ can be taken over. For example, when a cup is extracted as a landmark, the place where the cup is arranged is likely to be substantially limited, and the landmark is updated by taking over the evaluation values W _{1 to} W ₅ of other cups. The processing speed can be increased.

FIG. 11 is a table showing an example of landmark evaluation values. For each landmark (L1 to L4), the type of landmark detection sensor 2, position, feature amounts, a plurality of evaluation values _W 1 to _W-3, the detection number of times are stored. In FIG. 11, describes an evaluation value _W 1 to _W-3, are omitted in the evaluation value _W 4, _{W 5.} For example, when a new landmark L4 is detected, the stored feature amounts V1 to V3 of the landmarks L1 to L3 are compared with the feature amount V3 of the new landmark L4, and a landmark having the same feature amount V3 is compared. The evaluation value of L3 can be taken over.

  Next, the basic operation of the self-position estimation apparatus 41 according to the second embodiment will be described with reference to the drawings. Hereinafter, the control process in the position estimation ECU 43 will be described. Since the position estimation process is the same as the self-position estimation apparatus 1 of the first embodiment, it will be omitted and the map creation process and the map update process will be described.

  The map creation process executed by the position estimation ECU 43 will be described with reference to the flowchart of FIG. FIG. 12 is a flowchart showing an operation procedure of map creation processing executed by the position estimation ECU. In the detection data input (S1) and landmark extraction process (S2), the same process as the position estimation process of the first embodiment is executed.

The position estimation ECU 43 performs the evaluation value setting process after extracting the landmark (S23). Here, "Evaluation by landmark height", "Evaluation by landmark location", "Evaluation by spatial arrangement of landmarks", "Evaluation by landmark characteristics", "Evaluation by landmark detection sensor" depending on the ", to set the evaluation value _W 1 ~W ₅ of the landmark.

After setting the landmark evaluation values W _{1 to} W ₅ , the position estimation ECU 43 creates landmark data (map data) (S 24). The storage unit 5 stores the created landmark data as map data (S25).

  Next, map update processing executed by the position estimation ECU 43 will be described with reference to the flowchart of FIG. FIG. 13 is a flowchart showing an operation procedure of map update processing executed by the position estimation ECU. The map update process executes the map update process when the movement of the landmark is detected in the position estimation process and when a new landmark is detected.

The position estimation ECU 43 determines whether or not the movement of the landmark is detected (S31). When the movement of the landmark is detected in S31, an evaluation value resetting process is executed (S32). Here, "Evaluation by landmark height", "Evaluation by landmark location", "Evaluation by spatial arrangement of landmarks", "Evaluation by landmark characteristics", "Evaluation by landmark detection sensor" The landmark evaluation values W _{1 to} W ₅ and W _f are reset according to “Evaluation based on landmark movement results”. In addition, in the case where the evaluation value W ₂ on the basis of the "evaluation by the location of the landmark" has been changed, the evaluation value W ₂ of the landmark existing in the same area is also changed.

On the other hand, if it is not determined in S31 that the movement of the landmark has been detected, that is, if a new landmark has been detected, an evaluation value setting process is executed (S33). Here, the evaluation value of the landmark is set according to “evaluation by the type to which the landmark belongs”. When the evaluation value (past data) of the same type of landmark is stored, the past evaluation value is taken over. When the evaluation values of the same type of landmark are not stored, the evaluation values W _{1 to} W ₅ are newly set.

  The position estimation ECU 43 updates the landmark data (map data) after executing the evaluation value resetting process and the evaluation value setting process (S34). The storage unit 5 stores the updated landmark data as map data (S25).

  According to the self-position estimation device 41 according to the second embodiment, the same effects as the self-position estimation device 1 of the first embodiment can be obtained, and in addition, a map creation function and a map update function are provided. Therefore, landmark evaluation values can be set in accordance with changes in the actual environment. Therefore, it is possible to estimate the self-position of the moving object using the landmark evaluation value corresponding to the change in the actual environment, and the position accuracy is improved.

  As mentioned above, although this invention was concretely demonstrated based on the embodiment, this invention is not limited to the said embodiment. In the above embodiment, the evaluation criteria “evaluation by landmark height”, “evaluation by landmark location”, “evaluation by spatial arrangement of landmarks”, “evaluation by landmark characteristics”, “landmarks” The landmark evaluation value W is set based on the “evaluation by the detection sensor”, but the landmark evaluation value (landmark information reliability) may be set based on other evaluation criteria. In short, it is only necessary to increase the degree of influence of highly reliable landmark information and estimate the self-position of the moving object.

  In the above embodiment, the landmark evaluation value is set using a plurality of evaluation criteria. However, the landmark evaluation value may be set using a single evaluation criterion.

  Further, in the second embodiment, the map data creation process is executed by the position estimation ECU. However, the map data creation ECU that executes the map data creation process and the position estimation ECU are separately provided. The data may be transmitted / received between the map data creation ECU, the position estimation ECU, and the map data storage. The map data creation process does not need to be executed by the ECU mounted on the mobile body, and map creation and map update may be performed by an ECU installed separately from the mobile body.

  Moreover, in the said embodiment, although the self-position estimation apparatus is applied to moving bodies, such as a robot, you may apply to other moving bodies, such as a passenger car, for example.

  Moreover, in the said embodiment, although it is set as the self-position estimation apparatus provided with the several landmark detection sensor 2, it is not necessary to provide multiple landmark detection sensors, The self-position estimation apparatus provided with one landmark detection sensor But you can.

1 is a system configuration diagram showing a self-position estimation apparatus according to a first embodiment of the present invention. It is the schematic which shows an indoor environment, and shows an example of the landmark which sets the evaluation value based on "evaluation by the height of a landmark". It is the schematic which shows indoor environment, and shows an example of the landmark which sets the evaluation value based on "evaluation by the place of a landmark." It is the schematic which shows indoor environment, and shows an example of the landmark which sets the evaluation value based on "evaluation by the spatial arrangement | positioning of a landmark." It is the schematic which shows an example of the image imaged with the camera. It is the schematic which shows an example of the map data memorize | stored in the memory | storage part. It is a flowchart which shows the operation | movement procedure of the position estimation process performed with ECU for position estimation. It is a system block diagram which shows the self-position estimation apparatus which concerns on 2nd Embodiment of this invention. It is the schematic which shows an indoor environment, and shows an example of the landmark which sets the evaluation value based on "evaluation by the movement result of a landmark". It is the schematic which shows indoor environment, and shows an example of the landmark which sets the evaluation value based on "evaluation by the place of a landmark." It is a table | surface which shows an example of map data. It is a flowchart which shows the operation | movement procedure of the map creation process performed with ECU for position estimation. It is a flowchart which shows the operation | movement procedure of the map update process performed with ECU for position estimation.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1,41 ... Self-position estimation apparatus, 2 ... Landmark detection sensor (detection means), 3 ... ECU for position estimation, 5 ... Memory | storage part (reliability information storage means), 7 ... Map collation part (collation means), 8 ... position estimation part (self-position estimation means), 43 ... position estimation ECU, 47 ... evaluation value setting part (evaluation means).

Claims (9)

A self-position estimation device that estimates self-position with reference to landmark arrangement information data,
Collating means for collating landmarks with the landmark arrangement information data;
Self-position estimating means for estimating the self-position using the landmarks that can be collated by the collating means,
The self-position estimation means estimates a self-position using the extracted landmarks when a predetermined number or more of landmarks having a reliability greater than or equal to a threshold are extracted , and the reliability is greater than or equal to the threshold. When a predetermined number or more of landmarks are not extracted , the influence of the landmark with high reliability is made larger than the landmark with low reliability, and self-location is estimated using the landmark that has been verified. A self-position estimation apparatus characterized by:   2. The self-position estimation apparatus according to claim 1, wherein the landmark reliability is evaluated to be higher for a landmark having a low possibility of movement than a landmark having a high possibility of movement. It further comprises detection means for detecting a landmark,
3. The self-position estimation apparatus according to claim 1, wherein the landmark reliability that is easy to detect by the detection unit is evaluated higher than the landmark that is difficult to detect by the detection unit. . Provided with a plurality of types of detection means having different position detection accuracy,
The landmark detected by the detection unit having a high position detection accuracy is more highly evaluated as the reliability of the landmark than the landmark detected by the detection unit having a low position detection accuracy. The self-position estimating apparatus according to any one of Items 1 to 3.   The self-position estimation apparatus according to any one of claims 1 to 4, wherein the landmarks that have not moved are evaluated more highly than the landmarks that have moved.   The self-position estimation apparatus according to any one of claims 1 to 5, wherein the reliability of the landmark is evaluated according to a type to which the landmark belongs.   The self-position estimation apparatus according to any one of claims 1 to 6, wherein the reliability of the landmark is evaluated according to a place where the landmark exists.   The self-position estimation apparatus according to claim 1, further comprising an evaluation unit that evaluates the reliability of the landmark.   The self-position estimation apparatus according to any one of claims 1 to 8, further comprising a reliability information storage unit that stores in advance information related to the reliability of the landmark.

Images