JP2008165275A - Mobile body with self-position identification device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a self-position identification device of a mobile body capable of reducing a processing cost of landmark information necessary for self-position identification even a target environment is in a large scale and also flexibly dealing with changes in the environment. <P>SOLUTION: The mobile body with the self-position identification device includes a local map and a global map, and stores a landmark candidate and a landmark collation result in the local map to use for the next landmark collation. Further, while newly registering a landmark candidate without a corresponding landmark in the collation process, the mobile body deletes an already registered landmark that has not been observed for a long term. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a self-position identification device for a mobile body that moves while observing the environment to sequentially identify the self-position in the environment.

  The self-position identification device according to the present invention sequentially identifies the absolute position of the moving body with respect to a reference point determined in the environment based on the observation results of the external sensor and the internal sensor provided in the moving body.

  Patent Document 1 shows a basic configuration for self-position identification of a moving body. FIG. 3 is a block diagram showing the overall configuration of the self-position identification system, in which 301 is a landmark recognition unit that extracts a landmark from an image obtained by a camera, and 302 is obtained from the landmark recognition unit 301 as needed. An observation map creation unit 303 obtains a row of landmarks together with a positional relationship between the landmarks using a landmark recognition result, and creates / updates an observation map indicating the movement history of the moving object, and 303 stores observation map information Observation map information storage unit 304, a movement amount calculation unit 304 that obtains an approximate movement amount of the moving object with an internal sensor, 305 a global position calculation unit that obtains the approximate position of the moving object in the travel route based on the travel distance, etc. 306 A map data storage unit that stores map data obtained by collecting landmarks in the actual environment as information in advance, and 307 is a map data storage unit A map data management unit that extracts data around the approximate position obtained by the global position calculation unit 305 from 06, and 308 is a movement in the travel route by comparing the map data sent from the map data management unit with the observation map It is a collation part which identifies the self position of a body. The self-position identification system of Patent Document 1 identifies a self-position by collating an observation map obtained by connecting a plurality of landmark rows with map data, and this point is claimed as a feature.

  Patent Document 2 discloses an environment identification device for a mobile robot device in an environment where a landmark exists to identify a current environment from a plurality of registered environments. FIG. 4 is a block diagram thereof, in which 401 is an environment map construction unit that constructs an environment map having landmark information of the environment in which the robot apparatus is currently located, and 402 is location information of landmarks in the environment. And an environment map storage unit having a database composed of a plurality of environment maps and the environment IDs, and 403 is an environment map construction unit 401 that is currently constructed by the robot apparatus. An environment identification unit 404 that identifies the environment in which the robot apparatus is currently placed by comparing the degree of similarity between the environment map of the placed environment and a known environment map stored in the environment map storage unit 402; Is an environment search unit for walking the robot apparatus to recognize a new environment. The environment identification device in Patent Document 2 is different from Patent Document 1 in that the robot device is correctly identified even when the robot device is turned off and then moved to another environment and then restarted. The purpose is to do.

  Patent Document 1 and Patent Document 2 are common to many conventional self-position identification devices in that the self-position is identified by collating and matching the landmark arrangement stored in advance with the landmark observation result. ing.

  Patent Document 3 discloses an autonomous mobile device having a self-position identification function that pays attention to the possibility of causing an error in the identification position due to the calculation cost generated during matching and the presence of a similar environment. In Patent Document 3, the self-position recognition means for estimating and correcting the self-position is preset based on the estimated self-position when collating the environment information obtained by the environment information acquisition means with the environment information included in the map information. The collation is performed within a predetermined range, and when correction information cannot be obtained by this collation, measures are taken such that the predetermined range is expanded and collation is performed again, or the predetermined range is shifted and collation is performed again. By limiting the collation range to a predetermined range, the calculation cost for collation can be reduced. Further, by setting the predetermined range in the vicinity of the estimated self-position, it is possible to reduce the possibility of erroneous association with the similar environment in the map information.

JP-A-8-247775 (page 16, FIG. 1) JP 2004-110802 A (page 40, FIG. 5) Japanese Patent Laying-Open No. 2004-5593 (page 4-5)

  As described above, there is a method for identifying the self-position of the moving object by collating and matching the landmark arrangement stored in advance with the landmark observation result. However, in Patent Document 1 and Patent Document 2, as pointed out in Patent Document 3, there is a problem that an error may occur in the identification position due to the cost of calculation occurring at the time of collation or the presence of a similar environment. Japanese Patent Application Laid-Open No. H10-228561 provides a countermeasure for this. However, since the configuration is such that past verification results are not used when the environmental information obtained by the environmental information acquisition means and the environmental information included in the map information are collated, unnecessary calculation processing may occur. In addition, when environmental conditions change during driving or when accurate observation is not possible, the actual landmark observation result and the landmark arrangement stored in advance differ from each other. It was. That is, when a part of a previously memorized landmark is removed, when a landmark other than a memorized landmark is newly installed, a part of a landmark memorized in advance If the observation cannot be performed due to the shadow of another moving object or the like, the actual landmark observation result and the landmark arrangement stored in advance are different, and collation fails. This problem is caused by the assumption that the landmark is always present at a predetermined position in the environment and can be observed stably.

In consideration of the above problems, the present invention provides a self-position identification device for a mobile object that can identify a self-position with a small amount of calculation and that can identify the self-position while flexibly adapting to changes in the environment. The purpose is to do.

  In order to solve the above problem, the present invention constitutes the following mobile body self-position identification device.

The invention according to claim 1 includes a self-position identification device that identifies a position of the self by observing a landmark in the environment, and moves while identifying the self-position in the environment by the self-position identification device. In a mobile object, the self-position identification device stores a landmark candidate extraction unit that extracts a landmark candidate from an observation result of an external sensor that observes the landmark around the mobile object, and information related to the landmark candidate A map construction unit that constructs a local map and a global map that contains information about the landmarks in the entire area of the environment, a local map storage unit that stores the local map, and a global map storage that stores the global map An estimated position correction unit that calculates the latest estimated position of the moving object from the observation result of the external sensor and the information of the global map, A position estimation unit that outputs a self-position identification result in the environment from information obtained from an internal sensor of the mobile body and the mobile body position estimation value, and the map construction unit constructs the global map In determining whether the landmark candidate matches the registered landmark of the global map, and when it is determined that they do not match, the landmark candidate is additionally registered in the global map as the new landmark. The mobile body is equipped with a self-position identification device that performs the processing.
According to a second aspect of the present invention, in constructing the global map, the map constructing unit constructs the global map separately for each of a plurality of small regions, and uses the landmark observation result as the landmark candidate information. As a moving object, the self-position identification device according to claim 1 is stored for each small area.
According to a third aspect of the present invention, when the map construction unit determines that the landmark candidate matches the registered landmark by the determination, the landmark candidate and the global area are included in the local map. 3. A mobile body comprising the self-position identification device according to claim 1 or 2, wherein processing for storing the landmark candidates including the correspondence information with the registered landmarks on the map is performed.
According to a fourth aspect of the present invention, the map construction unit first determines the landmark candidate in the local map and the previous landmark candidate saved in the local map in the previous process before making the determination. The mobile body having the self-position identification device according to claim 3, wherein the determination is performed only for the landmark candidate for which a matching relationship with the previous landmark candidate is obtained in the verification.
The invention according to claim 5 refers to the correspondence information of the previous landmark candidate obtained in the collation with the landmark candidate when the map construction unit makes the determination, and 5. The mobile body including the self-position identification device according to claim 4, wherein the registered landmark having a coincidence relationship with a mark candidate is judged to coincide with the landmark candidate.
In the invention according to claim 6, the map construction unit obtains the coincidence relationship between the landmark candidate and the previous landmark candidate in the collation, but in the reference, the previous landmark candidate is the When the correspondence information with the registered landmark is not possessed, the number of matching matches is added, and the landmark candidate whose matching matching count is equal to or greater than a threshold value and the registered landmark are 6. A moving body provided with the self-position identification device according to claim 5 for determining whether there is a match.
In the invention according to claim 7, in the determination, the map construction unit identifies the small area where the landmark candidate exists, and sequentially stores information on the registered landmarks existing in the small area. Read, calculate the distance to each landmark candidate, and among the registered landmarks whose calculated distance is less than the threshold, the nearest registered landmark matches the landmark candidate. A mobile body provided with the self-position identification device according to claim 6 for discriminating from an object.
In the invention according to claim 8, the map construction unit stores the observation time in the global map together with the landmark observation result, and the registered landmarks that have passed a specified time or more from the current time. It was set as the mobile body provided with the self-position identification apparatus in any one of Claim 2 to 7 deleted from the said global map.
The invention according to claim 9 is the self-position according to claim 8, wherein the specified time is set to a value larger than a time required for one turn of the lap when the moving body circulates in the environment. It was set as the mobile body provided with the identification apparatus.

According to the first aspect of the present invention, even when the landmark environment changes, the global map held by the self-position identification device can be flexibly adapted and self-position identification can be continued.
According to the third, fourth, and fifth aspects of the present invention, it is possible to significantly reduce the amount of calculation required for landmark matching by using the landmark matching result of the previous cycle.
According to the invention described in claim 6, since only the landmark candidates that can be observed stably coincide with the landmarks of the global map, the amount of calculation can be reduced. Furthermore, it is possible to eliminate the influence of external sensor noise included in the observation results of the landmark candidates.
According to the sixth aspect of the present invention, it is possible to more reliably determine the coincidence between the observed landmark and the registered landmark.
According to the second and seventh aspects of the present invention, priority can be given to collation between landmark candidates that have a high possibility of corresponding to each other and registered landmarks, and the amount of calculation required for collation can be reduced.
According to the eighth aspect of the present invention, the landmark information removed from the environment can be erased from the global map, and the storage area can be saved.
According to the ninth aspect of the present invention, it is possible to delete the landmark information removed from the environment from the global map and save the storage area, particularly for a moving body that orbits the same orbit.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a block diagram showing a basic configuration of a mobile unit self-position identification apparatus according to the present invention. In the figure, a dotted line 100 is a self-position identification device, 101 is a position estimation unit, 102 is a landmark candidate extraction unit, 103 is a map construction unit, 104 is a local map storage unit, 105 is a global map storage unit, and 106 is an estimated position. The correction unit is a signal 121 and 122 which are input signals of the self-position identification device 100, and are external sensor measurement results and internal sensor measurement results, respectively, and a signal 123 is an output signal of the self-position identification device 100 and self It is a position identification result. The self-location identification apparatus 100 according to the first embodiment includes a computer having an external input / output interface and a plurality of tasks that operate inside the computer. Each task periodically executes the processes described below. Each component will be described below.

The position estimation unit 101 executes a process of calculating an estimated position in the environment of the moving body and outputting the result as a self-position identification result based on the observation result 122 of the internal sensor that measures the operating state of the moving body every cycle. Here, the internal sensor refers to a sensor for measuring information inside the moving body, such as a posture angle sensor or a wheel rotation speed sensor. The internal sensor in the first embodiment is a rotary encoder installed so as to be able to measure the rotation amount of the moving body wheel. The estimated position by the internal sensor gradually deviates from the true value due to sensor noise and wheel slipping that cannot be measured. Therefore, when the position estimation unit 101 receives a corrected self-position identification result from an estimated position correction unit 106 described later, the position estimation unit 101 replaces the self-position identification result with the corrected self-position identification result.
The position estimation unit 101 according to the first embodiment is an odometry position estimation task for estimating (odometry) a moving body position using the encoder signal and a kinematic relational expression specific to the moving body in the computer.

The landmark candidate extraction unit 102 executes a process of extracting a landmark candidate in the environment using the external sensor observation result 122, specifying a position with respect to the moving object, and sending the candidate to the map construction unit 103 every cycle.
Here, the external sensor refers to a sensor for measuring environmental information outside the moving body, such as an image sensor or a laser range finder. The external sensor in the first embodiment is a two-dimensional laser range finder.
A landmark candidate is an artificially installed mark such as a plate or reflector with a characteristic pattern, or an object or point in the environment that has a characteristic shape such as a corner or a plate of the object. This is extracted from the sensor observation result 122. The landmark candidate in the first embodiment is a corner (corner) having an angle within a predetermined range existing in an object in the environment.
The landmark candidate extraction unit according to the first embodiment is an angle extraction task for extracting a corner having an angle within a predetermined range from the observation result of the two-dimensional laser range finder in the computer and specifying a position with respect to the moving object. Here, the angle refers to an angle formed by the intersection of two line segments having a length equal to or greater than a predetermined value on the measurement surface of the two-dimensional laser range finder, such as an apex of a prism or a corner of a passage intersection. . In the present embodiment, the position information and the angle information are used by using the corner extracted in each cycle as a landmark candidate.
A landmark refers to a landmark candidate whose information is registered and stored in the global map storage unit 105 for use in self-location identification. A method for selecting landmark candidates to be stored as landmarks will be described later.
Note that the processing cycle of the landmark candidate extraction unit 102 does not need to be synchronized with the processing cycle of the position estimation unit 101.

The map construction unit 103 constructs and stores a new local map from the current landmark candidate information, which is the landmark candidate information extracted by the landmark candidate extraction unit 102, and the local map up to the previous cycle. The process of constructing and saving a new global map from the local map obtained in the previous step, the global map up to the previous cycle and the latest position estimation result is executed every cycle. Processing in the cycle of the map construction unit 103 will be described later.
Here, the local map is a local map of the surrounding environment measured by the external sensor,
-Previous landmark candidate information which is information of landmark candidates extracted by the landmark candidate extraction unit 102 in the previous cycle,
Current landmark candidate information, which is information of landmark candidates extracted by the landmark candidate extraction unit 102 in the current cycle,
The number of matching matches, which is the number of times that the previous landmark candidate and current landmark candidate match continuously,
-Previous Landmark Candidate-Landmark Correspondence Relationship that indicates which Landmark Candidate corresponds to which Landmark,
-Current landmark candidate-landmark correspondence relationship indicating which current landmark candidate corresponds to which landmark,
including.
The global map is a global map that covers the entire moving environment of the moving object to be used by the estimated position correction unit 106, and is an aggregate of the landmark information. The global map is the result of mobile location identification,
・ Landmark position identification result,
・ Landmark observation results,
including. The global map is constructed by dividing the environment into a plurality of small areas, and landmark information (landmark position identification results, landmark observation results) of the landmarks existing in the areas is Are saved together.
The processing cycle of the map construction unit 103 is started simultaneously with the end of the processing cycle of the landmark candidate extraction unit 102.
The map construction unit 103 in the first embodiment is a map construction task inside the computer. The map construction task starts its processing upon completion of the corner extraction task.

The process flow of the cycle in the map construction unit 103 will be described with reference to FIG. Details of the processing will be described later. FIG. 2 shows processing executed for each current landmark candidate, which is the latest landmark candidate extracted by the landmark candidate extraction unit 102.
In FIG.
S201 is a step of starting processing;
S202 stores the current landmark candidate information in a local map;
S203 branches the subsequent processing depending on the result of collating the current landmark candidate with the previous landmark candidate stored in the local map;
S204 is a step of branching the subsequent processing depending on the result of referring to the previous landmark candidate-landmark correspondence relationship of the previous landmark candidate corresponding to the current landmark candidate;
S205 determines that the landmark corresponding to the previous landmark candidate also corresponds to the current landmark candidate, updates the current landmark candidate-landmark correspondence relationship, and stores it in the local map;
S206 is a step of storing the current landmark candidate information in the global map as an observation result of the corresponding existing landmark;
S207 is a step of setting the number of matching matches, which is the number of times the current landmark candidate matches the matching result with the previous landmark candidate, as a value obtained by adding 1 to the number of matching matches of the previous landmark candidate;
S208 is a step of additionally storing the number of matching matches as part of the current landmark candidate information in a local map;
S209 is a step of branching the subsequent processing according to a result of comparing the number of matching matches with a preset threshold value;
S210 is a step of branching the subsequent processing according to the result of collating the current landmark with the existing landmark registered in the global map,
S211 is a step of additionally registering the current landmark candidate information in the global map as an observation result of the new landmark;
S212 is a step of initializing the number of matching matches of the current landmark candidate to zero,
S213 is a step of completing the map construction process.

Each step of one cycle of the map construction process described above will be described in more detail.
The information on the current landmark candidate extracted by the landmark candidate extraction unit 102 is stored in the local map in step S202. At this time, the estimated global position of the current landmark candidate is calculated from the estimated position of the moving object in the global map and the current landmark candidate information, and is also added to the local map as the current landmark candidate estimated position. save.

  Next, in step S203, the matching relationship between the current landmark candidate and the previous landmark candidate is collated. The collation is performed by the following method. That is, the previous landmark candidate estimated position stored in the local map in step S203 in the previous cycle is compared with the current landmark candidate estimated position, and the distance between the two is less than a predetermined threshold. It is assumed that the mark candidate having the closest estimated position matches the current landmark candidate. In this step, the current landmark candidate is not suddenly checked against all the landmarks in the global map. First, in the local map, the current landmark candidate is the previous landmark saved in the previous cycle. Check which of the mark candidates matches. By doing so, the target of collation is limited to the previous landmark candidates extracted in the previous cycle, so that the processing can be completed with a much smaller calculation amount than when collating directly with all the landmarks in the global map. As a result of the collation, if there is a previous landmark candidate that matches the current landmark candidate, the process proceeds to step S204, and if not, the process proceeds to step S212.

  Step S204 is a step of checking the match or mismatch between the current landmark candidate and the landmark in the global map using the previous landmark candidate-landmark correspondence relationship stored in the local map. Here, the previous landmark candidate-landmark correspondence relationship indicates which previous landmark candidate matches which landmark as described above, and is stored in the local map in the previous cycle. It is a corresponding relationship. Since the current landmark candidate that has proceeded to step S204 has a matching previous landmark candidate, the current landmark candidate-landmark correspondence relationship is referred to by referring to the correspondence relationship with the landmark of the previous landmark candidate. A part of the search process for checking That is, if the corresponding previous landmark candidate corresponds to any landmark in the global map, it is determined that the current landmark candidate also corresponds to the landmark, and the search process is omitted. Proceed to step S205. If the corresponding previous landmark candidate does not correspond to any landmark, the process proceeds to step S207 for search processing.

  In step S205, the current landmark candidate-landmark correspondence relationship determined to match in step S204 is stored in the local map, and the process proceeds to step S206.

In step S206, first, the current landmark candidate information is stored in the global map as the corresponding landmark observation result. Next, the landmark observation result is calculated using the recent (newest) moving object position identification result and the landmark position identification result (the result of identifying the current landmark candidate and the corresponding landmark). Calculated as the estimated observation result estimated above, and compares this predicted observation result estimated value with the actually measured landmark observation result to calculate the amount of divergence between them and reduce this Thus, the moving body position identification result and the landmark position identification result are corrected. At this time, the position of the moving object and the landmark position are regarded as the state quantities in the state equation, and the landmark observation results are regarded as outputs, and the estimated state quantities are calculated for each cycle by using the deviation between the estimated observation results and the actual observation results. Although it is close to the true value by calculation, a specific correction calculation method is not described here because it is known from the following document, for example.
MWMGamini Dissanayake et al, "A Solution to the Simultaneous Localization and Map Building (SLAM) Problem", IEEE Trans. R & A, Vol. 17, No. 3, 2001
Finally, the corrected moving body position identification result and landmark position identification result are stored in the global map, and the process is completed. It should be noted that the landmark observation results used for correcting the identification results should be as large as possible. This is because each observation result includes an error due to sensor noise or the like, and therefore the influence of the error can be reduced by using a plurality of landmark information.

  In step S207, when the current landmark candidate matches the previous landmark candidate, but the previous landmark candidate does not match the landmark of the global map (previous landmark candidate). -When there is no correspondence with reference to the landmark correspondence), the number of matching matches of the current landmark candidate is added, and the process proceeds to step S208.

  In step S208, the number of matching matches obtained in step S207 is stored in the local map, and the process proceeds to step S209.

In step S209, the number of matching matches obtained in step S207 is compared with the threshold value. If the threshold value is equal to or greater than the threshold value, the process proceeds to step S210.
The current landmark candidate that proceeds to step S210 has been continuously observed for the threshold number of times or more. By setting the threshold value, it is possible to ignore erroneous landmark candidate information that the landmark candidate extraction unit 102 accidentally outputs due to the influence of sensor noise or the like. As a result, the number of landmark candidates to be collated with the landmarks in step S210 can be minimized and the amount of calculation can be saved.

In step S210, when the current landmark candidate matches the previous landmark candidate, but there is no corresponding information that matches the previous landmark candidate with the landmark of the global map, In this step, it is determined whether or not the candidate matches the landmark of the global map.
The purpose of this step is to confirm whether or not there is a match with an existing landmark in the global map with respect to the current landmark candidate observed continuously for the threshold number of times.
Matching / non-matching is performed according to the following procedure. First, the current landmark candidate position estimated value stored in the local map in step S202 is read. Next, the small area in the global map where the current landmark candidate position estimated value exists is specified. After specifying the small area, the landmark position estimated values of the landmarks existing in the small area are sequentially read, and the distance from the landmark candidate estimated position is calculated. Then, it is assumed that the landmark having the closest position estimation value among the landmarks whose calculated distance is less than the predetermined threshold matches the current landmark candidate. As a result of collation, if there is a landmark that matches the current landmark candidate, the process proceeds to step S205, and if not, the process proceeds to step S211.

  In step S211, the current landmark candidate is registered in the global map as a new landmark, and the process proceeds to step S205. The initial value of the new landmark information is registered as a new landmark global position estimation result calculated based on the latest moving object position estimated value and the current landmark candidate information.

  In step S212, the number of matching matches of the current landmark candidate is initialized to zero and stored in the local map, and the process is completed.

  According to the above procedure, when a new landmark candidate appears, a landmark can be additionally registered in the global map, and a global map that more accurately reflects the actual situation of the environment can be constructed. Since landmarks that can be used for self-position identification increase, the self-position identification value can be obtained more accurately in the estimated position correction unit 106 described later.

On the other hand, landmarks that originally existed may not be observed due to environmental changes. In this case, landmark information that does not already exist is kept in the global map, which not only consumes a storage area but also increases the amount of calculation required for collation. In order to prevent this, the following processing may be added.
First, in step S206, the observation time is stored in the global map together with the landmark observation result. Further, the following processing is added to the subsequent stage of the processing of the map construction unit 103. That is, the observation time of the landmark in the global map is checked, and the landmark that has passed the specified time from the current time is deleted. The specified time is set to a sufficiently large value so that necessary landmark information is not removed. When the moving body circulates in the same environment, the specified time may be set to a value larger than the time required for one lap.

The estimated position correcting unit 106 corrects the moving body position identification result output from the position estimating unit 101. The value obtained by odometry in the position estimation unit 101 causes an error due to the difference between the kinematic relational expression to be used and the actual value. It cancels out.
The estimated position correction unit 106 in the present embodiment is an estimated position correction routine that is called later in the process of the map construction unit 103, reads the latest mobile body position identification result from the global map, and outputs the result to the position estimation unit 101. . The position estimation unit 101 receives the moving body position identification result and at the same time discards the previous value and replaces it with the received value.

  Since the above configuration is adopted, the self-position identification result 123, which is the output of the self-position identification device according to the present invention, is corrected based on the external sensor observation result 121, and therefore is estimated using only the internal sensor observation result 122. Will be closer to the true value than Unlike the conventional self-position identification apparatus (for example, Patent Documents 1 to 3), the present invention newly registers landmark candidates that are continuously observed in the landmark candidate extraction unit 102 as specified landmarks. By removing landmark information that is not observed over time, the global map can be flexibly adapted to changes in the environment. Also, by using the matching result of the previous landmark candidate, the number of direct matching between the current landmark candidate and the landmark of the global map is reduced as much as possible, and the calculation amount is greatly reduced.

  The global map can be flexibly adapted to changes in the environment, and the amount of computation required for landmark matching for self-location identification can be greatly reduced. It can be widely applied to all moving objects that move while identifying their own positions.

Configuration diagram of self-position identification device according to the present invention Flow chart of processing in map construction unit of self-position identification device according to the present invention Block diagram showing the configuration of a self-localization system in the prior art Block diagram showing the configuration of the environment identification device in the prior art

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Self-position identification apparatus 101 Position estimation part 102 Landmark candidate extraction part 103 Map construction part 104 Local map preservation | save part 105 Global map preservation | save part 106 Estimated position correction | amendment part 121 External world sensor measurement result 122 Internal world sensor measurement result 123 Self-position identification result 301 Landmark recognition unit 302 Observation map creation unit 303 Travel amount calculation unit 305 Map data storage unit 307 Map data management unit 308 Verification unit 401 Environmental map construction unit 402 Environmental map storage unit 403 Environment identification unit 404 Environment search unit

Claims (9)

In a moving body that comprises a self-position identification device that observes landmarks in the environment and identifies its own position, and moves while identifying the self-position in the environment by the self-position identification device,
The self-position identification device is
A landmark candidate extraction unit that extracts landmark candidates from observation results of an external sensor that observes the landmarks around the moving body;
A map construction unit for constructing a local map containing information on the landmark candidates and a global map containing information on the landmarks in the entire area in the environment;
A local map storage unit for storing the local map;
A global map storage unit for storing the global map;
An estimated position correction unit that calculates the latest moving object position estimated value of the moving object from the observation result of the external sensor and the information of the global map;
A position estimation unit that outputs a self-position identification result in the environment from the information obtained from the internal sensor of the mobile object and the mobile object position estimation value;
When the map construction unit constructs the global map, it judges the coincidence relationship between the landmark candidates and the registered landmarks of the global map. A mobile object equipped with a self-position identification device that performs a process of additionally registering as a landmark in the global map.   In constructing the global map, the map construction unit constructs the global map by dividing it into a plurality of small areas, and stores the landmark candidate information as landmark observation results for each of the small areas. A moving body comprising the self-position identification device according to claim 1.   When the map construction unit determines that the landmark candidate matches the registered landmark according to the determination, the map construction unit includes the landmark candidate and the registered landmark of the global map in the local map. The mobile object having the self-position identification device according to claim 1, wherein a process of storing the landmark candidates including correspondence information is performed.   Before making the determination, the map construction unit first collates the landmark candidate with the previous landmark candidate saved in the local map in the previous process in the local map, and the collation The mobile object having the self-position identification device according to claim 3, wherein the determination is performed only for the landmark candidate that has a matching relationship with the previous landmark candidate.   In making the determination, the map construction unit refers to the correspondence information of the previous landmark candidate obtained in the matching with the landmark candidate, and the registration having the matching relationship with the previous landmark candidate 5. The moving body having a self-position identification device according to claim 4, wherein a completed landmark is determined to coincide with the landmark candidate.   In the collation, the map construction unit obtains the matching relationship between the landmark candidate and the previous landmark candidate in the collation, but in the reference, the previous landmark candidate corresponds to the registered landmark. When the information is not owned, the number of matching matches is added, and it is determined whether there is a match between the landmark candidate whose registered matching count is equal to or greater than a threshold and the registered landmark. A moving body comprising the self-position identification device according to claim 5.   In the determination, the map construction unit identifies the small area where the landmark candidate exists, sequentially reads information on the registered landmarks existing in the small area, and distance from the landmark candidate Are calculated, and among the registered landmarks whose calculated distance is less than a threshold, it is determined that the nearest registered landmark coincides with the landmark candidate. A moving body comprising the self-position identification device according to claim 6.   The map construction unit stores the observation time together with the landmark observation result in the global map, and deletes the registered landmark that has passed a specified time from the current time from the global map. A moving body comprising the self-position identification device according to claim 2.   9. The self-position identification device according to claim 8, wherein the specified time is set to a value larger than a time required for one round of the lap when the moving body circulates in the environment. Moving body.

Images