JP2009163156A - Moving robot system and control method therefor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce a storage volume of a map, to quickly access to a map database in estimation, and to reduce a processing time for data preparation in accompaniment to moving of a robot, compatibly at the same time, in a moving robot system moving while estimating own positional posture by collating a sensing data with the map. <P>SOLUTION: This moving robot system is provided with an entire map storage means 160 for managing an entire map, a partial map storage means 170 for managing continuously map areas required for estimating the positional posture as partial maps, and a partial map updating means 150 for updating a content of the partial map storage means 170. The partial map updating means 150 selects and extends only a block not written yet on the partial map storage means, out of the map areas required for estimating the positional posture of the robot, based on a positional posture estimation initial value of the robot, and updates the stored content to make the spatially continuous areas get to continuous areas on a memory. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an autonomous mobile robot that can automatically reach a destination without human operation. More specifically, it estimates the position and orientation of itself using sensing data measured by a distance sensor mounted on the mobile robot and a two-dimensional environment map showing the geometric state of the environment, and autonomous movement based on the result The present invention relates to a robot system that performs control.

  First, a method for estimating the position and orientation of a mobile robot using a distance sensor and a two-dimensional environment map will be described.

  Here, the distance sensor can measure the distance to an obstacle in a plurality of directions around the distance sensor. Various specific specifications can be considered, but here, as shown in FIG. 2, 1 ° in a direction from −90 ° to + 90 ° with respect to the front of the distance sensor 210 on a horizontal plane at a certain height. Assume that the distance data to the obstacle 230 can be instantaneously measured for a total of 181 different directions for each. The distance sensor 210 is assumed to be mounted on the mobile robot 220.

FIG. 3 shows an example of sensing data measured using the distance sensor 210. Each sensing data is represented as (d ₀ , φ ₀ ), (d ₁ , φ ₁ ),..., (D ₁₈₀ , φ ₁₈₀ ) as a combination of the angle φ and the distance d. Here, the numerical suffix i of each variable is a data number, φ _i indicates the i-th measurement direction, and d _i indicates the distance to the obstacle in the φ _i direction. The measurement direction is a clockwise direction, indicating the angle formed with the main direction of the distance sensor, and data numbers are assigned in ascending order of the angle formed with the main direction. That is, φ ₀ = −90 °, φ ₁ = −89 °,..., Φ ₁₈₀ = + 90.0 °.
Here, considering the ability to measure the distance sensor, the measurable distance assumes greater 16.0m follows from 0.0 m, the range of values of d _i, and be limited to 0.0 m <d _i ≦ 16.0m . In addition, the 181 distance data are collectively referred to as sensing data 300 for one frame.

  On the other hand, the two-dimensional environment map is a digital image as shown in FIG. 4 showing the presence of an obstacle on the cross section measured by the distance sensor 210. Here, as shown in FIG. 4, a place 410 where no obstacle is present is represented as a pixel value indicating white, and a place 420 where the obstacle is present is represented as a pixel value indicating black.

  It is assumed that the mobile robot 220 moves on a flat ground. Therefore, the degree of freedom relating to the position and orientation that changes due to the movement is a total of three parameters, that is, two parameters (x, y) for the position and one parameter θ for the orientation angle that the mobile robot faces. It is assumed that the coordinate system representing these parameters matches the coordinate system representing the environment map 510 as shown in FIG.

  Estimating the position and orientation of the mobile robot means obtaining the three parameters (x, y, θ) from the sensing data of the mobile robot in an arbitrary position and orientation. Therefore, the functions targeted by the present invention are the position and orientation parameters (x, y, θ) of a mobile robot in an arbitrary position and orientation, sensing data from a distance sensor mounted on the mobile robot, and an environment prepared in advance. It will be obtained using a map.

  Several methods are conceivable as means for realizing this function. Here, a method based on matching processing, which is the most basic method, will be described.

  The matching method is a method of superimposing the sensing data on the environment map while changing the position and orientation (direction) of the entire sensing data measured in a certain frame, and the position and orientation that best matches the sensing data and the environment map. This is used as an estimated value of the position and orientation parameter of the mobile robot. The principle is based on the idea that when the mobile robot is in its position and posture, the same data as the measured data will be obtained with the highest probability.

  Specifically, such a method by matching processing is performed by processing as described below.

  (1) For all the sensing data in a certain frame (181 in the example of the distance sensor described above), this is rotated by θ on the map, and the position translated by (x, y) is obtained.

  (2) Refer to the pixel values on the environmental map at those positions, count the number of obstacles, and use this as the evaluation value for the parameter (x, y, θ).

  (3) For the combination of (x, y, θ) with all possible values, the solution (1) and (2) above is the solution with the highest evaluation value (x, y, θ). .

  FIG. 5 shows an outline of matching processing based on this concept. When the environment map 510 and the sensing data 300 are given, the situation determined to be the best match by this method is the situation shown on the right side of FIG. 5, and the translation component (x, y) and rotation at this time The component θ is the estimated position and orientation of the mobile robot.

  The mobile robot 220 sequentially obtains its own position and orientation in this way, determines the moving direction, moving speed, and the like by comparing the value with the route that it should take, and based on this, the moving mechanism of the mobile robot 220 The target movement is realized by controlling the unit. Here, in order to perform the movement accurately, it is necessary to perform the movement control in a cycle as short as possible, so that the position / orientation estimation process is also required to be performed at high speed. Specifically, it is desired to have a high speed capable of sequentially performing several tens of milliseconds to 100 milliseconds, that is, several tens of times per second.

It is generally considered that the position / orientation estimation process is realized by a combination of a central processing unit typified by a CPU, a main storage unit typified by a memory, and an auxiliary storage unit typified by a hard disk. Here, in order to perform the position and orientation estimation processing at high speed, it is necessary to store frequently accessed data, that is, the environment map 510 on a main memory such as a memory. However, assuming that the environment map used here represents the distance of 1 cm in the real environment as 1 pixel, the area that is the target of the real environment is, for example, a 300 m x 300 m area with a data size of 900 MB. If it becomes large, there is a possibility that the memory will not be able to be stored in a memory generally used at present. Also, when this is stored in an auxiliary storage device such as a hard disk, it is desirable to reduce the size as much as possible in consideration of the read / write processing time and the case of managing a plurality of these.
Therefore, it is conceivable to apply a method in which the environment map is compressed and stored in the auxiliary memory, and only the necessary area is expanded and expanded in the main memory for use when estimating the position and orientation. Note that processing time for decompression and expansion is newly required here. However, as described above, if the overall position / orientation estimation process is performed from several tens of milliseconds to 100 milliseconds, this process may take a long time from several milliseconds. It needs to be done in about 20msec.

  On the other hand, the amount of data to be read into the main memory, for example, because the sensor sensing range is 16m, the map area required for position estimation is about (16m × 2) × (16m × 2), This is about 10MB in size as uncompressed data. Considering the performance of current general CPUs, etc., all processing such as reading from the external storage device, decompression processing, writing to the main storage device, etc. is performed on the compressed data that becomes this size after decompression. It is extremely difficult to perform in time, and even if possible, this process is preferably as small as possible for the purpose of allocating computing resources to other processes.

  Therefore, the present invention provides a method for reducing the amount of processing required for position and orientation estimation when map management is performed using an auxiliary storage device and a main storage device together.

  As a publicly known technique related to the above-mentioned problem, in “JP-A-2006-39486 Map Information Processing Device”, a map image formed along a user's route is simply extracted as a small map, and is not interrupted along the user's route. A map information processing apparatus that can be displayed on a portable device carried by a user is presented. However, the present method is intended to display map data on a screen, and does not present a function for efficiently developing on a single memory.

In the "Japanese Patent Laid-Open No. 9-160482 navigation apparatus", an image obtained by bird's-eye photography of the ground is divided and compressed, and an image of an appropriate area is expanded and displayed based on the current position of the vehicle. In this method, a method is presented in which an image is divided and compressed, an appropriate one is selected from the divided and compressed images as the vehicle moves, and the image is efficiently stored and displayed in a buffer memory. However, with this method, since the buffer memory is not continuous, in order to display this as a continuous map, a separate process for transferring the buffer memory from the buffer memory to the display memory in an appropriate order is newly performed. This causes a problem that the processing time is excessive. That is, in such a method using a cache memory or a buffer memory, a continuous area as a map is not necessarily a continuous area in the memory space. In this case, there is a problem that the address calculation for memory access which is performed in large quantities at the time of estimation becomes complicated, resulting in a long processing time.
That is, in the present invention, the capacity to be stored in the storage device is reduced, and further, high-speed access to the map data when performing the position / orientation estimation processing based on matching is enabled, and the auxiliary storage device and the main storage device are used in combination. To solve the problem that map data processing must be performed at the same time.

JP 2006-39486 A JP-A-9-160482

  In order to solve the above-mentioned problem, the present invention solves the above problems in a mobile robot system that recognizes its position and orientation by comparing sensing data obtained by measuring a surrounding situation with a distance sensor and a map, and moves by movement control based on the recognition result. Then, the entire map storage means for managing the entire map, the partial map storage means for continuously managing the map area necessary for position and orientation estimation as a partial map, and the map data are selectively read from the entire map management means. And a partial map update means for updating the contents of the partial map storage means.

  By managing the map separately into the main memory and the auxiliary memory, even if the map is enlarged, it is possible to perform high-speed position and orientation estimation processing using only the limited main memory. By compressing and saving the entire map in the auxiliary memory, it is possible to reduce the amount of data to be stored and the reading processing time. In addition, the entire map is divided into a plurality of blocks and stored in the auxiliary memory, and only necessary blocks are selectively expanded and written on the main memory, so that data on the main memory can be prepared at high speed. it can.

  Embodiments of the present invention will be described below.

FIG. 1 is a functional block configuration diagram showing the overall configuration of a mobile robot system according to an embodiment of the present invention.
The mobile robot system calculates a moving mechanism unit 130 that can move in an arbitrary direction in a real environment having a flat floor by wheels and the like, a direction, a distance, a speed, and the like to be moved from a route and a position and orientation parameter of the robot. Based on the result, the movement control unit 120 that controls the movement mechanism 130, the distance sensor 210 that measures the environmental geometrical situation around the robot, and the matching of the position and direction of the robot with the map and sensor data described in the background section The position / orientation estimation unit 110 obtained by the above, the partial map storage unit 170 that stores the partial map 190 used when performing the position / orientation estimation, and the entire environment map 180 are divided into a plurality of blocks, and each is compressed and stored. Block area required to construct a partial map 190 from the entire map storage unit 160 and the entire map 180 Read by selecting, consisting of partial map update unit 150 to update the contents of the partial map 190. The partial map update unit 150 decompresses a read block selection instruction unit 152 that selects and instructs a block to be read, and a compressed block map 183 that has been selected and read, and uses this as an update block map 186 as a partial map storage unit 170. Is composed of a partial map data decompression unit 154 to be written in

  Further, it is assumed that the entire map storage unit 160 is implemented by an external storage device such as a hard disk from the viewpoint of managing a huge amount of data, and the partial map storage unit 170 is used for position and orientation estimation. It is assumed that it is mounted by a main storage device such as a memory for the purpose of speeding up access to data during calculation. Further, it is assumed that each process such as the selective reading process, the expansion process, and the position / orientation estimation process described above is performed by a program that operates a central processing unit such as a CPU.

A method for managing the entire map and the partial map in the above system configuration will be described. Here, in order to make the explanation easy to understand, the following explanation will be made using specific numerical values as an example. However, this numerical value does not limit the application of the present invention to this value.
First, the overall map 180 will be described. Here, a rectangular area of 400 cm × 400 cm is defined as one block image, and the whole is represented as a set of block images. Note that 1 cm in the actual environment is represented by 1 pixel on the image. It is assumed that each block image is compressed and stored on the hard disk. The compression method here uses a reversible compression method that does not change the contents even if compression and decompression are repeated. As the specific method, various methods such as a run length method can be considered. However, since the content of the method is irrelevant to the essence of the present invention, a detailed description thereof is omitted, and compression currently available generally. Assume that you use a program. It is also assumed that a decompression program for this compression method is also available in this system. These block images are allowed to be randomly accessed with respect to the block images arbitrarily designated by the program, for example, by storing them as different files.

  Next, the partial map 190 will be described. In the matching process in position and orientation estimation, uncompressed map data on the memory is accessed. The map data developed on this memory is the partial map 190 in the present invention. The partial map 190 only needs to be within a range necessary for the robot to estimate the position and orientation, and does not need to include the entire map 180 described above. That is, since it is a part of the entire map, the name “partial map” is used.

  In explaining which part of the entire map 180 the partial map 190 represents with reference to FIG. 6, the concept of a necessary partial map 620 will be introduced. The area that can be measured by the distance sensor 210 mounted on the mobile robot 220 is limited to a certain area around the candidate position 610 where the mobile robot will currently exist. That is, if there is a map in this area, it is sufficient for estimating the position and orientation of the mobile robot, and conversely, even if there is map data in a larger range, it will be wasted.

In determining this area, first, the possible range of the mobile robot 220 is considered. For example, assuming that the moving speed of the mobile robot is 7.2 km / h and the frame period of the position and orientation estimation described above is every 100 ms, the moving distance per period is 720000 cm / 60/60/1000 * 100 = 20 cm. Therefore, the position of the mobile robot 220 in the current frame is within a range of ± 20 cm both vertically and horizontally with respect to the position of the mobile robot 220 in the previous frame. Therefore, using the estimated position value of the frame as the position of the previous frame, and assuming that the error of the estimated value is about 5 cm at the maximum, the position of the mobile robot 220 of the current frame is within the range of ± 25 cm. Assuming that From this, it is defined that the possible range of the mobile robot 220 is within a rectangular range of ± 25 cm with respect to the position estimate value of the previous frame. The previous frame position estimation value is referred to as the candidate position 610 of the mobile robot 220 in FIG. The method of determining this value is not limited to using the position estimation value of the previous frame, but a method of calculating / predicting this from the moving speed of the robot, a method of estimating this from the wheel rotation information, etc. Is also possible.
In determining the area of the necessary partial map 620, the measurable distance of the distance sensor 210 is considered next. Since the maximum distance measurable by the distance sensor 210 is 1600 cm, the map area that can be used for position and orientation estimation is limited to a range of ± 1600 cm with respect to the position of the mobile robot 220.

  As described above, the area of the map used for position and orientation estimation is limited depending on the possible range of the mobile robot 220 and the measurable distance of the distance sensor 210. A map in the range of (1600 cm × 2 + 25 cm × 2) × (1600 cm × 2 + 25 cm × 2) = 3300 cm × 3300 cm centering on the position 610 is sufficient for position and orientation estimation.

  Considering the above, as shown in FIG. 6, in this embodiment, an area having a width of 4 blocks is defined as a necessary partial map 620 around the block including the candidate position 610 of the mobile robot. I will decide. Specifically, it is a 9 × 9 block, and since the size of one block is 400 cm × 400 cm, the entire area is 3600 cm × 3600 cm. Although it is a value with a slight allowance for the size of 3300 cm × 3300 cm described above, this can be adjusted as appropriate. If the necessary partial map 620 is prepared, the map area necessary for position estimation is always included regardless of which direction the mobile robot 220 faces.

  A method of preparing the partial map 190 using the concept of the necessary partial map 620 described above and first preparing the partial map 190 in FIG. 1 as the same as the necessary partial map 620 will be described with reference to FIGS. This will be described with reference to FIG. The basic idea at this time is that the necessary partial map 620 is always written in the partial map 190 without excess or deficiency. That is, the map data of the area of the necessary partial map 620 in FIG. 6 is stored in the memory as the partial map 190 in FIG. 1 or 7 as uncompressed data.

  This processing is performed by the partial map update unit 150 in FIG. 1, and the processing content will be described with reference to FIG. Here, the processing of the portion 860 surrounded by the dotted frame is processing performed by the partial map update unit 150.

  First, the candidate position 610 of the mobile robot is set as a position estimation value of the previous frame (820). When there is no estimated value of the previous frame at the beginning of the first movement, this is uniquely determined by setting a condition such that a person designates this or starts moving from a predetermined location.

Next, an area of the necessary partial map 620 is calculated from the candidate position 610 (825). The calculation method is as described in the definition of the necessary partial map described above.
Next, the entire map storage unit 160, that is, the hard disk is stored so that the content written in the partial map storage unit 170, that is, the partial map 190 on the memory, and the area indicated by the necessary partial map 620 are equal. A block of an insufficient area is selected from the entire map 180 (830).

  The read block selection instruction unit 152 in the partial map update unit 150 performs the above processing.

  Next, the data of the selected block is read (835), the expansion process is performed (840), and this is written in the partial map 190 (845). At this time, there is a problem in which part on the memory as an entity for storing the partial map, which will be described later.

  The read data decompression unit 154 in the partial map update unit 150 performs the above processing.

When this writing is completed, the partial map 190 necessary for the position / orientation estimation unit 110 to perform position / orientation estimation in FIG. 1 is prepared in the memory. Therefore, position / orientation processing and mobile robots using the partial map 190 are prepared. 220 movement control is performed (850).
A series of these processes is set as one cycle process (one frame process), and the movement control of the mobile robot 220 is performed by repeating this process.

  That is, the partial map update means selects only the blocks that have not yet been written on the partial map storage means from the map area necessary for the robot position and orientation estimation based on the initial position and orientation estimation values of the robot. It expands and updates the stored contents so that the space continuous area becomes a continuous area on the memory. Thus, access to necessary map data when performing the position / orientation estimation process is performed at high speed. Next, with reference to FIG. 9, a description will be given of which region of the entire map 180 is expanded as a partial map 190 on the memory as the mobile robot 220 moves.

  First, when the candidate position 610 of the mobile robot 220 is the position shown in FIG. 9, the area read in the memory as the necessary partial map 620 in the entire map 180 is as shown in the upper diagram of FIG. 9. It is confirmed that this area conforms to the definition of the necessary partial map described above. Even if this robot candidate position 610 moves with the movement of the robot, as long as it is within the block 900 in the figure, the area of the necessary partial map does not change and is thus expanded in the memory. There is no need to change the partial map. However, as shown in FIG. 9, when the robot's candidate position 610 moves across the block to a point 611 inside the block 903 as shown in FIG. 9, the area of the necessary partial map 620 is shown in the lower diagram of FIG. Changes to things. At this time, the area 980 indicated by hatching in the lower diagram of FIG. 9 is unnecessary from the memory, and the area 970 indicated by hatching is newly required on the memory, so that the area 980 is replaced with the area 970 in the memory. The contents will be updated.

  As described above, by updating the contents of the partial map 190 so as to be the same as the contents of the necessary partial map 620, the position / orientation estimation process can be performed using only a limited amount of memory. Become.

  However, with this method, the following problems occur when the robot moves near the block boundary. That is, when the frequency that the robot moves across blocks increases, the area of the necessary partial map 620 changes each time, the contents of the partial map 190 are updated, and the memory rewriting process frequently occurs. The problem is that the amount of processing for map reading, data compression / decompression, and memory writing increases.

  Therefore, in the present invention, the partial map 190 is configured not to include only the area indicated by the necessary partial map 620 but to include an additional area called an extended area. This will be referred to as a technique including an extended region, and will be described in detail with reference to FIG.

  In the method including the extended region, as shown in FIG. 10, a portion 770 surrounded by a dotted line in the entire map 180 is expanded on a memory as a partial map 190 as shown in FIG. The contents at this time are obtained by adding a region 767 for one vertical column and one horizontal column in addition to the necessary partial map 620 described in FIG. 6, and the memory stores data for this region. Prepare the size.

At this time, the area of the necessary partial map 620 is always written in the partial map 190 shown in FIG. This is performed by the partial map update unit 150 in FIG. 1, but the processing content is basically the same as that described in FIG.
However, in the method including the extended area, the frequency of updating the contents of the partial map 190 does not increase even when the robot moves more frequently across the blocks due to the addition of the extended area 767. It is possible to solve the problem that the processing amount of reading the entire map, compressing / decompressing data, and writing to the memory increases. The principle will be described with reference to FIG.

  First, when the candidate position 610 of the mobile robot 220 is the position shown in FIG. 12, the area read in the partial map memory in the entire map 180 including the expansion area 767 added to the necessary partial map 620 is shown in the upper diagram of FIG. 12. It is as the area | region 770 shown with the dotted line. As the mobile robot 220 moves, as shown in the lower diagram of FIG. 12, when the candidate position 610 of this robot moves to a point 611 across the block, the area of the necessary partial map 620 changes to that shown in the lower diagram of FIG. To do. At this time, the area 912 indicated by hatching in the lower diagram of FIG. 12 is unnecessary from the memory, and the area 915 indicated by hatching is newly required on the memory, so that the area 915 is replaced with the area 912. The contents will be updated. The processing here is almost the same as the method described in FIGS.

  However, as shown in FIG. 13, when the candidate position 611 of the robot changes to a position 612 across the blocks as the mobile robot 220 moves, the method of adding this extended area shows an effect. By this movement, the area of the necessary partial map 620 changes from that shown in the upper diagram of FIG. 13 to that shown in the lower diagram of FIG. However, due to the presence of the extended area 767, all areas of the necessary partial map 620 after the change have already been written in the memory as the dotted line part 770, that is, the partial map 190. That is, there is no need to update the partial map. This is to solve the problem of increasing the amount of processing of body map reading, data compression / decompression, and memory writing. It is obvious that this effect appears even when the mobile robot 220 repeatedly moves between the block including the position 611 and the block including the position 612 shown in the number of FIG. Note that it can be easily understood that the above-described effects appear not only when the two blocks described in FIG. 13 are straddled up and down, but also when they are straddled left and right.

  Next, how to manage the partial map 190 described above on the memory will be described in detail. It is assumed that the memory is managed by a combination of the address number and data written to the address.

  The memory is prepared as a continuous memory having a storage capacity of 10 × 10 blocks of pixels, that is, (40 × 10) × (40 × 10) = 4000 × 4000 pixels. The memory number starts from the top left and increases to the right.When the right end is reached, the number increases from the left end of the next lower row to the right again. This is the same as the general method of holding image data with the bottom end.

  That is, the partial map storage means is realized by a main storage device such as a memory, and this memory is (M × W) × (N × H) pixels or ((M + 1) × W) × ((N + 1) XH) The pixel has a size capable of storing pixel data, and the partial map is allocated on the memory so that the right end of the rectangular area is continuous with the left end and the lower end is continuous with the upper end. As a result, even if the mobile robot moves near the boundary where the necessary blocks are changed and this boundary is moved back and forth continuously, it is not necessary to repeatedly update the contents of the main memory, and the preparation of the map used for estimation is eliminated. Therefore, the processing amount can be significantly reduced.

FIG. 14 is a partial map 190 in the method including the extended region described above. FIG. 15 shows in correspondence with which position on the memory 776 the data of each block indicated by this partial map is written. The principle is that the continuity of the arrangement of blocks in the left and right vertical directions of the original partial map 190 is such that the right end is continuous to the left end and the lower end is continuous to the upper end in the arrangement on the memory 776, Is to be kept.
When the mobile robot 220 moves, it is necessary to change the contents of the partial map 190 from the one shown in FIG. 14 to the one shown in FIG. 16, and the newly added row (or column) The contents of (or column) shall be overwritten. This is shown in FIG.

  By writing the contents of the partial map 190 on the memory based on such a rule, the memory is scanned in the right direction from the memory number in which the contents of the upper left block of the partial map 190 are written. From this point, it is possible to access any point of the partial map 190 by scanning again from the top, further scanning downward, and when reaching the lower end, by scanning again from the upper end.

  Next, the effect of managing the partial map 190 on the memory by the method described above will be described.

  First of all, there is no need to prepare a correspondence table for associating positions on the map with memory addresses. If the contents are updated based on the general idea of buffer memory, the contents of the map at various locations are written to the same memory address, so it is necessary to manage all of this, but according to this method, Since the map position and the memory address are associated with each other based on the rules described above, the necessity is eliminated. This means that it is not necessary to refer to the correspondence table when accessing the map data in the memory at the time of position / orientation estimation, and as a result, the position / orientation estimation process can be speeded up.

  As a second effect, map data at consecutive positions on the map can be recorded continuously on the memory. In the matching process for position and orientation estimation described in the background art section, when the mobile robot is at the candidate position, the position indicated by the sensor data is used as a center, and the values of the surrounding map are referred to. In other words, since the data on the map to be accessed is often continuous in space, the time for memory access can be shortened compared to random access if it is written in a continuous address even in memory. It becomes possible.

  However, on the other hand, according to the memory array as described above, when the right end is reached, the process returns to the left end, the process for checking whether the right end is reached, and the process for returning to the upper end when the lower end is reached. The process of confirming whether or not the lower end has been reached is required every time a point on the map is accessed. This is illustrated in FIG. When accessing the area of 1500 in the figure, everything can be handled as a continuous memory, but when accessing the area of 1600 in the figure, it is confirmed whether or not the right end of the entire area is exceeded, and Sometimes it is necessary to access the leftmost region 1630. More specifically, in the program, a confirmation branch process corresponding to an if statement for confirming whether the memory address exceeds the right end of the figure is required. In general, this process is performed in an extremely short time so that it can be ignored as a single unit, but in the matching process for position and orientation estimation described in the background art section, it is repeated most frequently to calculate an evaluation function. In addition, since the other processes that are repeatedly performed in the same manner are very simple in nature, the effect on whether the confirmation branch process is performed or not is large on the overall processing time as a result. . For example, when the operation in the innermost part of the processing loop that is performed most frequently in the matching process is originally processing amount 3, and the processing amount of this confirmation branch is 1 processing amount as a single unit, the total processing amount is Theoretically, it will be increased to about 4/3.

  In order to solve this problem, the management method on the memory of the partial map 190 described above can be improved as follows. This will be referred to as an improved method for memory management of partial maps, and will be described in detail.

  In the improved method regarding the memory management of the partial map, as shown in FIG. 19, the memory has the number of pixels of 11 × 11 blocks, that is, the storage capacity of (40 × 11) × (40 × 11) = 4400 × 4400 pixels. Prepare as a continuous memory with. The method of assigning memory numbers is the same as that described with reference to FIG. In this improved method, new areas 1300 are provided in the uppermost column and the leftmost column of the memory area. Here, contents equivalent to the contents of the bottommost column in the memory area are written in the topmost column, and contents equivalent to the rightmost one column of the memory area are written in the leftmost column. By providing such an area and referring to the area 1670 instead of referring to the contents of the area 1600 and the area 1620 as shown in FIG. 20, the same contents can be referred to as a continuous memory area. As a result, when performing matching processing, it is not necessary to check whether the end of the memory has been reached when referring to continuous memory in a limited area, and the innermost processing loop that is most frequently performed in matching processing. The calculation amount of the part can be reduced, and as a result, the entire matching processing amount can be reduced. For example, according to the theoretical value described above, the processing amount can be reduced to 3/4.

  In the above description, the width of the new area 1430 is assumed to be equal to the width of the map block size, but it is reasonable to determine that the width of this area is equal to the width of the area accessed during matching. And it can be set like this.

  In the above description, the restriction on the rotation direction of the mobile robot 220 is not taken into consideration. However, the angle that can be rotated during one cycle described above is actually limited. By setting the area outside the necessary partial map 620, the size of the memory required for storing the partial map 190 can be reduced, and the amount of data read / write required for the update can also be reduced.

  FIG. 21 is a hardware configuration diagram showing an example of a hardware configuration for realizing the functions of the present invention described so far. The computer includes a CPU 20, a memory 30, a hard disk 40, a distance sensor interface (IF) 50, a moving mechanism interface (IF) 60, a network interface (IF) 70, a distance sensor 80, and a moving mechanism 90.

  The CPU 20 operates based on a program stored in the hard disk 40 or the memory 30 and controls each part. The hard disk 40 stores a program executed by the CPU 20 and stores an entire map. The memory 30 stores a program executed by the CPU 20 and stores a partial map. The distance sensor IF 50 operates the distance sensor 210 based on control by the CPU 20 and inputs sensing data. The moving mechanism IF 60 operates the moving mechanism 90 based on control by the CPU 20. The network IF 70 interfaces with a wireless or wired communication line, and performs program input / output, data input / output, external command input / output, and the like.

  The correspondence between these devices and each block in FIG. 1 will be described. The hard disk 40 realizes the entire map storage unit 160. The hard disk can be replaced with a memory-type external storage medium such as a compact flash (registered trademark), an external computer or a server system connected via a network. What implement | achieves the partial map update part 150 is a program which operate | moves by CPU20. The memory 30 implements the partial map storage unit 170. What implements the position / orientation estimation unit 110 is a program operating on the CPU 20 and a distance sensor IF 50. What implement | achieves the movement control part 120 is the program which operate | moves with CPU20, and IF60 for movement mechanisms. The moving mechanism unit 130 is realized by the moving mechanism 90.

It is a functional block block diagram which shows the whole structure of the mobile robot system which concerns on one Embodiment of this invention. It is a figure explaining the sensing method in the mobile robot system which concerns on one Embodiment of this invention. It is a figure explaining an example of the sensing data measured with the distance sensor. It is an example of the map used for position and orientation estimation. It is a figure explaining the structure which estimates a position and orientation by matching with a map and sensing data. It is a figure explaining the relationship between a whole map, the candidate position of a mobile robot, and a required partial map. It is a figure explaining the relationship between a partial map and the candidate position of a mobile robot. It is a flowchart explaining the flow of the process performed by the partial map update part. It is a figure explaining a mode that the area | region of a required partial map changes, when the candidate position of a mobile robot moves across a block. It is a figure explaining the area | region of a partial map in the method containing an extended area. It is a figure explaining the relationship between a partial map and the candidate position of a mobile robot in the method including an extended area. It is a figure explaining a mode that the area | region of a partial map changes, when the candidate position of a mobile robot moves across a block in the method containing an extended area. It is a figure explaining that it is not necessary to update the content of a partial map, when the candidate position of a mobile robot moves across blocks in the method including an extended area. It is a figure explaining the arrangement | sequence of the block of a partial map. It is a figure explaining the relationship between the block of the partial map of FIG. 14, and the position on memory. It is a figure explaining the arrangement | sequence of the block of a partial map when the area | region of a partial map changes. It is a figure explaining the relationship between the block of the partial map of FIG. 16, and the position on memory. It is a figure explaining the area | region of the memory which each accesses based on the value of two sensor data at the time of a matching process. It is a figure explaining how to arrange the block on a memory in the improvement method about memory management of a partial map. It is a figure explaining the area | region of the memory accessed based on the value of the sensor data in the matching process in the improvement method regarding the memory management of a partial map. It is a figure explaining the hardware constitutions in one embodiment of this invention.

Explanation of symbols

181 Compressed block map selected by read block selection instruction unit, 186 Block map for update in partial map storage unit

Claims (17)

In a mobile robot system that recognizes its position and orientation by comparing sensing data measured by a distance sensor with a map and the surroundings, and moves by movement control based on the recognition result,
The mobile robot system
An overall map storage unit for managing the entire map as a block map divided into a plurality of blocks,
A partial map storage unit for managing a map area necessary for position and orientation estimation in a continuous area as a partial map;
A position and orientation estimation unit for obtaining the position and orientation from the partial map and the sensing data;
A mobile robot system comprising: a partial map update unit that selects and reads the block map from the overall map management unit, and updates the content of the partial map storage unit using the read block map. The mobile robot system according to claim 1,
A mobile robot system characterized in that an area to be managed as the partial map is determined based on an initial position and orientation estimation value of the robot. The mobile robot system according to claim 1 or 2,
The entire map storage unit stores the block map as compressed data,
The mobile robot system, wherein the partial map update unit includes means for decompressing the read compressed data. The mobile robot system according to any one of claims 1 to 3,
2. The mobile robot system according to claim 1, wherein the partial map includes an extended partial map including a necessary partial map necessary for position and orientation estimation and a map of a peripheral portion thereof. The mobile robot system according to any one of claims 1 to 4,
The entire map is composed of a plurality of block areas that are rectangular areas of W × H pixels,
The necessary partial map is a rectangular area composed of M × N block areas,
The partial map obtained by combining the necessary partial map and the extended partial map is a rectangular area composed of (M + 1) × (N + 1) block areas,
When the robot moves and the initial position and orientation estimation value of the robot changes, the block map is read from the entire map so that the necessary partial map is always included in the partial map, and the partial map is updated. A mobile robot system characterized by that. The mobile robot system according to any one of claims 1 to 4,
The partial map storage area is composed of continuous physical memory,
The physical memory has a size capable of storing pixel data of (M × W) × (N × H) pixels,
The mobile robot system characterized in that the partial map storage area is allocated on the physical memory so that the right end of the rectangular area is continuous with the left end and the lower end is continuous with the upper end. The mobile robot system according to claim 5, wherein
The partial map storage area is composed of continuous physical memory,
The physical memory has a size capable of storing pixel data of ((M + 1) × W) × ((N + 1) × H) pixels,
The mobile robot system characterized in that the partial map storage area is allocated on the physical memory so that the right end of the rectangular area is continuous with the left end and the lower end is continuous with the upper end. The mobile robot system according to any one of claims 1 to 4,
The partial map storage area is composed of continuous physical memory,
The physical memory has a size capable of storing pixel data of (U + M × W) × (U + N × H) pixels using a predetermined width parameter U,
The partial map storage area is allocated on the physical memory so that the right end of the rectangular area is continuous with the left end and the lower end is continuous with the upper end.
A mobile robot system, wherein the content of the right end width U is copied to the left end width U portion of the physical memory, and the lower end width U content is copied to the upper end width U portion. The mobile robot system according to claim 5, wherein
The partial map storage area is composed of continuous physical memory,
The physical memory has a size capable of storing pixel data of (U + (M + 1) × W) × (U + (N + 1) × H) pixels, using a predetermined width parameter U,
The partial map storage area is allocated on the physical memory so that the right end of the rectangular area is continuous with the left end and the lower end is continuous with the upper end.
A mobile robot system, wherein the content of the right end width U is copied to the left end width U portion of the physical memory, and the lower end width U content is copied to the upper end width U portion. A distance sensor that measures the situation around the mobile robot system and generates sensing data;
An overall map storage unit for storing a plurality of block maps;
A partial map management unit for managing a partial map obtained by expanding some of the plurality of block maps stored in the overall map storage unit;
The block map included in the partial map managed by the partial map management unit is selected from the block map stored in the entire map storage unit, and the selected block map is expanded and managed by the partial map management unit A partial map updater for updating the map;
A position and orientation estimation unit that obtains a position and orientation of the mobile robot system by collating the sensing data with the partial map;
A movement control unit that controls a moving mechanism unit that moves the mobile robot system according to the obtained position and orientation;
The partial map update unit calculates a range of the necessary partial map based on the position and orientation, and selects a block map included in the partial map managed by the partial map management unit so as to include the entire necessary partial map. Mobile robot system characterized by In claim 10,
The mobile robot system, wherein the entire map storage unit stores the compressed block map. In claim 10,
The mobile robot system, wherein the necessary partial map is selected so that a block map corresponding to the position of the mobile robot system obtained from the position and orientation is centered. In claim 10,
The partial map has more block maps than the necessary partial map;
The range of the necessary partial map is changed according to a change in the position of the mobile robot system obtained from the position and orientation,
When the range of the necessary partial map is changed to include a block map that is not included in the partial map, the partial map update unit expands the block map from the overall storage map storage unit and manages the partial map A mobile robot system characterized by being added to a partial map managed by the department. In claim 13,
A mobile robot system, wherein a block map not included in the necessary partial map is deleted from the partial map management unit when the block map is added to the partial map.   15. The mobile robot system according to claim 14, wherein the block map to be added to the partial map is stored in a storage area of the partial map management unit storing the block map to be deleted. In a control method for a mobile robot system, which recognizes its position and orientation by comparing sensing data obtained by measuring a surrounding situation with a distance sensor and a map, and moves by movement control based on the recognition result,
The mobile robot system includes:
An overall map storage unit for managing the entire map as a block map divided into a plurality of blocks,
A partial map storage unit for managing a map area necessary for position and orientation estimation in a continuous area as a partial map;
A position and orientation estimation step in which a position and orientation estimation unit of the mobile robot system obtains the position and orientation from the partial map and the sensing data;
A partial map update unit of the mobile robot system comprising: selecting and reading the block map from the overall map management means, and updating the content of the partial map storage unit using the read block map. A mobile robot system control method. A distance measurement step in which the distance sensor measures the surrounding situation of the mobile robot system and generates sensing data;
A block map selection step in which the partial map update unit selects from the block map included in the partial map managed by the partial map management unit from an entire map storage unit that stores a plurality of block maps;
A partial map update step in which the partial map update unit extends the selected block map to update a partial map managed by the partial map management unit;
A position and orientation estimation unit, wherein the position and orientation estimation unit obtains the position and orientation of the mobile robot system by comparing the sensing data and the partial map;
A movement control unit comprising a movement step of controlling a movement mechanism unit that moves the mobile robot system according to the obtained position and orientation;
In the map selection step, the partial map update unit calculates a range of the necessary partial map based on the position and orientation, and the partial map managed by the partial map management unit includes the entire necessary partial map. A control method of a mobile robot system, characterized by selecting a block map.

Images