JP6123370B2 - Autonomous mobile device that moves by measuring the distance to the obstacle - Google Patents

Description

  In this specification, the distance and direction to the obstacle that becomes an obstacle to movement are measured, the non-existing range of the obstacle is specified from the measurement result of the distance to the obstacle and the direction, and the specified non-existing range of the obstacle Disclosed is a technology related to an autonomous mobile device that searches for a route that travels through the vehicle and moves.

Patent Document 1 discloses an autonomous mobile device. This type of autonomous mobile device moves while measuring the distance and direction to the obstacle. If the position and orientation of the mobile device are found and the measurement results of the distance and orientation from the mobile device to the obstacle are obtained, the location of the obstacle can be calculated.
(3) of FIG. 2 schematically shows an obstacle position group calculated from a measurement result group of distances by direction measured by the distance measurement apparatus by direction provided in the mobile device. In the figure, V is a laser beam for distance measurement, and is scanned in the W direction shown in the figure. The azimuth distance measuring apparatus measures the distance to the point where the laser beam V is reflected. If the distance is measured every predetermined time, the distance to the obstacle can be measured every predetermined angle. The direction-specific distance measuring device can be said to be a device that measures the distance and direction to an obstacle. The distance measurement result includes an error, and a distance shorter than the actual distance to the obstacle U may be measured or a long distance may be measured. U1 exemplifies a case where the measurement is short, and is a calculation result indicating that there is an obstacle at a position closer to the front side than the actual obstacle U. U2 exemplifies the case where the measurement is performed for a long time, and is a calculation result indicating that there is an obstacle at a position on the back side of the actual obstacle U.

(1) of FIG. 2 illustrates the result of calculating the position of the obstacle from the measurement result group of the distance and direction to the obstacle existing point P measured in the state where the autonomous mobile device is at the point Q. ing. If the distance and azimuth measurement results are accurate, the calculated position matches the true position P. However, the distance measurement result includes an error, and the position calculated from the measurement result deviates from the true position P.
For example, if a distance shorter than the true value is measured, a position on the near side (S side) from the true position P is calculated, and if a distance longer than the true value is measured, the position behind the true position P is calculated. The (R side) position is calculated. The positions Pa, Pb... Calculated from the measurement results are dispersed before and after the true position P. An arrow G indicates an error range, and reference symbols E1 and E2 indicate the magnitude of the error. Depending on the characteristics of the measuring device, E1 and E2 may be equal or different.
ing.

  Since each position calculation result group includes an error, there may be a method of performing an averaging process and correcting it to a value close to the true value. However, in order to perform the averaging process, it is necessary to repeat the distance measurement for each direction illustrated in (3) of FIG. In addition, an obstacle may move, and it is not guaranteed that the obstacle will approach the true value by averaging. Therefore, for the sake of safety, a technique is adopted in which the obstacle non-existing range is identified as the obstacle is located at the position closest to the obstacle in the position calculation result group.

For example, in the case of FIG. 2 (1), since the calculated positions are distributed within the range of the arrow G, if the obstacle non-existing range is specified assuming that there is an obstacle at the closest position, the range of the arrow T is It is an obstacle non-existing range. The arrow T is a range on the moving device side from a position closer to the moving device side from the true position P by the miscalculation magnitude E1. As shown in (3) of Fig. 2, when the positions of multiple measurement points are calculated, the presence range of obstacles and the non-existence range of obstacles are specified from the measurement result group of multiple positions. To do. In other words, the polygon that includes the measured position group is specified, the front side of the polygon is specified, the front side of the polygon is set as the obstacle nonexistence range, and the back side of the side is set as the obstacle. The range of existence. In the case of (3) in FIG. 2, the line closer to the front side by E1 than the front surface of the real obstacle U is the side on the near side of the polygon including the measured position group, and is closer to the front surface of the real obstacle U. The line on the back side by E2 is the back side of the polygon that includes the measured position group.
When the obstacle non-existing range is specified according to the above logic, it is possible to prevent a situation in which the mobile device and the obstacle come into contact due to a measurement error. Safety is improved.

JP 2004-280451 A

  According to the obstacle non-existing range specifying technique shown in FIG. 2A, it is possible to prevent the moving device and the obstacle from contacting each other. However, there is an opportunity to erroneously determine a position that can actually move as an immovable position. FIG. 1 shows a state in which the autonomous mobile device 2 moves into a narrow recess 30 a formed in the wall 30. In FIG. 3 (1), reference numeral 32 indicates a target position, which is taught to the autonomous mobile device 2 in advance. The autonomous mobile device 2 ends the movement when the reference point 4a matches the target position 32. FIG. 3A illustrates a case where the target position 32 is a position where the moving device 4 enters the recess 30 a when the reference point 4 a coincides with the target position 32.

  FIG. 3 (2) shows a state in which the obstacle non-existing range is specified from the result group measured by the distance / orientation measuring device 6 mounted on the autonomous mobile device 2. Depending on the measurement error, it may be calculated that there is a wall in front of the actual wall 30, or it may be calculated that there is a wall on the back side. The line 30s is a line surrounding the position calculated to be on the front side, and the line 30r is a line surrounding the position calculated to be on the back side. The distance between the two is equal to the error magnitude E1 + E2 described in FIG. The autonomous mobile device 2 identifies the range T on the near side of the front line 30s surrounding the position calculated from the measurement result as the obstacle non-existing range.

  If the obstacle non-existing range T is specified as described above, the specified obstacle non-existing range may be too narrow to actually match. In the case of FIG. 3B, the front end of the moving device 4 protrudes from the obstacle non-existing range T while moving until the reference point 4 a coincides with the target position 32. When the autonomous mobile device 2 moves to the instructed target position 32, the autonomous mobile device 2 expects to touch an obstacle. The autonomous mobile device 2 is programmed to take measures to avoid contact when it is expected to contact an obstacle.

  In reality, the distance measurement error decreases as the distance to the obstacle approaches. When approaching an obstacle, the obstacle non-existing range is accurately specified, and the target position 32 designated in advance and the accurately specified obstacle non-existing range often do not contradict each other. In other words, there was a case where it was a misjudgment that the obstacle non-existing range specified at the distance measurement stage from a distance was inaccurate, and it was determined that some kind of avoidance processing was necessary due to the inaccurate non-existing range. Arise. In practice, unnecessary movement processing is performed, so that the smooth movement of the autonomous mobile device 2 is prevented.

  In this specification, despite the fact that there is actually a movable obstacle non-existing range, the obstacle non-existing range is specified to be narrower than the actual range, and as a result, a situation in which it is erroneously determined that it is impossible to move is generated. A technique for reducing the frequency is disclosed.

The autonomous mobile device disclosed in this specification includes a mobile device, a device that identifies the position and orientation of the mobile device, a distance and direction to an obstacle that is mounted on the mobile device and prevents movement of the mobile device. An obstacle non-existing range specifying device for specifying a non-existing range of an obstacle from a distance / azimuth measuring device to be measured, a position and azimuth specifying result of the moving device, and a measurement result group of a distance and an azimuth to the obstacle, and And a route search device for searching for a route that moves within the non-existing range of the obstacle. The obstacle non-existing range specifying device specifies a polygon that includes the measured position group, specifies the near side of the polygon, and specifies the near side of the polygon as the non-existing range of the obstacle. To do. In addition to the above, the autonomous mobile device disclosed in the present specification includes an irregular error storage device that stores the irregular error size included in the distance measurement result by the distance / orientation measurement device, There is provided an irregular error correction device that corrects the distance measurement result by the azimuth measuring device longer by the irregular error stored in the irregular error storage device. The obstacle non-existing range specifying device specifies the obstacle non-existing range using the distance corrected by the irregular error correcting device.

  For example, when the position of Pa in FIG. 2 (1) is calculated from the measurement result, it is unclear whether it contains a distance error longer than the true value or a distance error shorter than the true value. It is unclear whether the calculated position is closer to the true value if a distance error is added, or the true value is approached if the distance error is subtracted from the calculated position. There is no significant method for correcting to a true value for an irregular error whose sign is unknown.

  Nevertheless, the technique disclosed in this specification treats the position as calculated from a distance including an error measured shorter than the true value, and corrects the error by a larger amount of irregular error. All the calculation positions Pa, Pb, etc. are corrected by the amount of irregular error. A correction method that is not mathematically guaranteed to approach the true value is adopted.

If it is normal, it would be strange to correct the measurement result of the distance including the irregular error whose plus / minus is unknown by the length of the irregular error.
However, when used in combination with a technique for identifying an obstacle nonexistence range from a calculation position group calculated from a measurement result group, it is meaningful to correct the distance to the longer side. As illustrated in FIG. 3 (3), when the position of the obstacle is calculated after correcting the distance measurement result group to be long, and the obstacle non-existing range is identified from the calculated position result group, the case of the prior art Thus, it is possible to specify an obstacle non-existing range whose outline is a line close to the actual obstacle range. Also in the case illustrated in (3) of FIG. 2, the distance measurement result group is corrected to be long, the position of the obstacle is calculated, the polygon that includes the measured position group is specified, and the front of the polygon If the side on the side is specified, the front side of that side is the obstacle nonexisting range, and the back side is the obstacle existing range, the boundary between the obstacle nonexisting range and the obstacle existing range is the obstacle It matches well with the front of the object U. After correcting the distance measurement result group for a long time, calculating the position of the obstacle, and specifying the obstacle non-existing range from the calculated position result group, there is actually a movable obstacle non-existing range Regardless, the obstacle non-existing range that can reduce the occurrence frequency of the situation that it is erroneously determined as immovable is specified.

  In some cases, the distance measurement result by the distance / orientation measuring device includes a regular error in which the plus / minus is determined in addition to the irregular error in which the plus / minus is not known. For example, when the true distance is 10 m, it may be regularly measured as 10.1 m. That is, the distance measurement result group may be irregularly distributed positively or negatively with an average value of 10.1 m. In this case, in addition to the regular error of measuring 0.1 meter longer, it can be said that an irregular error with no plus or minus is superimposed. FIG. 2B illustrates a case where the regular error F and the irregular error E are superimposed. The regular error F has plus and minus and is hereinafter referred to as a value. As for the irregular error E, plus or minus is unknown, but its magnitude can be measured.

If the distance measurement result includes a regular error, the regular error storage device that stores the value of the regular error, and the distance measurement result by the distance / orientation measurement device A regular error correction device that corrects only the stored regular error value is added, and the obstacle non-existing range specifying device is corrected by the regular error correction device, and further corrected by the irregular error correction device for a long time. It is preferable to specify the obstacle non-existing range using the measured distance.
For example, when the true distance is 10 m and when it is regularly measured as 10.1 m, the fact that a regular error of +0.1 m is included is stored, and the measurement result is the error of the regular error. Correct by the value, and further correct by the magnitude of the irregular error. Alternatively, when the true distance is 10 m and when regularly measuring 9.8 m, the fact that a memory error of −0.2 m is included is stored, and the measurement result is converted into a regular error. Is corrected by the value of, and is further corrected by the magnitude of the irregular error. For regular errors, the correction direction is switched depending on the value of the regular error. The irregular error is always corrected longer than the measurement result.

  As described above, the regular error included in the measurement value is corrected so as to approach the true value. After that, the boundary between the non-existing range and the existing range of the obstacle is specified after correcting the irregular error to the longer side. It is possible to accurately identify the non-existing range of the obstacle.

A mode that the autonomous mobile device of an Example moves to the narrow space enclosed by the wall is shown. The figure explaining the calculation result group of the position obtained by moving while measuring distance and direction. The figure explaining the process which specifies an obstacle non-existence range from the calculation result group of a position. The figure which shows a mode that the measurement result of distance disperses. The figure which illustrates the value of a regular error, and the magnitude | size of an irregular error. The figure which shows the process sequence which the autonomous mobile apparatus of an Example implements. The figure which shows the system configuration | structure of the autonomous mobile apparatus of an Example.

The features of the embodiments described below are listed.
(Feature 1) An obstacle non-existing range is specified by excluding a measurement result whose distance measurement result is outside the range of σ or 3σ.
(Feature 2) The distance is measured by the light reception time of the reflected light from the obstacle.
(Feature 3) The irregular error storage device stores the irregular error magnitude in association with the reflection intensity.
(Specific 4) The distance / orientation measuring device is an ultrasonic radar, millimeter wave radar, laser radar, a device that measures by the active stereo method (light cutting method, spatial coding method), or a device that measures by the passive stereo method. .

  FIG. 1 shows a process in which the autonomous mobile device 2 according to the embodiment moves to a space (depression 30 a) surrounded by a wall 30. The autonomous mobile device 2 includes a mobile device 4 incorporating a drive device and a steering device, and a distance / direction measuring device 6 that measures the distance and direction to the obstacle as viewed from the mobile device 4. The moving device 4 has a reference point 4a. As will be described later with reference to FIG. 7, the moving device 4 includes a self-position / self-direction specifying device 16 that specifies the position and orientation of the moving device 4. If the position and azimuth data by the self-position / self-direction specifying device 16 and the distance and azimuth data from the moving device 4 to the obstacle by the distance / azimuth measuring device 6 are obtained, vector calculation is performed to calculate the obstacle. The position can be calculated. The obstacle position calculation device 18 performs vector calculation.

FIG. 2 shows a calculation result group in which the position of the obstacle is calculated from the measurement result group of the distance and direction to the obstacle existing point P measured by the autonomous mobile device 2. The distance measurement result includes an error, and the position calculation result calculated from the measurement result deviates from the true position P. The positions Pa, Pb,... Exemplify each position calculated from the measurement result, and are distributed around the true position P. An arrow G indicates an error range, and reference numerals E1 and E2 indicate error magnitudes.
As will be described later, when the distance measurement result includes a regular error, the center position Pc of the error range G and the true position P are shifted. FIG. 2A illustrates the case where no regular error exists, and FIG. 2B illustrates the case where the regular error F exists.
The value and plus / minus of the regular error F can be measured in advance. The regular error can be corrected by the value of the regular error F measured in advance (the plus or minus is also known). Regarding the irregular error, the magnitude E of the error can be measured in advance, but the plus or minus of the irregular error included in the measured value is not known.

  FIG. 3 shows a process of calculating a position from a measurement result group including an error and specifying an obstacle non-existing range from the result group of the calculated position. The obstacle non-existing range is specified by specifying a boundary with the obstacle existing range. If the obstacle non-existing range is specified, the obstacle existing range is specified.

  FIG. 3A illustrates the periphery of the recess 30a surrounded by the wall 30, and the wall 30 is an obstacle. Although the dent 30a is narrow, it has a width and a depth into which the moving device 4 can enter. The moving device 4 can proceed until the reference point 4a coincides with the target position 32. When the reference point 4a coincides with the target position 32, the moving device 4 enters the recess 30a. The rear end of the moving device 4 does not protrude from the recess 30a.

  FIG. 3 (2) shows the position of the obstacle (wall 30) calculated from the result group measured by the distance / orientation measuring device 6 mounted on the autonomous mobile device 2, and the obstacle from the calculated position result group. It shows how the non-existing range is specified. Depending on the measurement error, it may be calculated that there is a wall in front of the actual wall 30, or it may be calculated that there is a wall on the back side. The line 30s is a line surrounding the position calculated to be on the front side, and the line 30r is a line surrounding the position calculated to be on the back side. The autonomous mobile device 2 identifies the range T on the near side of the front line 30s surrounding the position calculated from the measurement result as the obstacle non-existing range. According to this technique, when the autonomous mobile device 2 moves to the instructed target position 32, it erroneously determines that the mobile device 4 contacts the obstacle.

FIG. 3 (3) shows that the position of the obstacle is calculated after correcting the measurement result group of the distance / orientation measuring device 6 by the length of the irregular error E included in the measurement result group. A state in which the front surrounding line 30s1 is specified from the result group is shown. Obviously, the line 30s1 matches well with the actual wall 30 position. According to the technique of FIG. 3, an obstacle non-existing range T1 that closely matches the actual obstacle non-existing range is identified. In practice, it is possible to prevent erroneous determination of the movable target position 32 as an immovable position.
When the regular error F is included in the measurement result group of the distance / orientation measuring apparatus 6, the position is calculated after correcting the regular error. For example, when the distance / orientation measuring device 6 includes a regular error that measures 0.1 m longer and the distance / orientation measuring device 6 measures 10 m, the measurement result is corrected to 9.9 m. To do. After the regular error value F is corrected, the irregular error magnitude E is corrected. Irregular errors are always corrected longer.

FIG. 4 shows the relationship between the measurement result by the distance of the distance / orientation measuring apparatus 6 and the true value. This relationship can be measured in advance. The vertical axis represents the measurement frequency. FIG. 4 illustrates a case where a regular error is included. F shown in the figure indicates a regular error value included in the measurement result. In the case of FIG. 4, according to the distance / orientation measuring device 6, a case where the measurement is regularly shortened by F is illustrated. F may be positive or negative.
E1 and E2 illustrate the magnitude of the irregular error. Irregular errors can be positive or negative. E1 indicates 3σ when the measurement is short, and 99% or more of the measurement result is a longer distance. E2 indicates 3σ when measured for a long time, and 99% or more of the measurement result is a shorter distance. Normally, E1 = E2, but depending on the characteristics of the measuring device 6, E1 and E2 may not match. In the case of inconsistency, the irregular error magnitude E1 of the shorter measurement is used for the following correction. A measurement result that does not fall within the range of σ to 3σ is an abnormal value and is excluded from the subsequent processing. Only the measurement result within the range which is a normal measurement result is adopted to identify the obstacle non-existing range.

  FIG. 5A shows an example of regular error values, and the horizontal axis indicates the measured distance. Regular error values may vary depending on distance. In the case of illustration, the case where it is in the characteristic of measuring long at short distance and measuring short at long distance is illustrated. The regular error value may be positive or negative. The value of the regular error is determined by the characteristics of the distance / orientation measuring device 6. The graph shape shown in FIG. 5A can be measured for each distance / orientation measuring device 6.

  FIG. 5 (2) shows an example of the magnitude of the irregular error, and the horizontal axis shows the measured distance. The magnitude of the irregular error may vary depending on the distance. In the illustrated case, the irregular error is small at a short distance, and the irregular error is large at a long distance. Usually, the magnitude of the irregular error increases as the distance increases. The magnitude of the irregular error is always positive and not negative. The magnitude of the irregular error is determined by the characteristics of the distance / orientation measuring device 6. The graph shape of FIG. 5 (2) can be measured for each distance / orientation measuring device.

FIG. 6 shows a processing procedure performed by the autonomous mobile device 2. FIG. 7 shows a system configuration of the autonomous mobile device 2.
In step 2, the distance and direction from the moving device 4 to the obstacle are measured by the distance / direction measuring device 6. In this embodiment, the distance to the obstacle is measured every predetermined angle. In step S2, a plurality of distance data is obtained. Processing from step S2 to step S12 described later is performed for each distance data.
In step S4, the regular error is corrected. The regular error is often a function of distance as illustrated in FIG. The regular error model 8 stores the relationship between the distance and the regular error value. The regular error model 8 can be measured in advance and stored in advance. The regular error correction device 10 corrects using the regular error value described in the regular error model 8. Here, correction is performed according to the sign of the value of the regular error described in the regular error model 8.
In step S6, the distance is corrected farther away (to the longer distance side) by the irregular error. As illustrated in FIG. 5B, the magnitude of the irregular error is often a function of distance. The irregular error model 12 stores the relationship between the distance and the magnitude of the irregular error. The irregular error model 12 can be measured in advance and stored in advance. The irregular error correction device 14 corrects the irregular error using the irregular error size described in the irregular error model 12. Here, correction is always made on the far side.
In step S8, the position of the obstacle is calculated from “the distance to the obstacle corrected by the regular error and corrected to the longer side by the irregular error” and “the direction of the obstacle”. In step S8, the position of the obstacle is calculated using a sensor coordinate system fixed to the distance / orientation measuring apparatus 6.
In step S <b> 10, the position and direction of the moving device 4 in the absolute coordinate system are specified by the self-position / self-direction specifying device 16. This reveals the relationship between the sensor coordinate system and the absolute coordinate system. Note that the difference in execution time between step S2 and step S10 is short, and the movement of the moving body between them can be ignored.
In step S12, the position of the obstacle calculated in the sensor coordinate system is converted into a coordinate value in the absolute coordinate system. The obstacle position calculation device 18 executes the processes in steps S8 and S10. As described above, the process of step S12 is executed for all directions. When step S12 is executed, the measurement point position for each direction is determined.
In step S14, from the result group of the obstacle positions calculated by the obstacle position calculation device 18, the polygons enclosing the positions are specified, the sides of the polygons located on the side closer to the moving device are specified, A range closer to the moving device than the side is defined as an obstacle non-existing range (movable range), and a range farther than that is defined as an obstacle existing range. The calculation is performed by the obstacle non-existing range specifying device 20.
In step S16, a route that reaches the target position by moving within the obstacle non-existing range is searched. The route search device 22 searches for the travel route. If the route is not searched, it is determined in step S18 that the search is impossible, and a measure is taken when the search fails (step S26). In the present embodiment, since it is prevented that the obstacle non-existing range is narrowly recognized, the chance of being determined as unsearchable in step S18 is reduced.
In step S20, the speed and direction for moving along the searched route are calculated and commanded to the movement control device 26. The autonomous mobile device 2 moves along the searched route.
In step S22, it is determined whether the target position has been reached in the target posture. The above processing is repeated until the target position arrives at the target posture.

  According to the above embodiment, the route is searched after correcting the irregular error for a longer time. As shown in FIG. 5 (2), the magnitude of the irregular error becomes smaller as the obstacle is approached. Even if the irregular error is corrected to be long, the moving device and the obstacle do not come into contact with each other. According to the above embodiment, it is possible to remarkably reduce the case where a route is not searched even though there is actually a movable route.

In the present embodiment, the case of entering the dent has been illustrated, but the usefulness thereof is not limited to the case of entering the dent. It is also useful when passing through a bent obstacle non-existing range. In the present embodiment, the technology for distinguishing the obstacle existence range and the obstacle non-existence range on the two-dimensional map has been described. In practice, there are obstacles at various heights. It is preferable to measure the distance to the obstacle and the three-dimensional orientation, calculate the three-dimensional position from the distance and the three-dimensional orientation, and project the three-dimensional position onto a two-dimensional map.
FIG. 5 illustrates a case where the value of the regular error and the magnitude of the irregular error are uniquely determined with respect to the distance. In practice, the value of the regular error and the magnitude of the irregular error may change in some cases. For example, when measuring the distance and azimuth by measuring the time it takes for the light emitted toward a specific direction to return after being reflected by an obstacle, the error value or error magnitude depends on the reflected light intensity. May change. For example, when the reflectance of the obstacle is high, the distance tends to be shorter than the true value, and the variation in the detection result is small, whereas when the reflectance of the obstacle is low, the distance is longer than the true value. There is a case where there is a large variation in detection results. When the regular error value and the irregular error magnitude vary depending on the reflected light intensity, the graph showing the regular error value and irregular error magnitude shown in FIG. 5 is associated with the reflected light intensity. There are cases where it is better to memorize it.

  The technical elements described in this specification or the drawings exhibit technical usefulness alone or in various combinations, and are not limited to the combinations described in the claims at the time of filing. In addition, the technology illustrated in the present specification or the drawings achieves a plurality of objects at the same time, and has technical utility by achieving one of the objects.

2: Autonomous mobile device 4: Mobile device 6: Distance / orientation measuring device

Claims (3)

A mobile device;
A device for identifying the position and orientation of the mobile device;
A distance / azimuth measuring device that is mounted on the mobile device and measures the distance and direction to an obstacle that prevents the mobile device from moving;
From the result of specifying the position and direction of the mobile device and the measurement result group of the distance to the obstacle and the direction, specify a polygon that includes the measured position group, specify the near side of the polygon, An obstacle non-existing range specifying device that specifies the near side of the side as an obstacle non-existing range;
A route search device for searching for a route that moves within a non-existing range of obstacles;
An autonomous mobile device comprising
A random error storage device that stores the size of the irregular error included in the distance measurement result by the distance / orientation measurement device;
A random error correction device that corrects the distance measurement result by the distance / orientation measurement device longer by the size of the irregular error stored in the irregular error storage device is added,
An autonomous mobile device characterized in that an obstacle non-existing range specifying device specifies an obstacle non-existing range using the distance corrected by the irregular error correcting device. A regular error storage device for storing a regular error value included in a distance measurement result by the distance / azimuth measuring device;
A regular error correction device is added to correct the distance measurement result by the distance / orientation measurement device by the regular error value stored in the regular error storage device,
2. The obstacle non-existing range specifying device specifies an obstacle non-existing range using a distance corrected by a regular error correcting device and further corrected by the irregular error correcting device. Autonomous mobile device. The irregular error storage device stores the magnitude of the error in association with the distance,
The autonomous mobile device according to claim 2, wherein the regular error storage device stores an error value in association with the distance.

Images