JP2009178782A - Mobile object, and apparatus and method for creating environmental map - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an apparatus and a method for recognizing an environment, preventing erroneous detection of the environment, and also to provide a robot having the apparatus incorporated. <P>SOLUTION: The robot has: a moving means; a detection unit 18 for detecting a floor around the robot; an environmental map creation module 12 for comparing current detection results detected by the detection unit 18 with previously detected detection results to create or update the environmental map around the mobile object of the robot, based on the detection results when environmental conditions around the robot coincide; and a path planning module for planning a moving path of the moving means referring to the environmental map. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an environment map generation apparatus and method for generating an environment map from detection results of a distance sensor or the like, and a mobile body that makes a route plan based on the environment map.

  2. Description of the Related Art Conventionally, when a mobile robot moves, there is a technique for determining a movement path by measuring the state of a floor surface with a sensor and mapping it (for example, see Patent Document 1). In Patent Literature 1, an environment map is generated for determining whether or not the area can be moved even in an area where there is a level difference from the floor surface. The route planning unit that performs route planning based on the environmental map is considered not to be an obstacle in the environmental map on other planes other than the floor even if it is an obstacle on the environmental map on the floor. In this case, the route is a candidate for the movement route.

  However, in the technique described in Patent Document 1, there is a case where it is difficult to measure accurately due to vibration or shaking of the robot itself, and in some cases, an erroneous measurement is recognized as the floor surface. .

  By the way, the robot usually acquires position information from the odometry, acquires angle information from the gyro, and corrects the three-dimensional data (3D data). The accuracy of odometry in a biped robot depends on the rigidity of the mechanism and the quality of the feet. Mechanical rigidity cannot be easily increased by handling weight and electrical equipment. In addition, it is considered that footing is worse because it does not necessarily walk in an ideal environment such as walking on rough terrain. That is, it is not easy to improve the accuracy of odometry.

Although methods for identifying the absolute value of a robot using an external sensor such as motion capture and GPS (Global Positioning System) can be obtained with relatively high accuracy, there is a problem in terms of cost for environment construction. That is, there is a need for a solution other than directly measuring the vibration and vibration of the robot. Here, stable recognition is possible by intentionally selecting a frame with good 3D data accuracy and discarding the other frames.
JP 2005-92820 A

  As shown in FIG. 14, the 3D scan by the range sensor has an accuracy that can use raw data as it is, and can express a three-dimensional shape well. However, when mounted on a foot type robot, the state of the robot (sway, 3D accuracy may be reduced due to vibration. That is, it is difficult for a robot to generate an accurate environment map based on data from a range sensor or the like.

  The present invention has been made to solve such problems, and it is an object of the present invention to provide an environment recognition device, an environment recognition method, and a mobile body equipped with the environment recognition device that can prevent erroneous detection of the environment. And

  In order to solve the above-described problems, a moving body according to the present invention includes a moving means, a detecting means for detecting a floor surface around the moving body, and an environmental state around the moving body according to the movement of the moving means. The environmental map generation means for generating or updating the environmental map around the mobile body based on the detection result when the environmental state matches the prediction result predicted by the detection means and the detection result detected by the detection means, Route planning means for planning a movement route of the moving means with reference to an environmental map.

  In the present invention, the environment map generating means compares the predicted result of predicting the environmental condition around the moving object according to the movement of the moving object and the detection result detected by the detecting means, and only when the environmental condition matches. Generate and update an environmental map. As a result, when the predicted result is different from the detection result of the detection means, it is possible not to update the environment map, thereby reducing the environment grasp due to erroneous detection.

  The moving body according to the present invention compares a moving means, a detecting means for detecting a floor surface around the moving body, a current detection result detected by the detecting means, and a detection result detected before the present time, If the environmental conditions around the moving body match, based on the detection result, an environmental map generating means for generating or updating an environmental map around the moving body, and a movement path of the moving means with reference to the environmental map A route planning means for planning.

  In the present invention, when the detection results coincide with each other over a plurality of frames, the environment map is updated based on the detection result, so that an environment map (global map) including only the detected obstacles can be generated.

  The environment map generation means compares the current detection result detected by the detection means with the detection result detected before the present time, and the presence or absence of an obstacle and the position of the obstacle match. In this case, the environmental map can be updated based on the detection result. When the presence / absence of an obstacle and its position match, it can be recognized that the obstacle exists.

  Furthermore, the environmental map generation means compares the current detection result detected by the detection means with the detection result detected before the present time, and determines the height from the floor surface of the plane other than the floor surface and its method. If the line vectors match, the environment map can be updated based on the detection result. By recognizing a plane other than the floor as the height from the floor and its normal vector, the plane can be easily described.

  Furthermore, the environmental map generating means can update the environmental map with the detected frame when it is determined that the same plane is included for a plurality of consecutive frames among the detected frames detected by the detecting means. . Obstacle detection accuracy can be further improved by detecting the same plane for a plurality of frames.

  The environment map generation means compares the current frame, which is the current detection frame, among the detection frames detected by the detection means, and the previous frame detected before the present, and the surrounding environmental conditions do not match. In this case, it is possible to compare the environmental conditions around the next frame, which is the next detection frame, and the previous frame and / or the current frame, and discard the detection frames that do not match. This reduces false recognition of environmental maps.

  Further, the environmental map generation means includes an obstacle detection means for detecting a plane other than the floor surface from the detection frame detected by the detection means, and the obstacle detection means to the floor with a plurality of continuous detection frames. When detecting the same plane other than a plane, an obstacle registration unit that registers the plane as a fixed plane in the environmental map or plane registration unit, and an environmental map update unit that updates the environmental map by a detection frame having the fixed plane; It can have.

  Furthermore, when the obstacle detection unit detects a plane other than the floor surface in the detection frame by the obstacle detection unit, the obstacle registration unit determines whether the plane is a registered plane with reference to the environment map. The map update means can update the environmental map with the detection frame having the registered plane. Thereby, an environmental map can be updated using a detection frame with a low probability of misrecognition.

  The prediction result is a conversion result obtained by converting a detection result before the present into a global coordinate system, and the environment map generation unit converts the current detection result into a global coordinate system and converts the current detection result with the conversion result. A match can be determined. By using an environmental map of the global coordinate system, the recognition range can be expanded, and the reliability of the landing position plan can be improved.

  Further, the environment map generating means sets a predetermined area adjacent to the moving body as a plane determination area, sets a predetermined area on the side farther from the moving body than the plane determination area as a plane search area, The environmental map can be updated based on the detection result of the planar search area by the detection means. By dividing the periphery of the moving body into a plane determination area and a plane search area and updating the environment map based on the detection result for the plane search area, it is possible to reduce erroneous detection and efficiently update the environment map.

  Furthermore, the environment map generation means may set the plane fixed region based on observation results over a plurality of frames. By making the environmental map obtained by observing over a plurality of frames a definite region, useless observation can be reduced.

  Moreover, the said environment map production | generation means shall discard this, when it is planes other than the said floor surface in the said plane fixed area | region, and detects a new plane. Since the plane fixed area is an area where the environmental map is completed, if a new plane is detected in this area, it can be regarded as a false detection.

  Further, the obstacle detection means detects an obstacle from the detection frame, the obstacle registration means detects the obstacle in a plurality of successive detection frames, and registers the obstacle in the environmental map. The environmental map update means updates the environmental map with the detection frame having the determined obstacle, and the route planning means refers to the environmental map and plans a route that avoids the obstacle. Can do.

  Furthermore, when a new environment is detected in the plane determination area, the moving means can stop moving until the analysis of the new environment is completed. Thereby, it is possible to prevent an unreasonable route plan from being made, for example, when the plane is indefinite.

  Further, the detection means is a three-dimensional distance sensor, and a plane other than the floor surface has information about a height from the floor surface and a normal vector as attribute information thereof, and the route planning means includes the A route can be planned based on the attribute information.

  Further, the route planning means plans a route that moves on the plane when the direction of the normal vector is within a predetermined range and the height from the floor surface is not more than a predetermined value. Can do. Further, the route planning means can plan a route that avoids the plane when the direction of the normal vector is outside a predetermined range and the height from the floor surface is larger than a predetermined value. . If the detection plane has a slope larger than a certain value and the height is higher than a certain value, it is difficult to move on the detection plane, and a route that bypasses the detection plane can be planned.

  Furthermore, the environmental map generation means can generate the environmental map by converting the detection frame detected by the detection means into a global coordinate system by odometry. By using the global map, the recognition range can be expanded and the reliability of the landing position plan can be improved.

  A mobile route planning method according to the present invention. A detection step for detecting a floor surface around the moving body, an expected result of predicting an environmental state around the moving body according to the movement of the moving means included in the moving body, and a detection result detected in the detection step. Comparing, if the environmental conditions match, an environment map generating step for generating or updating an environment map around the moving body based on the detection result, and a route for planning the movement route of the moving means with reference to the environment map And a planning process.

  A path planning method for a moving body according to the present invention includes a detection step of detecting a floor surface around a moving body, a current detection result detected in the detection step, and a detection result detected before the present time. In comparison, if the environmental conditions around the moving body match, an environment map generating step for generating or updating an environmental map around the moving body based on the detection result, and the mobile body with reference to the environmental map A route planning step of planning a moving route of the moving means provided.

  An environmental map generation apparatus according to the present invention is an environmental map generation apparatus mounted on a mobile body, and includes an detection map that detects a floor surface, and an environmental map that generates an environmental map based on a detection result detected by the detection means. The environment map generating means compares the predicted result of predicting the environment around the moving body with the detection result detected by the detecting means according to the movement of the moving body, and matches Updates the environmental map based on the detection result.

  An environment map generation method according to the present invention is an environment map generation method that refers to a mobile object for route planning, and includes a detection step of detecting an environment around the mobile object by a detection unit, An environmental map generation step of comparing the predicted result of predicting the environment around the moving body with the detection result detected by the detection means according to the movement, and updating the environmental map based on the detection result if they match. It is what has.

  ADVANTAGE OF THE INVENTION According to this invention, the environment recognition apparatus which can prevent the erroneous detection of an environment, the environment recognition method, the mobile body carrying this, and the route planning method of a mobile body can be provided.

  Hereinafter, specific embodiments to which the present invention is applied will be described in detail with reference to the drawings. In this embodiment, the present invention is applied to a mobile robot that determines the movement route by grasping the state of the floor surface with a sensor and generating an environmental map.

  The mobile robot according to the present embodiment predicts a change in the surrounding environmental state measured by the sensor, compares the predicted environmental state change with the environmental state measured this time, and is the environment measured this time. The environment that is not in the state change that has been made is no longer adopted as an environmental map. In addition, the area measured by the sensor is divided into a fixed area near the robot and another exploration area, and the above-described cancellation is executed only for the fixed area. Further, when a new environment is measured in the fixed region due to the movement of the robot, the movement of the robot is prohibited until the analysis of the environment is completed. First, an example of such a robot will be described.

  FIG. 1 is a perspective view showing a robot according to an embodiment of the present invention. As shown in FIG. 1, in the robot 1, a head unit 2, two left and right arm units 3a and 3b, and two left and right leg units 4a and 4b are connected to a predetermined position of the trunk unit 5. . The robot 1 according to the present embodiment is configured to be movable by wheels built in the leg units 4a and 4b. The robot may be a biped or quadruped walking type robot.

  The head unit 2 includes an imaging unit 6 and can capture a predetermined range around the head unit 2. Moreover, the head unit 2 is provided with a microphone, a speaker, etc. (not shown), and can recognize or answer a call from the user. The head unit 2 is connected to the trunk unit 5 so as to be turnable in the left-right direction in a plane horizontal to the floor surface. The surrounding environment can be imaged.

  The arm units 3a and 3b and the leg units 4a and 4b are joint units included in the arm units 3a and 3b according to a predetermined control program by an arithmetic processing unit included in the control unit built in the trunk unit 5. The amount of driving of the wheels included in the leg units 4a and 4b is controlled, and the joint driving angle of each joint and the rotation angle of the wheel are determined, thereby taking a desired position and posture.

  FIG. 2 is a block diagram showing the robot according to the present embodiment. The robot 1 includes a control unit 101, an input / output unit 102, a drive unit 103, a power supply unit 104, an external storage unit 105, and the like.

  The input / output unit 102 includes a camera 121 such as a CCD (Charge Coupled Device) for acquiring surrounding images, one or a plurality of built-in microphones 122 for collecting surrounding sounds, and outputs audio to the user. A speaker 123 for performing a conversation, an LED 124 for expressing a response to the user, feelings, and the like, a sensor unit 125 including a touch sensor, and the like are provided.

  The drive unit 103 includes a motor 131 and a driver 132 that drives the motor, and operates the leg units 4a and 4b and the arm units 3a and 3b according to a user instruction. The power supply unit 104 includes a battery 141 and a battery control unit 142 that controls discharging and charging thereof, and supplies power to each unit.

  The external storage unit 105 includes a removable HDD, an optical disk, a magneto-optical disk, and the like, stores various programs and control parameters, and stores the programs and data in a memory (not shown) in the control unit 101 as necessary. To supply.

  The control unit includes a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), a wireless communication interface, and the like, and controls various operations of the robot 1. And this control part 101 produces | generates a route plan based on the environmental map generation module 12 which produces | generates an environmental map by analyzing the image | video acquired with the camera 121, for example according to the control program stored in ROM, and the produced | generated environmental map. A route planning module 13 to be performed, an action determining module 14 for selecting an action to be taken based on various recognition results, a voice recognition module 15 for performing voice recognition, and the like. In particular, in the present embodiment, the environment map generation module 12 generates an environment map in which erroneous detection is suppressed. Based on this, a route plan is created, an action is determined, and the drive unit 103 is controlled. Expresses more natural movement behavior.

  Next, the environment map generation module 12 mounted on the robot 1 according to the present embodiment will be described. FIG. 3 is a diagram showing an environment map generation apparatus according to the present embodiment. As shown in FIG. 3, the environment map generation device 200 can include a detection unit 18 including a three-dimensional distance sensor and the like, and an environment map generation module 12 as an environment map generation unit. The detection unit 18 detects the floor surface from an image obtained by photographing the surroundings of the robot. The environmental map generation module 12 compares the predicted result of predicting the environmental state around the robot 1 with the detection result detected by the detection unit 18 according to the movement means of the robot 1, in FIG. If the environmental conditions match, an environmental map around the moving body of the robot 1 is generated or updated based on the detection result. Alternatively, the environment map generation module 12 compares the current detection result detected by the detection unit 18 with the detection result detected before the present time, and if the environmental conditions around the robot 1 match, the detection result Based on the above, an environment map around the robot 1 is generated or updated.

  The environment map generation module 12 includes an obstacle detection unit 111, an obstacle registration unit 112, and an environment map update unit 113. The obstacle detection unit 111 detects a plane other than the floor surface from the detection frame detected by the detection unit 18. When the obstacle detection unit 111 detects the same plane other than the floor surface in a plurality of consecutive detection frames by the obstacle detection unit 111, the obstacle registration unit 112 is connected to the environment map or the environment map generation module 12 using the plane as a fixed plane. Register in the plane registration database 16. The environment map update unit 113 updates the environment map in the environment map database 17 with the detection frame having a confirmed plane.

  In the environmental map generation module 12, the detection unit 18 detects the floor surface and other planes. First, the plane detection method will be described. FIG. 4 is a diagram for explaining a plane detection method. As shown in FIG. 4A, a search is made as to whether or not there is a plane other than the floor surface from the image in the i frame obtained by capturing the traveling direction (front) of the robot 1. In this case, the flat surface 201 of the step A is detected in addition to the floor surface. In this case, the detection unit 18 includes height information from the floor surface and information on the normal vector of the plane as the plane attribute information. In the following (i + 1) frame, it is detected again whether there is a plane other than the floor. Here, whether or not it can be determined as the same plane 201 in the i frame and the (i + 1) frame is determined using the height and normal vector information as the attribute information. In addition, it is preferable not to use plane position information as attribute information. In other words, when the position is the center of gravity of the planar area as attribute information, it is affected by outliers and occlusion. Further, when a plane area is obtained and used as attribute information, it is necessary to calculate a connected area, and the processing becomes complicated. Similarly, as shown in FIG. 4C, whether or not the detected plane in the (i + 2) frame is the plane 201 is detected from the height from the floor and the normal vector.

  Next, the environment map stored in the environment map DB 17 will be described. FIG. 5 is a diagram illustrating a method for updating the environment map. The detection frame used by the detection unit 18 is an image expressed in a robot coordinate system (camera coordinate system or local coordinate system). The robot coordinate system is a coordinate system based on an object arranged in the 3D space, and is a coordinate system based on the object. Therefore, there is not one local coordinate. The robot 1 basically performs comparison with the past frame using the plane recognized in the robot coordinate system and its attributes, and if it is determined that the frame is highly reliable as a result of the comparison, the robot 1 is used as the robot coordinate. Convert from to global coordinate system. The global coordinate system is a coordinate system that represents the position of the three-dimensional space itself, and the coordinates specified by the global coordinates never move. For example, the plane 202 detected in the i frame (FIG. 5A) passes through the (i + 1) frame (FIG. 5B) and protrudes from the angle of view in the (i + 2) frame (FIG. 5C). ). In this case, the robot 1 cannot make a landing plan for the plane 202 without changing the robot coordinates. Therefore, as shown in FIG. 5D, the robot coordinates are transformed into global coordinates using odometry. With global coordinates, an image recognized once can be used, and the recognition range can be expanded. This improves the reliability of the landing position plan.

  Next, a method in which the environment map generation module 12 determines the plane by continuously tracking the plane will be described. FIG. 6 is a diagram illustrating a case where the plane is determined when the number of continuous tracking = 3. The environment map generation module 12 compares the current detection result detected by the detection unit 18 with the detection result detected before the present time, and if there is an obstacle and the position of the obstacle matches, The environmental map is updated based on the detection result. Specifically, when the height from the floor surface of the plane other than the floor surface and the normal vector thereof match, the environment map is updated based on this result.

  (I-1) Only the floor surface is recognized in the frame (FIG. 6A). It is assumed that the plane 203 at the step A is found in the i frame (FIG. 6B). Since the plane 203 is found, the obstacle detection unit 111 starts tracking the plane 203. It is assumed that the plane 203 is detected also in the (i + 1) frame (FIG. 6B), and the plane 203 is also detected in the (i + 2) frame (FIG. 6C). Here, the plane 203 is detected continuously for three frames from the i frame. For this reason, the obstacle registration unit 112 assumes that the plane 203 formed by the step A exists as an obstacle, and registers it as a confirmed plane in the plane registration DB 16 (FIG. 6F).

  As described above, when the obstacle registration unit 112 of the environment map generation module 12 determines that the same plane is included for a plurality of consecutive frames among the detection frames detected by the detection unit 18, it registers this in the plane registration DB 16, The environment map update unit 113 updates the environment map with the i to (i + 2) frames and stores it in the environment map DB 17 (FIG. 6 (g)).

  FIG. 7 is a diagram illustrating a case where a mistake occurs during continuous tracking of a plane. Similarly, the plane 203 is found in the i frame (FIG. 7A), and the plane 203 is traced in the (i + 1) frame (FIG. 7B). However, the plane 204 different from the plane 203 is detected in the (i + 2) frame (FIG. 7C). The plane 204 is a plane whose normal vector is different from that of the plane 203, for example. It is assumed that the plane 203 is found again in the next (i + 4) frame (FIG. 7D). Similarly, the plane 203 is also detected in the (i + 5) frame (FIG. 7E). In this case, at the time of the (i + 4) frame when the plane is detected again, the obstacle registration unit 112 registers the plane 203 as the fixed plane in the plane registration database 16 (FIG. 7 (g)). Further, the environment map update unit 113 converts the frames i, (i + 1), (i + 3), and (i + 4) in which the plane 203 is detected to global coordinates and updates the environment map (FIG. 7 (f)).

  Here, the obstacle registration unit 112 out of the frames detected by the detection unit 18, the current frame (current frame) ((i + 3) frame) and, for example, the previous frame ((i + 2) frame) detected before the present time. When the surrounding environmental conditions do not match, for example, when different planes are detected as in this example, the next detection frame (next frame) ((i + 4) frame) and the previous frame (( i + 2) frame) and / or the surrounding environmental conditions of the current frame ((i + 3) frame) and discard the mismatched detected frames. That is, in this example, the (i + 3) frame is discarded.

  In addition, the environment map generation module 12 predicts a state change in the environmental state around the robot 1 detected by the detection unit 18, such as the phase and posture of the robot 1, and the predicted environmental state and the environmental state measured this time The plane may be determined by comparing. If a plane is detected at the expected position, the plane tracking is further continued. For example, when a plane is detected at the expected position for three frames in succession, this can be determined as a definite plane. It is possible to reduce misrecognition by stopping the adoption of the environment to the global map due to the planned state change.

  Next, handling of the plane being tracked will be described. 8 and 9 are diagrams for explaining the handling of the plane being tracked. As shown in FIG. 8A, when the plane 205 is found in the i frame, the plane is not determined until the (i + 2) frame. For this reason, the landing position is planned on the plane 205 in the (i + 1) frame during the plane tracking, and the step is kicked. That is, in the (i + 2) frame in which the plane is fixed, it will already walk on the step.

  Therefore, when tracking is in progress and an undetermined plane exists ahead, as shown in FIG. 8 (b), for example, stepping is performed before the step at the time of (i + 1) frame. Thus, the plane is first determined. In this way, the route planning module 13 makes a route plan so as to step on the front of the indefinite plane. Then, when the plane is determined, a plan for overcoming or avoiding the obstacle (step) is made. The route planning module 13 checks the height of the plane from the attribute information of the plane 205 constituting the step, and if the height of the plane can be overcome, the route planning module 13 plans the route on the step, and if the height cannot be overcome. For example, a route that bypasses the step is planned (FIG. 9). Further, if the indefinite plane is not fixed even when stepping in front of the indefinite plane, that is, if the step cannot be recognized as a plane, it may be mapped to the global map as an obstacle, and a route plan that bypasses the step may be made.

  Next, the case where the recognition area is divided and the handling of the tracking plane is changed depending on the area will be described. FIG. 10 is a diagram illustrating a plane search area and a plane determination area set by the robot 1. Moreover, FIG. 11 is a flowchart which shows the environmental map update method concerning this Embodiment. As shown in FIG. 10A, the environment map generation module 12 sets a predetermined area adjacent to the robot 1 as the plane determination area 212, and sets a predetermined area on the side farther from the robot 1 than the plane determination area 212 as a plane. The search area 211 is set. Then, the environment map generation module 12 updates the environment map based on the detection result for the plane search area 212 of the detection unit 18. Note that the environment map of the robot is divided into grids of a predetermined size and assigned with a plane ID. The robot determines whether or not the grid has entered the plane determination area by determining what percentage of each grid belongs to the plane determination area.

  The environmental map generation module 12 extracts data other than the floor surface from the detection data from the detection unit 18 (step S1). Then, as shown in FIG. 10B, since the plane 213 starts to appear, the robot detects an obstacle (plane 213) (step S2). In this case, the plane 213 is an indefinite plane. Therefore, as shown in FIG. 10C, the robot 1 steps in front of the indeterminate plane 213 so as to make the indeterminate plane the final plane. As a result, the robot 1 can use the plane 213 as a fixed plane and make a route plan for passing over the plane 213 or bypassing the plane. That is, if an obstacle is detected, it is determined whether or not it is a plane (step S3). If it is a plane, it is determined whether or not it is a new plane (step S4). If it is a new plane, it is determined whether or not the area where the plane is found is a search area (step S5). If it is the search area, tracking of the plane is started (step S6). Then, as described above, a step is taken to determine the plane.

  On the other hand, when a new plane 214 is found in the plane determination area, the data is discarded as a false detection (step S7). That is, the plane determination area is an area where an environment map including a plane other than the floor surface has already been created by tracking three or more frames. Since no new plane is found in this area, if it is found, this is detected as a false detection and the data is discarded (FIG. 10 (e)). In such a case, the fixed plane 213 is detected in the subsequent frame.

  On the other hand, if the obstacle detected in step S3 does not include a plane, it is mapped to the global map as an obstacle (step S8). If the plane found in step S4 is a tracking plane, it is determined whether the tracking frame N is larger than a threshold, for example, 2 (step S11). If more frames than the threshold can be tracked, they are registered as a fixed plane in the plane registration database 16 (step S13) and mapped to the global map. If it is not the tracking plane in step S9, it is checked against the confirmed plane and mapped to the global map if necessary.

  As described above, the environment map generation module 12 tracks a plane when a plane is detected in the plane search area 211, and detects a new plane that is a plane other than the floor in the plane determination area 212. Discard. In addition, the obstacle detection unit 111 detects an obstacle from the detection frame, and the obstacle registration unit 112 detects the obstacle in a plurality of continuous detection frames, and registers the obstacle in the environment map. Then, the environment map update unit 113 updates the environment map (global map) with the detection frame having the confirmed obstacle. The route planning module refers to the environment map and plans a route that avoids obstacles, for example. Furthermore, when a new environment is detected in the plane determination area 212, the environment map generation module 12 causes the moving unit to stop moving until the analysis of the new environment is completed. Prevent you from planning your route.

  FIGS. 12 and 13 are diagrams showing detection frames actually acquired by the robot according to the related art and the present embodiment, respectively. In FIG. 12A, 301 indicates a floor surface, and 302 indicates a detected plane. As shown in FIG. 12B, a recognition error 303 occurs and is mapped as an obstacle. As shown in FIG. 12C, the plane 302 is traced again from the next frame. As a result, as shown in FIG. 12D, the global map becomes dirty due to the erroneous recognition.

  On the other hand, as shown in FIG. 13B, even if a recognition error 303 occurs in the same manner, the robot according to the present embodiment discards the new plane (303) in the plane determination area. . Accordingly, as shown in FIG. 13C, the plane 302 is traced again in the next frame, but unlike the conventional technique, the global map is not soiled due to a recognition error as shown in FIG. 13D.

  In the present embodiment, the predicted result of predicting the environmental condition around the robot 1 according to the movement of the moving means of the robot 1 is compared with the detection result detected by the detection unit 18, and when the environmental condition matches. Based on the detection result, an environment map around the robot is generated and updated. That is, since the plane is recognized only when the same plane is detected in a plurality of consecutive frames, obstacle detection due to erroneous recognition can be reduced. Also, by setting the area near the robot as the plane determination area and the outside as the plane search area, a new obstacle is detected due to a false detection in the plane determination area where the observation is completed and the environment map is completed. If this is the case, it can be discarded immediately. Furthermore, when a new plane is detected in the search area, the new plane can be reliably detected by tracking the plane using this as a tracking target.

  It should be noted that the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the scope of the present invention.

  For example, in the above-described embodiment, the hardware configuration has been described. However, the present invention is not limited to this, and arbitrary processing may be realized by causing a CPU (Central Processing Unit) to execute a computer program. Is possible. In this case, the computer program can be provided by being recorded on a recording medium, or can be provided by being transmitted via the Internet or another transmission medium.

1 is a perspective view showing a robot according to an embodiment of the present invention. It is a block diagram which shows the robot concerning embodiment of this invention. It is a figure which shows the environmental map production | generation apparatus concerning embodiment of this invention. It is a figure explaining the plane detection method of the environmental map production | generation apparatus concerning embodiment of this invention. It is a figure explaining the method to update the environmental map concerning embodiment of this invention. It is a figure which shows the case where it determines with a plane when the number of continuous trackings = 3. In the environmental map production | generation apparatus concerning embodiment of this invention, it is a figure which shows the case where a mistake generate | occur | produces during the continuous tracking of a plane. It is a figure for demonstrating handling of the plane currently tracked in the environmental map production | generation apparatus concerning embodiment of this invention. Similarly, it is a figure for demonstrating handling of the plane currently tracked in the environmental map production | generation apparatus concerning embodiment of this invention. It is a figure which shows the plane search area | region and plane fixed area | region which the environmental map production | generation apparatus concerning embodiment of this invention sets. It is a flowchart which shows the environmental map update method concerning embodiment of this invention. It is a figure which shows the detection frame actually acquired with the robot concerning the conventional. It is a figure which shows the detection frame actually acquired with the robot concerning embodiment of this invention. It is a figure explaining the problem in the conventional range sensor.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Robot 2 Head unit 3a Arm unit 3a, 3b Arm unit 4a, 4b Leg unit 5 Trunk unit 6 Imaging unit 12 Environmental map generation module 13 Path planning module 14 Action determination module 15 Speech recognition module 16 Plane registration database 17 Environmental Map Database 18 Detection Unit 101 Control Unit 102 Input / Output Unit 103 Drive Unit 104 Power Supply Unit 105 External Storage Unit 111 Obstacle Detection Unit 112 Obstacle Registration Unit 113 Environmental Map Update Unit 121 Camera 122 Built-in Microphone 123 Speaker 125 Sensor Unit 131 Motor 132 Driver 141 Battery 142 Battery control unit 200 Environmental map generation device 201-205, 213, 214 Plane 211 Plane search area 212 Plane determination area

Claims (26)

Transportation means;
Detecting means for detecting a floor surface around the moving body;
According to the movement of the moving means, the predicted result of predicting the environmental state around the moving body is compared with the detection result detected by the detecting means, and if the environmental state matches, the moving body is based on the detection result. An environment map generation means for generating or updating a surrounding environment map;
A moving body having route planning means for planning a moving route of the moving means with reference to the environment map. Transportation means;
Detecting means for detecting a floor surface around the moving body;
The current detection result detected by the detection means is compared with the detection result detected before the present, and when the environmental conditions around the moving body match, based on the detection result, the environmental map around the moving body Environmental map generation means for generating or updating
A moving body having route planning means for planning a moving route of the moving means with reference to the environment map. The environmental map generation means compares the current detection result detected by the detection means with the detection result detected before the present, and if there is an obstacle and the position of the obstacle matches, The moving body according to claim 1, wherein the environmental map is updated based on the detection result. The environmental map generation means compares the current detection result detected by the detection means with the detection result detected before the present time, and the height from the floor surface of the plane other than the floor surface and its normal vector The mobile body according to claim 1, wherein when the two match, the environment map is updated based on the detection result. The environment map generation unit updates the environment map with the detected frame when it is determined that the same plane is included for a plurality of consecutive frames among the detection frames detected by the detection unit. Item 3. A moving object according to item 1 or 2. The environmental map generation means compares the current frame, which is the current detection frame, among the detection frames detected by the detection means, and the previous frame detected before the present, and if the surrounding environmental conditions do not match 3. The moving body according to claim 1, wherein the next frame, which is the next detection frame, is compared with environmental conditions around the previous frame and / or the current frame, and the detection frames that do not match are discarded. The environmental map generation means includes
Obstacle detection means for detecting a plane other than the floor surface from the detection frame detected by the detection means;
When the obstacle detection unit detects the same plane other than the floor in a plurality of successive detection frames, the obstacle registration unit registers the plane as a fixed plane in the environmental map or the plane registration unit;
The moving body according to claim 1, further comprising: an environment map updating unit that updates the environment map using a detection frame having the fixed plane. The obstacle registration unit determines whether a plane other than the floor surface is detected in the detection frame by the obstacle detection unit with reference to the environment map, and determines whether the plane is a registered plane.
The mobile object according to claim 7, wherein the environmental map update unit updates the environmental map with a detection frame having the registered plane. The expected result is a conversion result obtained by converting a detection result before the present into the global coordinate system,
The mobile object according to claim 1, wherein the environmental map generation unit converts the current detection result into a global coordinate system and determines a match with the conversion result. The environment map generating means sets a predetermined area adjacent to the moving body as a plane fixed area, sets a predetermined area farther from the moving body than the plane determined area as a plane search area, and detects the detecting means. The moving body according to any one of claims 1 to 9, wherein the environmental map is updated based on a detection result of the planar search area. The moving body according to claim 10, wherein the environment map generation unit sets the plane fixed region based on observation results over a plurality of frames. The moving body according to claim 10 or 11, wherein the environmental map generation unit discards a new plane that is a plane other than the floor in the plane determination area. The obstacle detection means detects an obstacle from the detection frame;
The obstacle registration means detects the obstacle in a plurality of successive detection frames, registers the obstacle in the environmental map,
The environmental map update means updates the environmental map with a detection frame having the determined obstacle,
The mobile object according to claim 7 or 8, wherein the route planning means plans a route that avoids the obstacle by referring to the environmental map. 13. The moving device according to claim 10, wherein, when a new environment is detected in the plane fixed region, the moving unit stops moving until the analysis of the new environment is completed. Moving body. The detection means is a three-dimensional distance sensor,
The plane other than the floor has information about the height and normal vector from the floor as attribute information,
The mobile object according to claim 1, wherein the route planning unit plans a route based on the attribute information. The route planning means plans a route that moves on the plane when the direction of the normal vector is within a predetermined range and the height from the floor surface is a predetermined value or less. The moving body according to claim 15. The path planning means plans a path that avoids the plane when the direction of the normal vector is outside a predetermined range and the height from the floor surface is larger than a predetermined value. The moving body according to claim 15 or 16. The mobile object according to any one of claims 1 to 17, wherein the environmental map generation unit generates the environmental map by converting a detection frame detected by the detection unit into a global coordinate system by odometry. . A detection step for detecting a floor surface around the moving body;
According to the movement of the moving means included in the moving body, the predicted result of predicting the environmental state around the moving body is compared with the detection result detected in the detection step, and if the environmental state matches, the detection result An environment map generating step of generating or updating an environment map around the moving body based on
A route planning method for a moving object, comprising: a route planning step of planning a movement route of the moving means with reference to the environment map. A detection step for detecting a floor surface around the moving body;
The current detection result detected in the detection step is compared with the detection result detected before the present, and if the environmental conditions around the moving body match, based on the detection result, the surrounding of the moving body An environmental map generation process for generating or updating the environmental map of
A route planning method for a moving body, comprising: a route planning step of planning a moving route of a moving means included in the moving body with reference to the environment map. An environmental map generation device mounted on a mobile body,
Detection means for detecting the floor surface;
Environmental map generation means for generating an environmental map based on the detection result detected by the detection means,
The environment map generating means compares the predicted result of predicting the environment around the moving body with the detection result detected by the detecting means according to the movement of the moving body, and if they match, based on the detection result, An environmental map generator that updates environmental maps. The expected result is a conversion result obtained by converting a detection result before the present into the global coordinate system,
The environment map generation device according to claim 21, wherein the environment map generation unit converts the current detection result into a global coordinate system and determines a match with the conversion result. The environment map generation unit updates the environment map with the detected frame when it is determined that the same plane is included for a plurality of consecutive frames among the detection frames detected by the detection unit. Item 21 or 22 is an environment map generation device. The environmental map generation means includes an obstacle detection means for detecting a plane other than the floor surface from a detection frame detected by the detection means,
When the obstacle detection unit detects the same plane other than the floor surface in a plurality of successive detection frames, the obstacle registration unit registers the plane as a fixed plane in the environment map;
The environment map generation device according to any one of claims 21 to 23, further comprising: an environment map update unit that updates the environment map with a detection frame having the fixed plane. The obstacle registration unit determines whether a plane other than the floor surface is detected in the detection frame by the obstacle detection unit with reference to the environment map, and determines whether the plane is a registered plane.
The environment map generation device according to claim 24, wherein the environment map update means updates the environment map with a detection frame having the registered plane. A method for generating an environmental map to be referred to a mobile object for route planning,
A detecting step of detecting an environment around the moving body by a detecting means;
An environment in which the predicted result of predicting the environment around the moving object is compared with the detection result detected by the detecting means according to the movement of the moving object, and the environment map is updated based on the detected result if they match. An environmental map generation method comprising a map generation step.

Images