JP2769972B2 - Autonomous driving system for mobile robots on uneven terrain - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an autonomous traveling system for an irregular terrain mobile robot capable of carrying observation equipment and the like to a place where no one can access, such as a dangerous disaster area or a crater.

[0002]

2. Description of the Related Art In areas where natural disasters such as pyroclastic flows due to volcanic eruptions and debris flows due to torrential downpours may occur, the Ground Self-Defense Force is currently dispatched and armed vehicles are used by members to monitor and monitor. I have.

[0003] However, if a vehicle that can be remotely controlled can be assigned to the above-mentioned duties without a member getting on board, a surveillance and warning activity can be performed safely without exposing valuable human lives to the violence of nature.

In the field of unmanned dump trucks traveling on rough terrain outdoors, an invention has been known in which a traveling road surface is assumed to be a two-dimensional plane, and a dump route is taught to autonomously travel. I have.

[0005] With regard to autonomy of this kind of outdoor robot, a wheel type or a crawler type is used as a moving mechanism of the robot, and the action is about moving on a flat ground and avoiding obstacles. In other words, remote control or
Some functions have been automated.

[0006]

However, the above-mentioned system cannot cope with the monitoring of debris flows caused by pyroclastic flows and heavy rains associated with volcanic eruptions due to unpredictable violence of nature. In remote control, attendants may be involved in disasters.
Therefore, there is a need for the development of a safe, autonomous vehicle that requires only command and supervision at the rear point without attendants.

By the way, there are various types of moving mechanisms of a robot that moves on irregular terrain due to pyroclastic flows due to volcanic eruptions or debris flows due to heavy rain. In addition, the site where the autonomous disaster relief vehicle actually runs is not a flat road surface with only ruggedness like a dump truck runs,
It is a three-dimensional undulating natural disaster area with cliffs, rivers, slopes, mountains, valleys, etc. Moreover, since the area is a natural disaster area, the terrain is constantly changing, and new obstacles may be newly generated which are initially determined not to exist.
In order to guide a vehicle traveling on a road surface that changes in a fluid manner as described above, it is not possible to directly adopt the above-described guidance method in which a traveling road surface is assumed to be a two-dimensional plane and a planned route is taught. For this reason, there is a demand for the development of a device that can reliably travel in a dynamically changing natural disaster area.

Accordingly, the present invention has been made in view of the above-described problems, and has as its object to provide a safer and more unmanned environment even on a site where the road surface is an irregularly changing surface such as a natural disaster site. An object of the present invention is to provide an autonomous traveling system of an irregular terrain mobile robot that can reliably perform autonomous traveling.

[0009]

To achieve the above object, an autonomous traveling system for an irregular terrain mobile robot according to the present invention is provided with a moving mechanism having a motor, and a posture for traveling on a road surface while keeping a body level. A mobile robot capable of changing the shape, distance information, contact pressure information such as the hardness and shape of the ground, terrain irregularity information, and presence / absence information of a convex object, and the traveling mobile robot A detection unit mounted on the mobile robot, comprising: an acceleration sensor, an angular velocity, an inclination angle of a vehicle body, an inner field sensor for detecting a rotation angle and a load of the motor, and a detection unit mounted on the mobile robot. And a control device for controlling the moving mechanism in accordance with an action pattern determined based on the detection information of the autonomous control of the mobile robot. Based on shape information, height information of the terrain in front of the mobile robot is generated from a distance image in a relatively narrow field of view, and a boundary between a flat ground and an uneven ground / obstacle is edged based on graphic information by the detection unit. Extracting as information, performing a region division based on the edge information and the graphic information by the detection unit to detect a non-traveling region, performing object recognition based on the region division information and the graphic information by the detection unit, A first image processing unit that obtains information indicating the shape and type of the moving object moving around the mobile robot; and a distance image with a relatively wide field of view based on the distance information from the detection unit. Is divided into a concave area, a convex area, a flat ground area, and a blind spot area according to the height information obtained from, and a polygonal approximation is performed by drawing a boundary line between a travelable area and a non-travelable area in each of the divided areas. The said move A second image processing unit that approximates the polygon-approached terrain with a polyhedron using the terrain inclination information in the area due to the difference in distance from the bot's body to the terrain; In addition to generating a contact pressure signal such as hardness or shape of the ground for determining whether the mobile robot can pass or not, one-dimensionally in front of the mobile robot based on contact information with the ground by the detection unit. Signal generating means for generating a contact signal indicating unevenness of the terrain, a signal processing unit for generating a signal corresponding to the presence or absence of a convex object based on the presence or absence information of the convex object by the detection unit, and the first image processing unit Based on the graphic information and the object information, the distance image and the polyhedral approximation information by the second image processing unit, and the contact pressure signal by the signal generating means, the three-dimensional topographic shape, the hardness of the ground, the moving robot An environment recognizing unit that recognizes the environment of a moving object that moves around the vehicle, and recognizes the self-position of the mobile robot based on acceleration information and rotation angle information obtained by the detecting unit. A self-position recognition unit that corrects a position error using the rotation angle information and the load information as correction information to obtain three-dimensional position information of the mobile robot;
A global movement route is planned based on the starting point and the target point of the input map information, and the map information is input, or the environment information recognized by the environment recognition unit and the self-location recognition unit recognizes the environment information. With the input position information as input, a terrain information input / output unit that updates the map while matching on the map, and a knowledge database on terrain such as uneven terrain and how to cross the terrain, posture are stored in advance,
While generating a path for obstacle avoidance based on polygon approximate information by the image processing unit and information on the global movement path of the terrain information input / output unit, and planning the moving speed and direction of the mobile robot, The environment recognition unit infers a behavior pattern in accordance with the knowledge database when moving on a known uneven surface having a known way to cross from the environment information recognized by the environment recognition unit and the position information recognized by the self-position recognition unit. When moving on an unknown bumpy terrain that does not know how to cross from the environmental information recognized by the self-position recognition unit and the position information recognized by the self-position recognition unit, the behavior pattern of the mobile robot is newly planned and recognized by the environment recognition unit. Environmental information, the position information recognized by the self-position recognition unit, the behavior pattern at the time of avoiding an obstacle in accordance with the performance information of the moving mechanism, Behavior pattern of movement of the behavior patterns, known to have been found way beyond inferred of uneven land when,
An integrated control unit for selecting any one of the newly planned movement patterns of the movement of the unknown uneven terrain, and generating a direction / speed and attitude command according to the action pattern selected by the integrated control unit; A moving mechanism control unit that controls rotation and stop of a motor of the moving mechanism so as to unconditionally avoid uneven terrain and obstacles existing near the mobile robot.

[0010]

The external sensor in the detection unit mounted on the mobile robot obtains graphic information, distance information, contact pressure information such as the hardness and shape of the ground, terrain unevenness information, and presence / absence information of a convex object. The internal sensor in the detection unit detects the acceleration, angular velocity, inclination angle of the vehicle body, rotation angle of the motor, and load of the traveling mobile robot. When the starting point and the target point of the map information are set by the operator by the terrain information input unit and the general moving path is planned, the control device mounted on the mobile robot then outputs the detection information of the detecting unit. The mobile robot is autonomously controlled according to the behavior pattern determined based on the mobile robot. In determining the behavior pattern, the self-position recognition unit obtains the three-dimensional position information of the mobile robot, and the environment recognition unit obtains the three-dimensional topography, the rigidity of the ground, and the movement around the mobile robot. Recognizes the environment of a moving mobile object and generates a path for avoiding obstacles based on approximate polygon information obtained by image processing and information on a global moving path, and plans the moving speed and direction of the mobile robot. I do. And
An action pattern is inferred according to a knowledge database stored in advance when moving on a known bumpy land where the way to go is known from the environment information recognized by the environment recognition unit and the position information recognized by the self-position recognition unit. On the other hand, in the case of moving on an unknown bumpy land whose way of movement is unknown from the environment information recognized by the environment recognition unit and the position information recognized by the self-position recognition unit, the behavior pattern of the mobile robot is newly planned. The integrated control unit infers the behavior pattern when avoiding obstacles, the behavior pattern when moving on a flat ground, according to the environment information recognized by the environment recognition unit, the position information recognized by the self-position recognition unit, and the performance information of the moving mechanism. One of a behavior pattern of movement of a known uneven terrain with a known way of crossing and a behavior pattern of movement of a newly planned unknown terrain is selected. Then, the moving mechanism control unit generates a direction, speed and posture command according to the action pattern selected by the integrated control unit, controls the rotation and stop of the motor of the moving mechanism, and causes the mobile robot to travel autonomously.

[0011]

FIG. 1 is a functional block diagram showing an embodiment of an autonomous traveling system for an irregular terrain mobile robot according to the present invention, and FIG.
FIG. 2 is a detailed block diagram of the system.

The autonomous traveling system according to this embodiment includes a robot vehicle 1 as a tracked vehicle for disaster relief traveling on uneven terrain that changes fluidly such as a natural disaster area,
The control device 2 for autonomously controlling the robot vehicle 1 is roughly configured.

Here, the irregular terrain in the embodiment is defined by topography and obstacles. The terrain includes soil / rock quality determined by the type of component particles, water content and wet density, inclination angle, ground surface Obstacles are defined by the curvature of the forest, the shape determined by the rate of change of the surface irregularities, and the obstacles are forests, rocks, rivers, lakes, marshes, grasses,
It is defined by natural obstacles caused by ice and artificial obstacles caused by natural obstacles, cities, artifacts and minefields.

The robot vehicle 1 comprises a tracked vehicle having a four-wheel track frame structure having four crawlers on the front, rear, left and right sides of a vehicle body in which a change in posture is introduced. More specifically, in the robot vehicle 1, the vehicle body is leveled,
A moving mechanism 4 for traveling on a road surface such as uneven terrain, a detecting unit (detecting means) 5 mounted on the vehicle body for recognizing surrounding conditions and a traveling state of the vehicle body, and a control device 2. Have been.

The moving mechanism 4 comprises four moving motors 4a for moving the respective crawlers in the front-rear direction and four rotating motors 4b for rotating the respective crawlers within a range of 360 ° around a fixed axis. Have two motors. Then, the robot vehicle 1 is configured such that the motors 4a and 4b are independently controlled based on a command from the control device 2, and travel while effectively gripping an uneven road surface in a natural disaster area.

The robot vehicle 1 may be any moving body that travels on a road surface.
And the like are optional.

The detection unit 5 mounted on the robot vehicle 1 includes:
It recognizes the surrounding situation based on various information such as environment, self-position, map, and vehicle body.
It is composed of As shown in FIG. 2, for example, an external sensor 7 is a CCD (ch) for obtaining graphic information and distance information.
arge coupled device) 7a and an LRF (Laser Range Fi) for obtaining distance information from reflection by emission of laser light.
nder) 7b, a tactile sensor 7c for obtaining contact pressure information such as the hardness and shape of the ground, and terrain unevenness information, and an ultrasonic sensor 7d for obtaining information on the presence or absence of a convex object. I have.

As shown in FIG. 2, for example, the inner field sensor 8 is composed of an inertia sensor 8A and a body control sensor 8B. The inertia sensor 8A is an accelerometer 8Aa for detecting the acceleration of the traveling robot vehicle 1. And a FOG (Fiber Optic) for detecting the angular velocity of the traveling robot vehicle 1.
cal Gyro) 8Ab and an inclinometer 8Ac for detecting the inclination angle of the vehicle body. Also, the vehicle control sensor 8
B is an encoder 8Ba for detecting the rotation angle of the motors 4a and 4b for controlling the rotation of each crawler;
and a motor load meter 8Bb for detecting the loads a and 4b.

The control device 2 includes a CCD image processing unit 9, an LR
F image processing unit 10, haptic mechanism unit 11, ultrasonic signal processing unit 12, environment recognition unit 13, self-position recognition unit 14, terrain information input / output unit (setting means) 15, integrated control unit 16, moving mechanism control unit ( (Movement control means) 17,
The components are electrically connected to each other via a communication interface 18 such as RS232C or Ethernet. Here, the CCD image processing unit 9 and the LRF image processing unit 1
0, a tactile mechanism section 11, an ultrasonic signal processing section 12, an environment recognition section 13, a self-position recognition section 14, and an integrated control section 16 constitute an action pattern determination means 2a.

Each component will be further described with reference to FIG. 2. The CCD image processing unit 9 includes a distance image generation unit 9
a, edge extracting means 9b, area dividing means 9c, and pattern recognizing means 9d. The distance image generation means 9a generates high-resolution height information of the terrain in front of the robot vehicle 1 from the distance image in a relatively narrow field of view based on the graphic information from the CCD 7a. Edge extraction means 9
“b” extracts a boundary between a flat land and a bumpy land / obstacle, which is a portion where the terrain changes, based on the graphic information from the CCD 7a at high speed as edge information. The area dividing means 9c is provided with the edge information extracted by the edge extracting means 9b and the CCD 7a.
Area is divided for each pattern by a predetermined texture on the basis of the graphic information from, and a running impossible area is detected.
The object recognizing means 9d performs object recognition based on the area division information obtained by the area dividing means 9c and the graphic information from the CCD 7a, and outputs a moving object as scene information such as a moving object moving around the robot vehicle 1. Information on shape and type is obtained.

The LRF image processing unit 10 includes a distance image generating unit 10a, a region dividing unit 10b, a polygon approximating unit 10
c, a polyhedron approximating means 10d. The distance image generation unit 10a generates a distance image with a relatively wide field of view based on the distance information of the LRF 7b. The area dividing means 10b performs coordinate conversion on the distance information of the LRF 7b, and converts the distance information into four areas of a concave area, a convex area, a flat ground area, and a blind spot area by height information obtained from the distance image of the distance image generating means 10a. It is roughly divided. The polygon approximating unit 10c approximates each of the four regions divided by the region dividing unit 10b with a polygon by drawing a boundary line between a region that can travel and a region that cannot travel. Polyhedron approximation means 10
d is a polyhedral approximation of the terrain polygonally approximated by the polygon approximation means 10c using the inclination information of the terrain in the area due to the difference in the distance from the vehicle to the terrain.

The tactile mechanism 11 includes a contact pressure signal generating means 1
1a and a contact signal generating means 11b. The contact pressure signal generating means 11a generates a contact pressure signal such as hardness or shape of the ground for determining whether or not the robot vehicle 1 can pass based on the contact pressure information of the tactile sensor 7c, If it can pass, the ON contact pressure signal
When the passage is impossible, the contact pressure signal is turned off. The contact signal generating means 11b is provided with a tactile sensor 7c.
A high-speed contact signal indicating one-dimensional terrain unevenness in front of the robot vehicle 1 is generated based on contact information with the ground by the robot. Generates off contact signals.

The ultrasonic signal processing unit 12 includes the ultrasonic sensor 7
If there is a convex object based on the convex object presence information from d, a convex object signal is generated by a fuzzy vector expressing the direction and distance as an ambiguous vector, and an off convex object signal is generated if there is no convex object. doing.

The environment recognition unit 13 includes environment recognition means 13a,
Caution generation means 13b, sensor selection control means 13c
It is provided with. The environment recognition means 13a uses CC
Based on the distance image and scene information from the D image processing unit 9, the distance image from the LRF image processing unit 10, the polyhedral approximation information, and the three-dimensional topographic shape based on the contact pressure signal from the haptic mechanism unit 11,
It recognizes the hardness of the ground and the environment such as a moving object moving around the robot vehicle 1. Caution generating means 13b
Error is detected based on the edge information from the CCD image processing unit 9, the presence or absence of a convex object from the LRF image processing unit 10, the tactile mechanism unit 11, or the on / off signal indicating the irregularities of the terrain from the ultrasonic signal processing unit 12. In order to avoid unnecessary stopping due to recognition, to detect the true change from flat ground to uneven terrain / obstacles, to generate a caution fuzzy vector with the direction and risk level as parameters, to improve the detection ability A caution signal is being generated. The sensor selection control means 13c includes:
In addition to selecting an external sensor 7 necessary for obtaining information that is insufficient in the environment recognized by the environment recognizing unit 13a, the controller controls a mechanism of the selected external sensor 7, for example, a camera platform, a visual field, and a zoom.

The self-position recognizing unit 14 recognizes the self-position of the robot vehicle 1 based on the acceleration information from the accelerometer 8Aa and the rotation angle information from the FOG 8Ab, and the tilt information of the inclinometer 8Ac, the encoder 8Ba, and the motor. Load meter 8
A signal from Bb is used as correction information, and a position error caused by a drift of the inertial sensor 8A is corrected based on the correction information to obtain highly accurate three-dimensional position information.

The terrain information input / output unit 15 includes a global path generating unit 15a and a map updating unit 15b. When the starting point and the target point are set in the ambiguous map information including information missing or error input by the operation of the operator,
The position and speed between the starting point and the target point are represented by points (× in FIG. 3).
(Corresponding to a mark), and interpolation is performed between the points (corresponding to the broken line in FIG. 3) in order to perform a smoother traveling, thereby planning a global movement route. The map updating unit 15b receives the map information input by the operation of the operator as input, or receives the environment information recognized by the environment recognizing unit 13a and the position information recognized by the self-position recognition unit 14 on the map. The map is updated while matching.

The integrated control section 16 includes a local path generating means 16a, a database storing means 16b, an action pattern inference means 16c, an action pattern generating means 16d, and an action pattern selecting means 16e. The local path generation unit 16a generates a path for obstacle avoidance based on the polygon approximation information from the polygon approximation unit 10c and the global path information based on the global path input from the global path generation unit 15a. The moving speed and direction of the robot vehicle 1 are planned. The database storage unit 16b stores a knowledge database that is preliminarily made into a database on terrain such as uneven terrain, how to cross the terrain, and posture. More specifically, the database storage means 16b includes a moving mechanism database in which the running performance of the moving mechanism 4 is described, an irregular terrain database in which an irregular terrain model is described, and a possible attitude of the moving mechanism 4. Is stored, and an action pattern database expressed by a state transition diagram in which the action on the terrain is one attitude and the action operator is the attitude transition. The action pattern inference means 16c stores the information in the database storage means 16b when moving on a known uneven surface whose way of passing is known from the environment information recognized by the environment recognition means 13a and the position information recognized by the self-position recognition unit 14. Inference for selecting an action pattern according to the obtained action pattern database. The behavior pattern generation means 16d includes:
When the user moves on an unknown bumpy terrain that cannot be determined from the environment information recognized by the environment recognition unit 13a and the position information recognized by the self-position recognition unit 14, the posture is determined using the posture database stored in the database storage unit 16b. The behavior pattern of the robot vehicle 1 such as the change, speed, direction, etc., is newly planned. The action pattern selecting unit 16e is configured to perform an action pattern when avoiding an obstacle and an action when moving on a flat ground according to the environment information recognized by the environment recognition unit 13a, the position information recognized by the self-position recognition unit 14, and the performance information of the moving mechanism 4. Either the movement pattern of the known uneven terrain that is known to be crossed inferred by the pattern or action pattern inference means 16c, or the movement pattern of the unknown uneven terrain newly planned by the action pattern generation means 16d Have made a choice.

The moving mechanism control unit 17 includes a vehicle body monitoring unit 17
a, command means 17b, and running / posture control means 17c. The vehicle body monitoring means 17a includes an on / off signal from the tactile mechanism section 11 indicating irregularities of the terrain, an on / off signal from the ultrasonic signal processing section 12 indicating on / off of a convex object, an inclinometer 8Ac, and a body control section. Based on the signal from the sensor 8B, it monitors whether the robot vehicle 1 is operating normally. The command unit 17b generates a command signal for each of the direction, speed, and posture according to the action pattern selected by the action pattern selection unit 16e. When a caution signal is input from the caution generation unit 13b, the reaction vector generated by the caution fuzzy vector from the caution generation unit 13b and the planned movement path are combined to generate a vicinity of the robot vehicle 1. Each direction, speed and attitude command signals are generated so as to unconditionally avoid uneven terrain and obstacles existing on both sides of the vehicle body (for example, on both sides of the vehicle body). Is automatically stopped. The running / posture control means 17c is configured to control the robot vehicle 1 in accordance with each command signal from the command means 17b and each signal from the vehicle body monitoring means 17a.
Of the eight motors 4a, 4b of the moving mechanism 4 at
The robot controls the stop independently, for example, on a flat ground, keeps the car body horizontal by turning the crawler around a fixed axis to change the posture, thereby keeping the car body horizontal. The running control of the vehicle 1 is performed smoothly.

In the autonomous traveling system configured as described above, when the power is turned on, map information is input from the terrain information input / output unit 15, and a global moving route connecting the starting point and the target point is determined. The environment recognizing unit 13 and the self-position recognizing unit 14 determine the environment, based on various information obtained by the detection signal from the detecting unit 5 mounted on the robot vehicle 1.
Recognize the self-position and the vehicle state. Then, the integrated control unit 16 generates and selects an action pattern based on the recognized information on the environment, the self-position, and the vehicle body, and outputs a command signal. The moving mechanism control unit 17 controls the direction, speed, and attitude of the robot vehicle by a command signal associated with the selected action pattern, and causes the robot vehicle 1 to autonomously travel.

Further, the autonomous running operation of the robot vehicle according to the change in the terrain from the starting point to the target point will be described with reference to FIG. FIG. 3 shows an example of the autonomous traveling operation of the robot in the above system. When the robot vehicle 1 autonomously travels on the movement route, a caution is generated when an obstacle exists on or near the movement route, and when there is uneven ground on the movement route, and the robot vehicle 1 The direction, speed and attitude to be taken are controlled.

First, in section A, when a caution is generated by detecting the presence of an obstacle near the moving route, a reaction vector generated by a caution fuzzy vector using the direction and the degree of danger as parameters and the movement vector of the moving route are determined. The command value is a vector obtained by combining the vector and
The direction, speed, and attitude of the robot vehicle 1 are controlled in accordance with the command values, and the robot vehicle 1 autonomously travels by a reaction action to avoid obstacles unconditionally.

Next, in section B, when a caution is generated upon detecting the presence of an obstacle on the moving route,
A moving path for obstacle avoidance is generated, and the direction, speed, and attitude of the robot vehicle 1 are controlled according to the moving path,
The robot vehicle 1 runs autonomously so as to avoid obstacles.

Next, in section C or section D, when a caution is generated by detecting the presence of a bumpy land (super levee or super bunker) which is known to cross over on the moving route, various cautions are stored in advance as a database. Of the robot vehicle 1 is selected next, and the direction, speed and attitude of the robot vehicle 1 are controlled in accordance with the selected behavior pattern.
Self-driving over a known bumpy super-bank or super-trench.

Next, in section E, when a caution is generated by detecting the presence of an unknown uneven land due to a moving object or a soft land on the moving route, a new behavior pattern is generated by itself.
The direction, speed, and attitude of the robot vehicle 1 are controlled according to this behavior pattern, and the robot vehicle 1 travels autonomously over unknown uneven terrain.

Therefore, according to the autonomous traveling system configured as described above, when the starting point and the target point are determined on the map, the robot vehicle 1 is autonomously operated according to the change in the terrain without the intervention of the operator. You can run.
In addition, since the moving route is corrected along with updating the map in parallel with the autonomous traveling of the robot vehicle 1, the posture can be improved even in a dangerous site such as a natural disaster area where the road surface is a fluidly changing irregular surface. The robot vehicle 1 in which the change is introduced can autonomously and safely travel autonomously, and labor and labor can be saved in simple tasks (for example, landmine processing, detection, transportation, reconnaissance, etc.).

When the robot vehicle 1 travels, since a true change from a flat ground to a bumpy terrain / obstacle is detected, only when there is an obstacle on the moving route in traveling on a flat terrain. Alternatively, even when an obstacle is present near the moving route, the robot vehicle 1 can be autonomously driven to avoid the obstacle without needlessly stopping.

Further, since the posture of the robot vehicle 1 is always kept horizontal while traveling, it is possible to carry out traveling while holding a working device such as a manipulator mounted on the robot vehicle 1 in a stable state. it can.

Furthermore, when there is a change in the terrain such as uneven terrain, the behavior pattern of the terrain that can be crossed by the robot vehicle 1 is stored in advance as a database, and the behavior pattern is generated for unknown terrain by itself. Therefore, no special input operation is required, the work load on the operator can be reduced, and the robot vehicle 1 can travel according to the terrain.

[0039]

As described above, according to the autonomous traveling system of the mobile robot for rough terrain according to the present invention, the following effects can be obtained. (1) When the starting point and the target point are determined on the map, the mobile robot can be made to travel autonomously according to the change in the terrain without the intervention of the operator. In addition, since the traveling route is corrected along with updating the map in parallel with the autonomous traveling of the mobile robot, the posture change can be made even on dangerous sites such as natural disaster areas where the road surface changes dynamically. In this way, the mobile robot incorporating the system can autonomously and safely travel autonomously, and can save labor and labor in simple operations such as landmine processing, detection, transportation, and reconnaissance. (2) When the mobile robot travels, it detects a change from a flat ground to a bumpy terrain / obstacle. Even if there is an obstacle in the vicinity of the vehicle, the mobile robot can autonomously travel while avoiding the obstacle without needlessly stopping the mobile robot. (3) When there is a change in the terrain such as uneven terrain, the behavior pattern of the terrain that can be passed by the mobile robot is stored in advance as a database, and the unknown terrain generates its own behavior pattern. This eliminates the need for a simple input operation, reduces the work load on the operator, and enables the mobile robot to travel according to the terrain.

[Brief description of the drawings]

FIG. 1 is a functional block diagram showing an embodiment of an autonomous traveling system for an irregular mobile robot according to the present invention.

FIG. 2 is a detailed block diagram of the system.

FIG. 3 is a diagram showing an example of an autonomous traveling operation of the robot in the system.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Robot vehicle, 2 ... Control device, 2a ... Behavior pattern determination means, 4 ... Moving mechanism, 5 ... Detection part (detection means), 7
... external sensor, 8 ... internal sensor, 9 ... CCD image processing unit, 10 ... LRF image processing unit, 11 ... tactile mechanism unit, 12
... Ultrasonic signal processing unit, 13 ... Environment recognition unit, 14 ... Self position recognition unit, 15 ... Terrain information input / output unit (setting means), 16
... Integrated control unit, 17 ... Moving mechanism control unit (movement control means).

Continuation of front page (58) Field surveyed (Int.Cl. ⁶ , DB name) G05D 1/02

Claims (1)

(57) [Claims] 1. A vehicle having a moving mechanism having a motor,
The posture can be changed so as to run on the road surface while keeping it level
Mobile robot and contact information such as figure information, distance information, ground hardness, shape, etc.
Information to obtain information about the terrain
Field sensor, acceleration and angular velocity of the traveling mobile robot
Angle, body tilt angle, motor rotation angle and load.
To the mobile robot with an internal sensor for
A detection unit mounted on the mobile robot;
The moving mechanism according to the action pattern determined based on the
A control device for controlling the mobile robot by autonomous control.
The control device has a ratio based on the graphic information by the detection unit.
From the distance image in a relatively narrow field of view,
Generates terrain height information and converts it into graphic information by the detection unit.
Edge information on the boundary between flat ground and uneven ground / obstacle based on
As the edge information and the graphic information
Segmentation is performed based on the information to detect unrunnable areas
Based on the area division information and the graphic information from the detection unit.
And move around the mobile robot.
Image processing for obtaining information indicating the shape and type of a moving object to be moved
And a relatively wide field of view based on the distance information from the detection unit.
Of the distance image, and height information obtained from the distance image
Divided into concave, convex, flat ground, and blind spot areas
In each of the divided areas, the area that can run and the area that cannot run
Polygonal approximation by drawing the boundaries of
Of the terrain in the area due to the difference in the distance from the vehicle body to the terrain
Using the slope information, approximate the terrain approximated by the polygon to a polyhedron
A second image processing unit similar to the moving robot based on contact pressure information from the detection unit;
The shape and hardness of the ground to determine whether
It generates the contact pressure signal such Jo, the detection unit
Based on the contact information with the ground
A signal generator that generates a contact signal indicating one-dimensional terrain irregularities
Generating means for detecting whether a convex object exists
A signal processing unit that generates a signal according to the absence, graphic information and object information by the first image processing unit,
The distance image and the polyhedron approximate information by the second image processing unit
Information and a contact pressure signal by the signal generation means.
Dimensional terrain shape, ground hardness, around the mobile robot
An environment recognition unit for recognizing the environment of the moving moving object; and an acceleration recognition unit and a rotation angle information obtained by the detection unit.
To recognize the self-position of the mobile robot,
Tilt information, motor rotation angle information and load information.
Correcting the position error as the correct information,
A self-position recognition unit that obtains dimensional position information, and globally based on the starting point and target point of the input map information
Planning the travel route and entering the map information
Or the environment information recognized by the environment recognition unit and
The position information recognized by the self-position recognition unit is used as an input.
Update the map while matching on the map
Terrain information input / output unit, and knowledge of terrain such as uneven terrain, how to cross the terrain, and attitude
Knowledge database is stored in advance, and the image processing unit
Global approximation of polygon approximation information and the terrain information input / output unit
Generates a route for obstacle avoidance based on the information on the travel route
To plan the moving speed and direction of the mobile robot.
And the environment information recognized by the environment recognition unit and
From the position information recognized by the self-position recognition unit, the way to cross
When moving on known uneven terrain,
The environment recognition unit infers an action pattern according to the database.
The environment information recognized by the
On unknown bumpy terrain where it is not clear how to cross from the location information
New behavior pattern of the mobile robot
Environment information recognized by the environment recognition unit,
Position information recognized by the position recognition unit, performance information of the moving mechanism
According to the behavior pattern when avoiding obstacles and when moving on flat ground
Known behavior patterns, inferred ways to cross
Movement pattern of uneven terrain, newly planned unknown
Any of the movement patterns of the movement pattern on uneven terrain
An integrated control unit for selecting a direction, and a direction /
Generates speed and attitude commands to generate a
Before unconditionally avoiding uneven terrain and obstacles that exist next to you
Moving mechanism control to control rotation and stop of motor of moving mechanism
Autonomous system of uneven terrain robot, characterized by comprising a control unit.