JPH05285868A - Walking control device for leg type moving robot - Google Patents

Abstract

PURPOSE:To perform stable walking by a method wherein the walking mode of a leg type moving robot having a plurality of leg part links is classified into a plurality of walking modes, a target value is described through combination of the walking modes, and the drive control value of the leg part link is decided based on the target value. CONSTITUTION:A square sensor 32 comprising a CCD camera is arranged to the base body 24 of a two-feet walking leg type moving robot 1 having right and left leg part links 2 and an output therefrom is inputted to a control unit 26 through an image processing unit 34. In which case, evaluation is effected by means of a first evaluation function for throttling the whole shape feature point of a whole object where environment acknowledge information is hierachically described, a list of a feature point candidacy is produced in the order of the evaluated results, and based on the list, walking plan processing is performed. Moving operation of a robot is divided into five kinds of straight advancing, walking up and down stairs, conversion of a direction, regulation of a walking width (in a longitudinal direction), and a regulation of a walking width in a lateral direction with a stop therebetween. Through combination of the operations, walking operation is executed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a walking control device for a legged mobile robot, and more particularly, to a legged mobile robot that moves in a predetermined environment to facilitate its walking control.

[0002]

2. Description of the Related Art Recently, a legged mobile robot such as that disclosed in Japanese Patent Laid-Open No. 62-97005 has been proposed.

[0003]

When such a legged mobile robot actually walks, it is necessary to control walking so that it can move while avoiding obstacles in an environment where there are many obstacles. Actually, it is not easy to realize stable walking control even in such an environment.

SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to provide a walking control device for a legged mobile robot capable of performing stable walking control even in an environment where many obstacles are present.

[0005]

In order to solve the above-mentioned problems, the present invention provides a walking control device for a legged mobile robot comprising a plurality of leg links, as set forth in claim 1, for example. The mode is classified into a plurality of walking patterns,
The target value is described by a combination of the plurality of walking patterns, and the drive control value of the leg link is determined to be the target value.

[0006]

[Function] The walking mode is classified into a plurality of walking modes, the target value is described by the combination, and the drive control value of the leg link is determined so as to be the target value, thereby moving in a complicated environment. At this time, the description of the target route becomes clear, control becomes easy, and stable walking control can be realized.

[0007]

Embodiments of the present invention will be described below. FIG. 1 is an explanatory skeleton diagram generally showing a walking control device for a legged mobile robot according to the present invention, in which each of the left and right leg links 2 has six joints (axes) (for convenience of understanding). Each joint (axis) is exemplified by an electric motor that drives it). The six joints (axes) are, in order from the top, joints (axes) 10R and 10L for rotating the legs of the waist (R on the right side, L on the left side).
And The same shall apply hereinafter), joints (axis) 12R, 12L in the waist pitch direction (around the y axis), joints (axis) 14R, 14L in the same roll direction (around the x axis), and joints (axis) 16R in the knee pitch direction. , 16L, pitch direction joints (axes) 18R, 18L, and roll direction joints (axis) 20
R, 20L, with a foot (foot) 2 below it
2R and 22L are attached, a base body (body portion 24) is provided at the uppermost position, and a control unit 26 equipped with a microcomputer is housed therein, as will be described later. In the above, the hip joint is a joint (axis) 10R
(L), 12R (L), 14R (L), and the ankle joint is composed of joints (axes) 18R (L), 20R (L). Further, the hip joint and the knee joint are connected by thigh links 28R and 28L, and the knee joint and the ankle joint are connected by lower leg links 30R and 30L.

The leg link 2 is given 6 degrees of freedom for each of the left and right feet, and these 6 × 2 = 1 during walking.
By driving each of the two joints (axes) at an appropriate angle, it is possible to give the desired movement to the entire foot, and
It is constructed so that it can walk in a dimensional space. As described above, the above-mentioned joint is composed of an electric motor, and is further provided with a speed reducer that boosts its output, and the details thereof are described in the application previously proposed by the present applicant (Japanese Patent Application No. 1-324221).
No. 8, JP-A-3-184782) and the like, and since they are not the gist of the present invention, further description thereof will be omitted.

Here, the base body 24 of the robot 1 shown in FIG.
One visual sensor 32 (monocular), which is a known CCD (solid-state image device) camera, is arranged in the camera, and its output is sent to an image processing unit 34, which is a microcomputer. Further, a well-known 6-axis force sensor 36 is provided on the ankle, and x, y, z transmitted to the robot via the foot
The force components Fx, Fy, Fz in the direction and the moment components Mx, My, Mz around the direction are measured to detect the presence or absence of landing of the foot and the magnitude and direction of the force applied to the supporting leg.
In addition, a pair of tilt sensors 40, 42 is provided on the upper portion of the base body 24.
Is installed, and the inclination with respect to the z axis in the xz plane and its angular velocity and angular acceleration, as well as the inclination with respect to the z axis in the yz plane and its angular velocity and angular acceleration are detected. Further, the electric motor of each joint is provided with a rotary encoder for detecting the amount of rotation thereof (only an electric motor for the ankle joint is shown in FIG. 1). These outputs are sent to the control unit 26 described above.

FIG. 2 is a block diagram showing the details of the control unit 26, which is composed of a microcomputer. The outputs of the tilt sensors 40 and 42 are A
It is converted into a digital value by the / D converter 50, and its output is sent to the RAM 54 via the bus 52. The output of the encoder arranged adjacent to each electric motor is input to the RAM 54 via the reversible counter 56. In addition, the keyboard and mouse for the user to input the target position etc. 5
8 is provided (not shown in FIG. 1), and its output is sent into the microcomputer through the user interface 60 composed of the microcomputer. Further, the output of the above-mentioned image processing unit 34 is likewise taken into the microcomputer through the communication interface 62. An arithmetic unit 64 is provided in the control unit, and based on the environment map stored in the ROM 66, the current position is recognized as described later in detail to determine the walking control value and the reversible counter 56. Calculate the speed command value of the electric motor from the deviation from the measured value sent from
It is sent to the servo amplifier via the / A converter 68.

Next, the operation of this control unit will be described.

FIG. 3 is a functional block diagram showing the operation focusing on the environment recognition processing. As shown, the visual sensor 3 is based on the environment map stored in the ROM 66.
The movement control is performed while recognizing the current position from the image information obtained through 2.

The details will be described below with reference to the flow chart of FIG. 4 (PAD diagram (structured flow chart)).

First, in S10, an approximate value of the initial current position in the moving environment, that is, the position and direction in which the robot 1 is first standing is input, and then in S12, the moving target, that is, the position of the target point and the robot. Enter the direction and. This input is performed through the keyboard, mouse 58 and user interface 60 shown in FIG.

In this embodiment, it is assumed that the moving environment is an indoor environment as shown in FIG. That is, the robot goes up and down the stairs according to the given command, selects a route for passing between the two pillars and reaching the target point,
The movement operation is executed with reference to the route. As shown in FIG. 6, the environment map is hierarchically described from the highest level to the second level to the third level. The highest level shows the connection relationship within a building and is described as the connection of areas. The area consists of rooms and links. Here, a room means a space that can be considered as a group of areas such as a room or a corridor, and a link means a place such as a staircase or a door that must pass through to move between rooms. Below the highest level, object placement information is described as a second level, and further below that, information on an inaccessible area, shape feature information, etc. for each object are described. The details are shown in FIGS. 7 and 8. Such information is stored in the ROM 66 as an environment map. Note that the following description will be limited to the movement within the room 1 shown in FIG.

Next, in the flow chart of FIG.
Proceed to 14 to perform route planning processing. Legged mobile robot,
Particularly in the case of bipedal walking, it is not easy to freely control the traveling direction such as traveling a designated distance or moving on a designated route. However, in movement control, realization of such movement is essential. Therefore, in this embodiment, the moving operation is divided into several basic walking functions with "stop" interposed therebetween and executed. This facilitates the calculation of the constraint conditional expression, and the movement control can be realized only by the control processing for executing the basic walking function. As shown in FIG. 9, there are five basic walking functions: (a) going straight, (b) going up and down stairs, (c) changing direction, (d) step adjustment (front and back), and (e) step adjustment (left and right). ..
At the time of the route planning processing in S14, the basic walking function is expressed as described above, and the trajectory is not directly expressed.

Specifically, the route planning process in S14 generates an essential passing point as shown in FIG. 10 (S16),
This is to generate an obstacle avoidance route as shown in FIG. 12 while considering interference with the obstacle as shown in FIG. 11 (S18). Here, the essential passing points are the position and direction at the stairs target point A, stairs starting point B, and staircase area exit point C. FIG. 13 shows this in detail.

In the flow chart of FIG.
In step 20, the movement control to the stairs target point (point A) is performed.

This will be described with reference to the flow chart of FIG. 14 showing the subroutine. First, in S100, the current position is recognized. Specifically, first, preparation for selection processing is performed in S102. FIG. 15 is a flow chart showing the subroutine, and the preparation for selection processing is S10.
As described in 20, S1021, preparation for the visual field calculation of the camera (visual sensor 32) at the estimated current position is performed. It is shown in FIG.

In the flow chart of FIG. 14, the process then proceeds to S104 to narrow down the shape feature points. FIG. 17
Is a flow chart showing the subroutine. The shape feature point narrowing-down process is performed in order to calculate shape feature point candidates from the shape feature points described in the environment information.
All shape feature points (shown in FIG. 8) of all objects of the environmental knowledge information described hierarchically are evaluated by the first evaluation function for narrowing down, and feature point candidates are selected in the order of the evaluation result. Create a list. As the first evaluation function, the distance from the robot is used as the evaluation function related to the size of the recognition error because of the general property that the recognition accuracy is poorer as the point is farther away and the possibility of erroneous recognition with other shapes is high. Adopted. Since it is sufficient to prepare the number of feature point candidates required for the next optimum feature shape point selection processing, the number is set in advance and unnecessary portions are discarded. In principle, a list of the order of all shape feature points may be obtained without discarding.

Specifically, S1 in the flow chart of FIG.
As shown in 040 to S1048, shape feature points are selected from the node data (shown in FIG. 8) of the target object described in the environment information, and whether the shape feature points are within the field of view of the camera at that time is determined. Find out. Next, the distance from the robot to the shape feature point is calculated. Since the visual processing is performed by a single eye, as shown in FIG. 18, the distance from the robot to the shape feature point depends on the distance from the camera to the shape feature point that crosses the height of the shape feature point described in the environment information. Find the distance.

Since only the distance to the shape feature point can be measured by this method, it is necessary to obtain the distance to at least two shape feature points in order to obtain the current position of the robot. In this embodiment, the position is determined from the minimum required distance of two points. On the other hand, in this method, if the angle between the two points is small, the measurement error in the direction becomes large.
Therefore, in the flow chart of FIG.
In the process of selecting the optimum shape feature point of 6, the measurement error is prevented from increasing.

That is, in the optimum shape feature point selection processing shown in the subroutine flow chart of FIG. 19, the angle sandwiched between the upper two points of the shape feature point candidates is smaller than the angle (allowable angle) determined from the allowable amount of error. If it is big, just 2
The point is determined as the optimum shape feature point, and the current position of the robot is calculated from the two points (S1060 to S1062).
If it is smaller than the allowable angle, it is determined whether the angle sandwiched by two points including other shape feature point candidates in the field of view can be increased (S1063). At this time, if all the combinations are examined, the processing time rapidly increases when the number of candidates increases. Therefore, in this embodiment, one of the upper two points is exchanged and the combination is changed. Among them, the set of two points having the maximum included angle is determined as the optimum shape feature point. For example, as shown in FIG. 20, when the included angle α between the candidates 1 and 2 is smaller than the allowable angle, one of the candidates 1 and 2, in this case 1 is replaced with 4, and a new included angle β is obtained to determine the allowable angle. I tried to compare with.

Further, as the evaluation function, when there is a shape that is easily erroneously recognized in the vicinity, a function relating to the recognition probability of the visual recognition algorithm is used so as to avoid the use of the feature points of such shape. Is also good. In addition, when moving in a complicated environment, it is possible to recognize the current position using multiple environment recognition means such as multiple visual recognition and sonar. The shape feature point may be determined by setting an evaluation function in consideration of the ease of recognition and recognition accuracy.

Returning to the flow chart of FIG. 14, S10
After recognizing the current position through the position measurement of the feature point selected in 7, the process proceeds to S108 and the walking planning process is performed. This means the work of expanding to the basic walking function so as to become the designated movement target as shown in FIG. Specifically, description will be made according to the flow chart of FIG. 22 showing the subroutine. In S1080, the turning angle and the like are determined, and in S1081, the number of steps and the remainder are determined. More specifically, as noted in the lower part of FIG. 22, first α degrees are changed (mode P1), and n steps forward (mode P).
2) Move forward through the remainder l'm with stride adjustment (mode P
3) Finally, plan walking by β-degree conversion (mode P4) (note that (a) added after each mode indicates the corresponding basic walking function shown in FIG. 9).

In the flow chart of FIG. 14, subsequently, the process proceeds to S110, and the walking control process is performed. This is S1
As shown in FIG. 12, it means the sequential execution of the basic walking mode according to the walking plan. Specifically, the subroutine of FIG.
As indicated by S1120 to S1123 in the flow chart, it means that the control value is determined so as to sequentially execute the modes P1 to P4. However, it should be noted here that the control at this stage is determined by the angle command values for the twelve joints shown in FIG. 1 so as to realize the corresponding basic walking function. Specifically, each motor is driven via a servo amplifier so as to eliminate the deviation between the angle command value and the actual value obtained through the encoder. However, the details thereof are not the object of the present invention, and therefore further description will be omitted.

In the flow chart of FIG. 14, the movement error recognition processing is performed in S114, that is, the estimated value of the current position after movement based on the command value to the movement target (down the stairs) is the same as in S100. A process of recognizing the current position and recognizing a movement error from the command value is performed, and S116 to
When it is determined in S120 that the error exceeds the predetermined value,
The correction movement is repeated until the error becomes equal to or less than the predetermined value.
FIG. 24 shows a state in which a movement error is obtained from the current position estimated from the movement command and the current position obtained through the visual sensor. Note that FIG. 25 is an explanatory diagram of the confirmation work of the shape feature points by image processing.

Returning to the flow chart of FIG.
Go to 22-S28 to climb stairs, etc.
This is to recognize the current position and perform the walking (movement) control, and to recognize the movement error and perform the correction movement in the same manner as described in the movement to the target point under the stairs in S20, so the details are omitted. To do.

In this embodiment, as described above, when the walking plan is realized, the walking mode is classified into five basic walking functions with "stop" in between and the combination is described.
The drive control value of each joint is determined so as to achieve the basic walking function described in the walking control process. Therefore,
The constraint condition expression can be easily calculated, the route can be easily described even when moving in a complicated environment, an obstacle can be reliably avoided, and stable walking control can be realized.

The basic walking function described in FIG. 21 can be expanded in various ways. For example, FIG.
In the case of 6, for example, the left and right strides may be adjusted before proceeding straight, or as in the case of FIG. 21, the direction may be changed to proceed straight and then change direction again.

Further, although the present invention has been described with respect to a bipedal legged mobile robot, the present invention is not limited to this and is also applicable to a legged mobile robot having one leg or three or more legs.

[0032]

According to the first aspect of the present invention, in a walking control device for a legged mobile robot having a plurality of leg links, the walking mode is classified into a plurality of walking patterns, and the plurality of walking patterns are classified into a plurality of walking patterns. Since the target value is described by a combination of walking patterns and the drive control value of the leg link is determined so as to be the target value, it becomes easy to describe the route even when moving in a complicated environment, and to avoid obstacles. It is possible to surely avoid it and to realize stable walking control.

In the apparatus according to claim 2, the combination of the plurality of walking patterns is at least stop → straight ahead → stop, stop → climb → stop, stop → fall → stop, stop → direction change → stop, Since it is configured as one of stop → step adjustment → stop, it becomes easy to calculate the constraint condition expression and it becomes easier to describe the route even when moving in a complicated environment, and obstacles are surely avoided. It is possible to realize more stable walking control.

[Brief description of drawings]

FIG. 1 is an overall schematic view of a walking control device for a legged mobile robot for a mobile body according to the present invention.

FIG. 2 is an explanatory block diagram of a control unit shown in FIG.

FIG. 3 is a functional block diagram showing an operation of the control unit shown in FIG.

4 is a flow chart showing the operation of the control unit shown in FIG.
It is a chart.

FIG. 5 is an explanatory diagram of a planned moving environment in the flow chart of FIG.

FIG. 6 is an explanatory diagram specifically showing a description example of the environment map of FIG.

FIG. 7 is an explanatory diagram showing the third level of the description example of FIG. 6 in more detail.

8 is an explanatory diagram showing the third level of the description example of FIG. 6 in more detail, and particularly an explanatory diagram showing details of feature points in detail.

FIG. 9 is an explanatory diagram showing a basic walking function planned in the flow chart of FIG. 4;

FIG. 10 is an explanatory diagram showing generation of essential passing points in the flow chart of FIG. 4;

FIG. 11 is an explanatory diagram showing the generation of intermediate target points in the flow chart of FIG. 4;

FIG. 12 is a similar explanatory diagram showing the generation of intermediate target points in the flow chart of FIG. 4;

13 is an explanatory diagram showing a route set in the generation of an intermediate target point in the flow chart of FIG.

FIG. 14 is a flow chart showing a subroutine for moving to a target point under the stairs in the flow chart of FIG. 4;

FIG. 15 is a flow chart showing a subroutine of preparation for selection processing in the recognition of the current position in the flow chart of FIG.

FIG. 16 is an explanatory diagram showing the operation of the flow chart of FIG. 15;

17 is a flow chart showing a subroutine for narrowing down the shape feature points in the flow chart of FIG.

FIG. 18 is an explanatory diagram showing the operation of the flow chart of FIG. 17;

FIG. 19 is a flow chart showing a subroutine for selecting an optimum shape feature point in the flow chart of FIG. 14;

FIG. 20 is an explanatory diagram showing the operation of the flow chart of FIG. 19;

FIG. 21 is an explanatory diagram showing designation of a moving target and expansion to a basic walking function.

FIG. 22 is a flow chart showing a subroutine of a walking planning process in the flow chart of FIG.

23 is a flow chart showing a subroutine of sequential execution of a basic walking mode according to a walking plan in the flow chart of FIG.

FIG. 24 is an explanatory diagram showing an error between the current position estimated from the movement command and the recognition result of the current position of the robot.

FIG. 25 is an explanatory diagram showing recognition of shape feature points by image processing.

FIG. 26 is an explanatory diagram showing another variation of the development to the same basic walking function as in FIG. 21.

[Explanation of symbols]

1 leg type mobile robot (bipedal walking robot) 24 base 26 control unit 32 visual sensor 34 image processing unit 58 keyboard, mouse 60 user interface

Claims (2)

[Claims] 1. A walking control device for a legged mobile robot comprising a plurality of leg links, classifies a walking mode into a plurality of walking modes, and describes a target value by combining the plurality of walking modes. Then, the walking control device of the legged mobile robot, wherein the drive control value of the leg link is determined as much as the target value. 2. The combination of the plurality of walking patterns is at least stop → straight ahead → stop, stop → climb → stop, stop → fall → stop, stop → direction change → stop, stop → step adjustment → stop. It is or, It is characterized by the above-mentioned.
A walking control device for a legged mobile robot according to item.