JP2006051585A - Robot device and its walking control method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a robot device and its walking control method capable of enhancing the walking performance. <P>SOLUTION: The legged locomotion robot device and its walking control method are configured so that the toe installed at the forefront of each sole body part is rotated in the hop-up direction during the movable legs being in idling while the heel part installed at the tail of the sole body part is rotated in the down-direction, and when the movable legs are grounded, the toes are rotated in the down-direction as the second step. Also the robot device is furnished with a plurality of toe parts installed at the forefront of the sole body part in such a way as rotatable round a joint axis stretching in the direction different from the pitch axis direction of the machine body of the robot and heel parts installed at the tail of the sole body part rotatably. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a robot apparatus and its walking control method, and is suitable for application to, for example, a biped walking robot apparatus.

  In recent years, research on biped robots has been actively conducted in various research institutions and companies. Biped walking is unstable and difficult to control posture and walking compared to the cloak type, 4 or 6 feet, etc., but walking surfaces and stairs with irregularities on the work path such as rough terrain and obstacles Or it is excellent at the point which can respond | correspond to a discontinuous walk surface, such as raising / lowering a ladder, and can implement | achieve flexible movement work.

  Many techniques relating to posture control and stable walking related to biped robots have already been proposed. The stable “walking” as used herein is defined as “moving using legs without falling down”. The overturn of the airframe means that the work being performed by the robot is interrupted, and considerable labor and time are spent to get up from the overturned state and resume the work. In addition, there is a danger that the fall may cause fatal damage to the robot itself or the object on the other side that collides with the robot. For this reason, posture stability control for avoiding falls is positioned as one of the most important issues in the development of a biped robot.

However, in a robot that performs biped walking, the normal upright posture as the basic posture is unstable in the first place. In many cases, ZMP (Zero Moment Point) is used as a norm for determining the stability of walking for posture stability control of a biped robot. The standard for discriminating the stability by ZMP is the principle of D'Alembert that gravity and inertia force from the walking system to the road surface, and these moments balance with the floor reaction force and the floor reaction force moment as a reaction from the road surface to the walking system. based on. As a result of dynamic reasoning, there is a point where the pitch axis and roll axis moments are zero, that is, ZMP, on or inside the side of the support polygon (that is, the ZMP stable region) formed by the sole contact point and the road surface ( For example, refer nonpatent literature 1).
"LEGGED LOCOMOTION ROBOTS" by Miomir Vukobratovic (Ichirou Kato's "Walking Robot and Artificial Feet" (Nikkan Kogyo Shimbun))

  By the way, in the case of a biped robot, in general, the single leg is supported by the left leg when the right leg is lifted, the both legs are supported when the right leg is grounded, and the right leg is lifted when the left leg is lifted. It is performed by repeating the walking cycle divided into four movement periods of the single leg support period and the both leg support period in which the left leg is grounded.

In this case, the conventional robot has the following problems because it has a single-plate structure with a flat sole.
1. Leg posture is bent to avoid singular posture. The feeling of bending the knee is particularly strong. For this reason, the movable angle of the ankle is also limited.
2. Since the leg postures that can be selected are limited, there is a limit to the ground adaptability.
3. Because the tip (foot) speed is limited, the moving speed is naturally limited.

  As a technique for solving such a problem, in recent years, as shown in FIG. 50, it has been proposed to attach the toe 201 to the foot main body 200 with a degree of freedom to jump up to the foot (for example, Http://www.jsk.t.u-tokyo.ac.jp/research/h6/H6_H7.html (as of September 11, 2003)).

According to this technique, the presence of the toe 201 causes the following 1. Since the substantial leg length is extended by the amount of the toe 201, it is possible to walk at high speed. 2. Because you can lift your foot higher by the tip of the toe 201, you can walk on higher steps. Since the walking posture is beautiful, it is possible to obtain the effect that the knee bends less.

  Also, by providing a movable foot piece on the foot of a biped robot, walking stability can be increased both when walking on a flat walking surface and when climbing or descending stairs. (For example, JP-A-5-318335). In this case, a movable foot piece that can move in a direction that changes the area of the effective ground contact surface with respect to the walking surface is provided in the foot main body portion, and each movable foot piece is moved in a predetermined direction by the operating mechanism.

  By providing a movable part such as a toe or a foot piece on the foot, while providing a great merit, the mechanism becomes complicated, causing a problem that the weight is increased and the movable angle is partially reduced. In addition, since it is sufficient to determine the sole position and posture for normal walking, there is still a viewpoint that the toe has an extra degree of freedom.

  For these reasons, while the effects and possibilities of introducing a toe are recognized, the introduction is only at the research level and has not yet been put into practical use. Of course, it is certain that the control algorithm considering the tip of the toe is much higher level.

  On the other hand, conventionally, as a heel structure of a legged walking type robot, the foot heel 210 is made of a flexible material or the like as shown in FIG. 51 for the purpose of optimally balancing shock absorption / relaxation and posture stabilization after landing. It has been proposed to configure using (Japanese Patent Laid-Open No. 2001-129774).

  In this heel structure, the heel 210 of the foot is composed of a hard member (protrusion 211) and a soft member (filler 212) filled therebetween, and the protrusion 211 has a comb-like shape in a cross-sectional view and a plan view. In FIG. 4, the linear shape or the non-linear shape is continuous, and the rigidity against the force acting from the other direction is lower than the rigidity against the force acting from the gravity axis direction.

  However, according to such a heel structure, it can be expected that some effects will be obtained with respect to shock absorption / relaxation when landing on the legs, but the sole shape is always fixed when viewed as the whole sole. However, it is difficult to obtain sufficient contact pressure and support moment on uneven terrain, and there is still a problem that adaptability to rough terrain is still insufficient.

  The present invention has been made in consideration of the above points, and intends to propose a robot apparatus capable of improving walking performance and a walking control method thereof.

  In order to solve such a problem, in the present invention, in a legged mobile robot device having a movable leg, a toe portion rotatably attached to a front end portion of a foot main body portion, and a rear end portion of the foot main body portion A hook part rotatably attached to the head, a driving means for rotationally driving the toe part and the hook part in the jumping-up direction or the lowering direction, and an external jumping direction or lowering direction applied to the toe part or the hook part. Shock absorbing means is provided, and the drive means rotates the toe portion in the jumping-up direction and the buttocks in the downward direction during the free leg period of the movable leg. The part was rotated in the downward direction.

  As a result, in this robot apparatus, the impact at the time of landing of the movable leg can be absorbed and alleviated by the toe portion and the heel portion. Further, at this time, the toe portion and the heel portion are rotationally driven in the jumping direction or the lowering direction, so that the toe portion and the heel portion can be grounded following the unevenness of the grounding surface, and sufficient grounding pressure can be obtained even in the uneven landform And support moment can be obtained.

  Further, in the present invention, in the walking control method for the legged mobile robot apparatus having the movable leg, the toe portion attached to the front end portion of the foot body portion is rotated in the jumping-up direction during the free leg period of the movable leg. In addition, a first step of rotating the heel part attached to the rear end of the foot main body part in the downward direction and a second step of rotating the toe part in the downward direction when the movable leg lands are provided. I did it.

  As a result, according to the walking control method of the robot apparatus, the impact at the time of landing of the movable leg can be absorbed and reduced at the toe portion and the heel portion. Further, at this time, the toe portion and the heel portion are rotationally driven in the jumping direction or the lowering direction, so that the toe portion and the heel portion can be grounded following the unevenness of the grounding surface, and sufficient grounding pressure can be obtained even on uneven terrain. And support moment can be obtained.

  Further, in the present invention, in the legged mobile robot device, a plurality of toe portions rotatably attached to the front end portion of the foot main body portion by joint axes in directions different from the pitch axis direction of the robot body, A heel part rotatably attached to the rear end part of the foot main body part is provided.

  As a result, in this robot apparatus, the toe portion and the heel portion can be grounded following the unevenness of the ground contact surface, and sufficient contact pressure and support moment can be obtained even on uneven landforms.

  According to the present invention, in a legged mobile robot device having a movable leg, a toe portion that is rotatably attached to the front end portion of the foot main body portion and a rear end portion that is rotatably attached to the foot main body portion. And a driving means for rotationally driving the toe portion and the heel portion in the jumping-up direction or the lowering direction, and a buffer for relaxing the impact in the jumping-up direction or the lowering direction applied to the toe portion or the heel portion from the outside. And the drive means rotates the toe part in the jumping-up direction and the buttocks part in the lowering direction during the free leg period of the movable leg, and moves the toe part in the lowering direction when the movable leg lands. By rotating it, the impact when landing on the movable leg can be absorbed and mitigated by the buffering means via the toe part or the heel part, and the toe part and the heel part can be made to follow the unevenness of the ground contact surface even in uneven terrain. Enough ground pressure It is possible to obtain the support moment, thus possible to realize a robot apparatus capable of improving walking ability.

  According to the invention, in the walking control method for the legged mobile robot device having the movable leg, the toe portion attached to the front end portion of the foot main body portion is rotated in the jump-up direction during the free leg period of the movable leg. And a second step of rotating the toe portion in the downward direction when the movable leg is landed, and a first step of rotating the heel portion attached to the rear end portion of the foot main body portion in the downward direction. By providing it, the impact at the time of landing of the movable leg can be absorbed and mitigated by the buffering means via the toe part or the heel part, and the toe part and the heel part are made to follow the unevenness of the ground contact surface even in the uneven terrain. It is possible to realize a walking control method for a robotic device that can be grounded to obtain sufficient ground pressure and support moment, and thus improve walking performance.

  Furthermore, according to the present invention, in the legged mobile robot device, a plurality of toe portions rotatably attached to the front end portion of the foot main body portion by joint axes in directions different from the pitch axis direction of the robot body. Since the toe part and the heel part can be grounded following the unevenness of the ground surface by providing a heel part rotatably attached to the rear end part of the foot body part, the uneven terrain A sufficient ground pressure and supporting moment can be obtained even in this way, and thus a robot apparatus capable of improving walking performance can be realized.

  Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.

(1) Configuration of Robot According to this Embodiment In FIGS. 1 and 2, reference numeral 1 denotes a robot according to this embodiment as a whole, and a head unit 4 is connected to an upper portion of a body unit 2 via a neck 3. The arm unit 5A, 5B is connected to the upper left and right side surfaces of the body unit 2 and a pair of leg units 6A, 6B are connected to the lower part of the body unit 2. ing.

  In this case, the neck 3 is held by a neck joint mechanism 13 having degrees of freedom around the neck joint pitch axis 10, the neck joint yaw axis 11, and the neck joint pitch axis 12, as shown in FIG. The head unit 4 is attached to the tip of the neck 3 with a degree of freedom around the neck roll shaft 14 as shown in FIG. As a result, in the robot 1, the head unit 4 can be directed in the desired directions of front and rear, left and right, and diagonally.

  Each arm unit 5A is composed of three blocks, an upper arm block 15, a forearm block 16, and a hand block 17, as clearly shown in FIGS. 1 and 2, and the upper end of the upper arm block 15 is shown in FIG. 3 is connected to the body unit 2 via a shoulder joint mechanism 20 having degrees of freedom around the shoulder pitch axis 18 and the shoulder roll axis 19.

  At this time, the forearm block 16 is connected to the upper arm block 15 with a degree of freedom around the upper arm yaw axis 21 as shown in FIG. Further, the hand block 17 is connected to the forearm block 16 with a degree of freedom around the wrist yaw axis 22 as shown in FIG. Further, the forearm block 16 is provided with an elbow joint mechanism 24 having a degree of freedom around the elbow pitch axis 23.

  As a result, the robot unit 1 can move the arm units 5A and 5B as a whole with almost the same degree of freedom as a human arm unit, thus greeting with one hand raised or dancing to swing the arm units 5A and 5B. Various actions using the arm units 5A and 5B can be performed.

  Further, five finger portions 25 are attached to the distal end portion of the hand block 17 so as to be able to bend and extend, respectively, so that an object can be picked or grasped using these finger portions. ing.

  On the other hand, each leg unit 6A, 6B is composed of three blocks, a thigh block 30, a shin block 31, and a foot block 32, as is apparent in FIG. 1 and FIG. As shown in FIG. 3, the upper end portion of 30 is connected to the body unit 2 via a hip joint mechanism portion 36 having degrees of freedom around the hip joint yaw axis 33, the hip joint roll shaft 34, and the hip joint pitch axis 35.

  At this time, the thigh block 30 and the shin block 31 are coupled via a knee joint mechanism 38 having a degree of freedom around the shin pitch axis 37 as shown in FIG. Are connected via an ankle joint mechanism 41 having degrees of freedom around an ankle pitch axis 39 and an ankle roll axis 40 as shown in FIG.

  As a result, the robot 1 can move these leg units 6A and 6B with almost the same degree of freedom as a human leg, and thus various operations using the leg units 6A and 6B such as walking and kicking a ball. Can be expressed.

  In the case of the robot 1, each hip joint mechanism 36 is supported by a waist joint mechanism 44 having degrees of freedom around the trunk roll axis 42 and the trunk pitch axis 43 as shown in FIG. The body unit 2 can be freely tilted in the front-rear and left-right directions.

Here, in the robot 1, as described above, as a power source for moving the head unit 4, the arm units 5A and 5B, the leg units 6A and 6B, and the body unit 2 as shown in FIG. In addition, actuators A _{1 to} A _{17 corresponding} to the number of degrees of freedom are arranged in portions having respective degrees of freedom including the joint mechanisms such as the neck joint mechanism 13 and the shoulder joint mechanism 20. The actuators A _{1 to} A ₁₇ are provided with an arithmetic circuit formed as an IC chip inside the case, a current detector for detecting a drive current value, and the like, and also have a communication function with an external device. Yes (see, for example, Japanese Patent Application No. 2000-38097).

  The body unit 2 stores a main control unit 50 that controls the operation of the entire robot 1, a peripheral circuit 51 such as a power supply circuit and a communication circuit, a battery 52 (FIG. 5), and the like. In the constituent units (the body unit 2, the head unit 4, the arm units 5A and 5B, and the leg units 6A and 6B), the sub control units 53A to 53A electrically connected to the main control unit 50, respectively. 53D is stored.

  Further, as shown in FIG. 5, the head unit 4 includes various external devices such as a pair of CCD (Charge Coupled Device) cameras 60A and 60B that function as “eyes” of the robot 1 and a microphone 61 that functions as “ears”. A sensor, a speaker 62 functioning as a “mouth”, and the like are disposed at predetermined positions. A touch sensor 63 as an external sensor is disposed at each predetermined portion such as the back surface of the foot block 32 in each leg unit 6A, 6B.

Further, various internal sensors such as a battery sensor 64 and an acceleration sensor 65 are disposed in the body unit 2, and each component unit is associated with each actuator A _{1 to} A _17. Potentiometers P _{1 to} P ₁₇ are provided as internal sensors that detect the rotation angles of the output shafts of the actuators A _{1 to} A ₁₇ .

  Each of the CCD cameras 60A and 60B images the surrounding situation, and sends the obtained image signal S1A to the main control unit 50 via the sub control unit 53B (not shown in FIG. 5), while the microphone 61 Then, various external sounds are collected, and the audio signal S1B thus obtained is sent to the main control unit 50 via the sub-control unit 53B.

  Each touch sensor 63 detects a physical action from the user or a physical contact with the outside, and the sub-control units 53A to 53D (not shown in FIG. 5) corresponding to the detection result as the pressure detection signal S1C. To the main control unit 50.

  Further, the battery sensor 64 detects the remaining energy of the battery 52 at a predetermined cycle, and sends the detection result to the main control unit 50 as a remaining battery signal S2A, while the acceleration sensor 65 has three axes (x axis, y axis). And the z-axis) are detected at a predetermined cycle, and the detection result is sent to the main control unit 50 as an acceleration detection signal S2B.

Further, each of the potentiometers P _{1 to} P ₁₇ detects the rotation angle of the output shaft of the corresponding actuator A _{1 to} A ₁₇ , and the corresponding sub-control units 53 A to 53 A as the angle detection signals S ₂ C _{1 to} S ₂ C ₁₇ at predetermined intervals. The data is sent to the main control unit 50 via 53D. Each of the actuators A _{1 to} A ₁₇ calculates its own output torque based on the drive current value detected by the above-described current detector provided therein, and the calculation results are output torque detection signals S2D _{1 to} S2D. ₁₇ is sent to the main control unit 50 via the corresponding sub-control units 53A to 53D.

The main control unit 50 has a microcomputer configuration including a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), and the like, and includes CCD cameras 60A and 60B, a microphone 61, and touch sensors 63. External sensor signals S1 such as image signal S1A, audio signal S1B, and pressure detection signal S1C supplied from various external sensors, and various internal sensors such as battery sensor 64, acceleration sensor 65, and potentiometers P _{1 to} P ₁₇ Energy remaining signal S2A, acceleration detection signal S2B and angle detection signals S2C _{1 to} S2C ₁₇ respectively supplied from the actuators, output torque detection signals S2D _{1 to} S2D ₁₇ respectively supplied from the actuators A _{1 to} A _17, etc. Based on the internal sensor signal S2, the robot 1 External and and internal conditions to determine whether the physical urging of the user or the like.

  The main control unit 50 then continues the robot 1 based on the determination result, a control program stored in the internal memory 50A in advance, various control parameters stored in the external memory 66 loaded at that time, and the like. The control command based on the determination result is sent to the corresponding sub-control units 53A to 53D (FIG. 4).

As a result, based on this control command, the corresponding actuators A _{1 to} A ₁₇ are driven under the control of the sub-control units 53A to 53D, thus swinging the head unit 4 up and down, left and right, Various actions and actions such as raising the unit units 5A and 5B and walking are expressed by the robot 1.

  In this way, the robot 1 can act autonomously based on external and internal situations.

(2) Configuration of the foot block 32 of the robot 1 according to the present embodiment (2-1) Basic structure (2-1-1) Toe placement and movable range with two or more degrees of freedom Most moving surfaces are uneven Even if a flexible material is provided on the sole, it is difficult to set the contact load (pressure distribution) appropriately.

  However, appropriately controlling the ground contact state greatly affects the movement performance of the robot. This influence becomes a problem to be sufficiently examined and solved as the size of the legged mobile machine becomes smaller, that is, as the robot body becomes smaller, the size effect of the moving surface on the motion performance of the robot body becomes larger.

  When ZMP is used as a stability criterion, it is also important to select a support polygon appropriately in order to generate a support moment appropriately. In principle, it is necessary to increase the area of the support polygon, but the shape of the support polygon greatly affects walking performance. Note that it is not simply determined by the ground contact area.

  In this case, it is not sufficient to equip a toe with one degree of freedom per leg as in the prior art described above, and it is desirable to prepare a mechanical system having toes with more than two degrees of freedom.

  FIG. 6 shows, as an example, a foot structure in which two finger portions 70A and 70B are provided on the toe. In this example, two foot joints 73A and 73B having joint axes parallel to the pitch axis direction (direction perpendicular to the traveling direction) are provided at the front end edge of the foot main body 72 connected to the ankle joint 71, respectively. Finger parts 70A and 70B are rotatably attached.

  In addition to providing two finger portions 70A and 70B, this foot structure is also characterized by the movable range of these finger portions 70A and 70B. That is, these two finger portions 70A and 70B jump upward with respect to the foot main body 72 as shown in FIG. 7A (the direction indicated by the arrow a, hereinafter referred to as the jumping direction). In addition, it can also operate in a direction (a direction indicated by an arrow b, hereinafter referred to as a lowering direction) that is lowered with respect to the foot main body 72 as shown in FIG. 7B.

  As a movable range of these two finger portions 70A and 70B, about 60 [°] in the jump-up direction and about 15 [°] in the lowering direction can be considered. However, it is necessary to change this movable range depending on the scale of the robot and the leg control method. For example, the foot length is about 110 mm, the foot width is about 70 mm, and the height is 600 mm. In the case of a robot having a degree of about 35 [°] in the jumping direction and about 15 [°] in the lowering direction, it is sufficient.

  Such movement of the fingers 70A and 70B in the jumping direction is effective for kicking the moving surface and obtaining a propulsive force when switching from the support leg to the free leg during walking or running. The movement of the fingers 70A and 70B in the downward direction is effective when adjusting the ground pressure or making the sole follow the moving surface as shown in FIG. 34, and as shown in FIG. You can also avoid stepping on it. Further, by allowing each of the finger portions 70A and 70B to be moved independently, as shown in FIG. 36, it is possible to make the sole follow the moving surface with higher accuracy.

(2-1-2) Arrangement of movable heel and movable range As shown in FIG. 6, the foot structure according to the present embodiment has another feature that a degree of freedom is also provided in the heel portion of the foot. To do. In the example of FIG. 6, a heel part 75 is attached to the rear end edge of the foot main body part 72 via a heel joint 74 having a joint axis parallel to the pitch axis.

  Then, like the finger portions 70A and 70B, the heel portion 75 is not only in the jumping direction (arrow c) as shown in FIG. 7A but also in the foot main body portion 72 as shown in FIG. 7B. Further, it can be operated in the downward direction (arrow d) with respect to the foot main body 72.

  In addition to buffering at the time of landing, the collar portion 75 has a function of assisting improvement in ground adaptability together with the two finger portions 70A and 70B of the toe. The foot according to the present embodiment is considered to adjust the state of the grounding portion of the foot body, although it constitutes a supporting grounding surface in the same manner as the two finger portions 70A and 70B of the toe. It can be said that the function of the collar portion 75 in the flat structure is specifically described.

  As the movable range of the flange portion 75, about 30 [°] in the jumping direction and about 15 [°] in the lowering direction can be considered. However, this movable range also varies depending on the scale of the robot and the leg control method.

(2-1-3) Direction of joint axis of each finger part 121A, 121B FIG. 6 illustrates an example of a foot structure in which each joint axis of each finger part 70A, 70B is aligned with the pitch axis direction. However, by providing these joint axes with an angle with respect to the pitch axis direction, the ground angles can be made different between the finger portions 70A and 70B and the foot main body portion 81.

  8 and 9 in which the same reference numerals are assigned to the corresponding parts as in FIG. 6, the configuration example when the mounting direction of the joint shaft of each finger 80A, 80B, 85A, 85B is different from the pitch axis direction Is shown. As shown in FIGS. 8 and 9, it is possible to appropriately change the sole of the foot by appropriately tilting the joint axis.

8, each finger portion 80A, the 80B, the joint axis at the foot structure example in which the front edge of the inclined angle theta ₁ to the lower side of the roll direction foot body 81 with respect to the pitch axis In this way, since the contact pressure at the sole of the foot can be secured high, an appropriate road surface friction can be secured, and an appropriate support moment can be generated.

Further, FIG. 9 shows the foot main body 85A, 85B by tilting the joint axis with respect to the pitch axis direction by an angle θ ₂ downward in the roll direction and an angle θ ₃ backward in the yaw direction. This is an example of a foot structure attached to the front edge of 86, and by doing this, it is possible to secure a high ground pressure on the inside and outside of the foot by rotating the finger portions 85A and 85B downward in the roll direction. Therefore, an appropriate road surface friction can be ensured, and an appropriate support moment can be generated.

  According to such a foot structure, the angles of the finger portions 80A, 80B, 85A, 85B with respect to the foot body portions 81, 86 can be changed according to the load level applied to the sole. it can.

  In addition, if the center of the sole is convex, it is difficult to ensure an appropriate support moment. Therefore, it is better to set the support point on the outer side as much as possible. In order to achieve this object, it is desirable to change the axial direction of the joint axis as described above so that the outer peripheral portion is lowered. Exhibits an effect just like an arch.

  Specifically, the attachment angle of the finger portions 80A, 80B, 85A, 85B with respect to the pitch axis direction is, for example, about −5 to 15 ° on the lower side in the roll direction, and 30 to 45 ° on the backward side in the yaw direction. ] Set to a range of about. Since it is desirable to increase the contact pressure when the sole is moved downward, the mounting angle is basically selected so that the ground angles of the fingers 80A, 80B, 85A, 85B are tight.

(2-1-4) Position in the height direction of the joint axis of each finger 121A, 121B In the case of having a structure in which two fingers 70A, 70B are provided at the front end of the foot main body 72 as shown in FIG. In addition to the joint axis direction (inclination with respect to the pitch axis) of these two finger portions 70A and 70B, the position of the joint axis in the height direction is also important.

  In practice, for example, when the joint shaft 90 is set high from the sole as shown in FIG. 10, the toe 91 can be bent only in the jumping direction, the joint torque of the toe 91 is increased, and the toe 91 is further bent. Since a large gap is opened between the foot main body portion 92 and the foreign body may be caught, problems such as lack of ground adaptability and danger arise.

  In order to avoid such a situation, as shown in FIG. 11, the joint shaft 94 of each finger part 93 is set as low as possible from the sole, and as shown in FIG. 12, each finger part 95A, By devising such as configuring the finger joints 97A and 97B so that the rear end edge of 95B and the front end edge of the foot body portion 96 are engaged, the intended effect can be obtained substantially.

  As shown in FIG. 13, the specific height position of the joint shaft 99 of each finger portion 98 is preferably set so as to be located within the thickness h of the foot. In this case, considering the problem of the gap between the toe 91 and the foot main body 92 as described above with reference to FIG. 10, the joint shaft 99 (FIG. 13) should be lowered to the vicinity of the moving surface. It is difficult. Actually, it seems that the joint axis 99 of each finger part 98 is set to a height position of about 2/3 of the height from the sole to the upper surface of the foot.

(2-1-5) Arrangement of Drive Actuator In general, a legitimate mobile robot should be designed to concentrate heavy objects on the body and to suppress the weight of the tip that increases speed. There is. However, in reality, at least 6 actuators per leg must be arranged (see Fig. 3), and there is a design that distributes to each part, and we are considering an appropriate weight arrangement within that range. Is real.

  As described above, when the mechanism is provided with the two finger portions 70A and 70B on the toe of the foot, if an actuator is further disposed in the foot, the space is reduced. Assume that it is attached to a place where it can be arranged in accordance with the property of the robot to operate, that is, according to the performance required for the actuator for driving each finger 70A, 70B. When the actuator is arranged at a place other than the finger part or the foot main body part, a power transmission means such as a link, a wire, a belt, or a gear is separately required.

  Depending on the moving speed, the operating speed, the body scale, and the function, the following control method can be selected and applied to the finger drive mechanism.

1. When the robot moves relatively slowly and mainly follows scanning control on the moving surface:
The actuator is placed on the finger or foot main body. This is because the actuator is small and a simple design is suitable.
2. If the robot moves relatively fast, especially when it moves its legs in a fast cycle:
In order to reduce the inertial force, it is necessary to make the tip as light as possible. In such a case, the actuator is placed on the leg or the trunk, but in that case, a transmission mechanism is required.
3. When the robot can jump and run:
It is thought that the construction method of the actuator will change. If a large output (instantaneous force) is required as a whole, it may be better to place it at the tip with space.

  As described above, it is desirable to change the arrangement of the actuators according to the size of the airframe and the operating conditions.

  FIG. 14 shows an example of arrangement of actuators on the movable legs. Biped walking robots generally tend to be heavy in order to generate strong output to the foot mechanism in order to avoid the influence of inertial force, etc. However, it is not necessary to place the actuator on the foot mechanism. As shown in FIG. 15, it may be arranged on each part of the shin or thigh or the trunk, and the power may be transmitted via a transmission mechanism in which wires, gears, links and the like are combined.

  In order to realize a powerful output in a specific joint actuator, it is possible to use the principle of interference driving, arrange actuators for the degree of freedom of toes, and drive by combining any of the actuators for joint axes. . For example, it can be said that it is possible to operate two degrees of freedom of the ankles and two degrees of freedom of the toes by adding two motors to interfere with the two-axis motors of the ankles.

(2-1-6) When equipped with passive mechanical elements It can be said that a dynamic mechanical system can obtain high performance if active control using a servo actuator is performed. Performance can be further improved by performing predictive control using abundant sensor groups. However, there is a drawback that it becomes impossible to adapt gradually as the frequency increases.

  Therefore, the present inventor has proposed a design theory of a composite control system that incorporates a passive mechanism control system to compensate for response characteristics in a high frequency band such as disturbance input, that is, a composite control system composed of a combination of active control elements and passive control elements. Et al. Propose.

  On the other hand, the passive mechanism control system is configured by a combination of mechanical elements such as springs and dampers. A variable nonlinear compliance mechanism is introduced as a passive control element. It should be noted that a simple spring is not introduced. Examples of the elements introduced here include the following.

1. Spring (including non-linear characteristics, preload, limiter, etc.)
2. Damper (non-linear characteristics, characteristics changed by compression / extension or negative characteristics for speed, etc.)
3. Friction (nonlinear component with low speed dependence)

  FIG. 15 shows an example of a linear motion mechanism 100 that realizes the concept of a mixed control system including an active servo control system and a passive mechanism control system.

  In the example shown in the figure, a mechanism system 102 capable of responding in a passive and broadband manner is connected in series between an active control servo drive system 101 including a servo motor and an output unit (effector such as a joint). Has been placed. The passive mechanism system 102 can be configured by non-linear elements such as preload, friction, and the like in addition to compliance such as a spring and a damper. Further, a torque limiter, another link, a joint, or the like may be further included between the passive mechanism system 102 and the output unit. The mechanism element may be an electric system such as a coil.

  Here, in order to realize a passive compliance mechanism, it is difficult to achieve a sufficient performance with a simple spring, and it is preferable to include a nonlinear mechanism element. With respect to this part, the constant or the like can be changed by an actuator or the like. The passive control processing system 103 is basically realized by machine elements, but it is also assumed that control means for assisting it is combined.

  In the example shown in FIG. 15, a switching mechanism 104 such as a solenoid for turning on / off the operation of the passive mechanism system 102 is arranged, and the passive mechanism system 102 is locked, released, and controlled. That is, the high-frequency response action in the passive mechanism system 102 can be energized or deenergized by the switching mechanism control signal S10 from the control processing system 103.

  Of course, it is possible to achieve the desired operation and effect without combining the passive mechanism system 102 and the control processing system 103 as described above, but it is possible to perform more sophisticated and precise control. .

  Below, each element which comprises the passive mechanism system 102, such as a spring and a damper, is demonstrated.

(2-1-6-1) Spring FIG. 16 shows the elastic characteristics of a normal linear spring. Applying such a simple spring is inappropriate for ensuring stability. This is because even if an attempt is made to generate the necessary support moment, if the displacement is small, it is not sufficiently transmitted and may cause a fall.

  In this case, when the load (displacement) of the spring is small, if the spring is too soft, the displacement becomes too large and it is difficult to generate a sufficient support moment. On the other hand, if the spring constant is increased, the characteristic cannot be adapted to the buffer. That is, it is difficult to directly use it for a legged mobile robot, and there may be problems such as an unnecessarily large stroke or vibration. Also, the spring element is generally difficult to control to determine the position.

  As a means for solving such a problem, there is a preload mechanism that does not operate in a low load region and elastic characteristics are activated when a predetermined load is exceeded, and elastic deformation stops when a displacement amount reaches a predetermined value. Apply a limiter mechanism to the spring.

  The elastic characteristics in this case are as shown in FIG. 17, and an offset (bias) is given to moderately suppress the spring constant and improve the characteristics in the low load region. By obtaining an elastic characteristic as shown in FIG. 17, for example, it can act as a substantially rigid body at normal times and operate only in an emergency, or can be displaced with a load at the time of foot contact.

  However, in practice, linear characteristics are not necessary so far. Further, instead of the mechanical positional relationship, a similar elastic characteristic may be realized by controlling a switching mechanism, and in this case, the elastic characteristic is as shown in FIG. In addition, it is possible to improve the consistency between the movement performance and the control characteristics by intentionally imparting non-linearity to the elastic characteristics.

  Combining functions with preload, displacement stopper, non-linear characteristic spring, etc. is effective. In addition, since the connection of these elements with an extreme step causes disturbance in behavior, it is necessary to consider characteristics that change smoothly.

(2-1-6-2) Damper FIG. 19 shows the characteristics of a damper serving as a basic model. The damper shown in FIG. 19 is a damping element having linear viscosity. If such a simple characteristic is applied, the resistance force becomes too large against a fast disturbance input. That is, the resistance force to a high-speed response becomes large, and when considered with a walking machine, it acts as a rigid body against a sudden force. For this reason, after all, it causes a fall moment. The following two methods are introduced as countermeasures.

1. 1. Characteristics change in the direction of expansion and contraction. Negative characteristics with high tracking capability for high-speed input.

  FIG. 20 shows the characteristics of a damper that introduces the first method and gives different characteristics depending on directions. In other words, it is effective to change the damping rate according to the direction in which the load is applied. When receiving an impact, the damping force is reduced to absorb quickly, and when extending, the damping force is set to be large, and the falling moment of jumping is reduced. Try to suppress the occurrence.

  FIG. 21 shows a speed-dependent damper characteristic in which the second method is introduced and the characteristic is switched according to the speed at which the system operates. The behavior due to the impact force is extremely fast, and it is desirable to reduce the displacement and impact of the upper body by quickly displacing the resistance force as the speed increases. This function can be realized by giving the damper a negative characteristic as shown in FIG.

  By freely combining the damper characteristics shown in FIGS. 20 and 21, it is possible to appropriately generate a force for releasing and supporting the impact force. This will be further expanded to give non-linear characteristics to prevent abrupt changes by devising such as reducing the influence of the natural frequency of the aircraft. FIG. 22 shows an example of the damper characteristic in this case. Since the connection of these elements with an extreme step causes disturbance in behavior, it is necessary to consider characteristics that change smoothly.

  In addition to the spring and the damper, characteristics can be improved by appropriately introducing factors such as friction.

(2-1-6-3) Switching mechanism In order to switch the characteristics of the passive mechanism control system, a preload, a latch or solenoid, a brake mechanism, or a combination thereof can be introduced.

  When the load or speed increases, the operation starts passively or the switching operation is clarified by giving control. Or it can be used to create the properties themselves.

  FIG. 23 shows a configuration example of a nonlinear compliance mechanism in which the above-described spring, damper, and switching mechanism are organically combined.

  In this non-linear compliance mechanism 110, the elements of the non-linear spring 111 are constituted by an indefinite pitch, multiple springs, or the like. The limit mechanism 112 has a buffer mechanism 113, and it is preferable to use a buffer material instead of stopping it suddenly with a stopper.

  A preload mechanism 114 acts on the nonlinear spring 111. A brake or a latch is used as a locking mechanism. Both passive and active control systems are equipped with a locking mechanism. In addition, as the nonlinear damping mechanism including the nonlinear spring 111 and the damper 115 in FIG. 23, an electric system such as a solid friction or a coil can be used in addition to a fluid.

  FIG. 24 shows a configuration example of a mechanism unit that organically combines a spring, a damper, and a switching mechanism. In the example shown in FIG. 24, the non-linear compliance mechanism 110 shown in FIG. 23 is used in combination with a lock mechanism 116 that appropriately locks passive control elements.

  Due to the mechanism configuration according to the present embodiment, it is necessary to mount a strong actuator on the foot mechanism body, but it is difficult to mount due to other design matters such as weight and dimensions. In this case, a mechanism that can lock (brake) is effective, and it cannot be said that the finger mechanism at the tip of the foot is always required to move. Alternatively, the brake mechanism is effective when the operation is such that it is pushed back to the external force while outputting a certain resistance force (torque).

(2-2) Specific Structure of Foot Block 32 of Robot 1 According to the Present Embodiment Here, FIGS. 25 to 27 show the robot 1 according to the present embodiment formed in consideration of the various items as described above. A specific structure of the foot block 32 is shown.

  As is clear from FIGS. 25 to 27, the foot block 32 has two finger portions 121 </ b> A and 121 </ b> B disposed on the front end edge of the foot main body portion 120 and the rear foot portion 120. The flange 122 is disposed at the end edge.

  In this case, in each finger part 121A, 121B, the rear end edge formed in a comb-teeth shape so that its joint axis is positioned within the thickness of the foot main body part 120 is similarly combed. The foot body 120 is attached to the foot body 120 so as to be fitted with the front end edge of the foot body 120 formed in a tooth shape.

  As a result, in the foot block 32, the fingers 121A and 121B are both raised and lowered without causing a gap between the fingers 121A and 121B and the foot main body 120 during walking or running. It is possible to drive freely, thus, by moving in the jumping direction, you can kick the moving surface when walking or running to obtain sufficient propulsive force, and by adjusting the ground pressure by moving in the downward direction The sole can be made to follow the moving surface.

  In addition, each finger 121A, 121B is formed in a trapezoidal shape so that a gap of an appropriate size is generated between each of the adjacent fingers 121A, 121B. When the 121A and 121B are moved in the jumping-up direction or the lowering direction, the finger portions 121A and 121B can be prevented from interfering with other adjacent finger portions 121A and 121B.

  Furthermore, each finger part 121A, 121B is provided on the foot body part 120 so that its joint axis is inclined about 30-45 [deg.] In the backward direction and about 5-15 [deg.] Downward with respect to the pitch axis direction. Thus, the support point at the time of grounding is positioned outside the bottom surface of the foot so that an appropriate support moment can be obtained.

  On the other hand, in the heel portion 122, the front end edge formed in a comb-teeth shape is similarly formed in a comb-teeth shape so that the joint axis is positioned within the thickness of the foot main body portion 120. The foot main body 120 is attached to the foot main body 120 so as to be fitted to the rear end edge.

  As a result, in the foot block 32, the heel portion 122 can be freely raised and lowered without causing a gap between the heel portion 122 and the foot main body portion 120 during walking or running. Thus, the movement in the jumping direction reduces the impact at the time of landing when walking or running, and the movement in the lowering direction adjusts the ground pressure or makes the sole follow the moving surface. It is made to be able to

  In the case of this embodiment, although not shown, the nonlinear compliance mechanism described above with reference to FIG. 23 is provided between each finger 121A, 121B and the foot main body 120 and between the foot main body 120 and the heel 122. Is provided as shown in FIG. 24, so that response characteristics in a high frequency band such as disturbance input can be improved.

In addition to this configuration, in the case of this embodiment,
1. 1. Each finger 121A, 121B and foot main body 120 has a concave shape at the center of the back surface (surface to be grounded). 2. Each finger 121A, 121B has a shape in which the outer peripheral side is more convex than the inner peripheral side so as to be grounded on the outer peripheral side of the back surface 3. A shape in which a convex portion is also provided on the rear end side (that is, the base side) of the back surface of each finger portion 121A, 121B to enable the tiptoe to stand. In order to satisfy the requirement for maintaining a stable stance or walking state that the foot main body portion 120 is also convex toward the corner so as to be grounded on the outer peripheral side, the sole of the foot block 32 is shown in FIG. ) To (C) are formed to be curved in a concave shape as indicated by the curves k _{1 to} k ₄ .

In practice, in this robot 1, each finger portion 121A, in 121B, the back surface, as shown respectively by curves _{k 2} in the curve _{k 1} and FIG. 28 (B) in FIG. 28 (A), the width direction Is an arcuate curved surface as a whole so that the center of the foot main body 120 in the width direction is farther from the grounding surface, and the longitudinal direction perpendicular thereto is farther from the grounding surface in the middle of the fingers 121A and 121B. It is formed in the concave shape which has.

Also similarly the rear surface of the foot body portion 120, as indicated by the curve k ₄ in the curve k ₃ and FIG. 28 (C) in FIG. 28 (B), the of the foot main body 120 in the width direction and the longitudinal direction both The center is formed in a curved concave shape that is recessed in an arch shape so as to be farther away from the ground contact surface.

In still heel 122, the back surface, as shown respectively by curves _{k 4} in the curve _{k 3} and FIG. 28 (C) in FIG. 28 (B), the from the ground plane as the center of the heel portion 122 in the width direction In the longitudinal direction, it is formed in a curved concave shape that is recessed in an arch shape as a whole so as to be smoothly connected to the curved shape of the foot main body 120.

  As a result, in the robot 1, such a sole has a curved concave shape as a whole, thereby forming a arch as a human being, and reducing unnecessary ground contact parts of the sole when standing or walking. It has been made so that it can.

  In this way, the robot 1 can take a wide support polygon when standing or walking, and can adjust the surface pressure to ensure an appropriate frictional force with the ground contact surface. 29 and 30, when supporting the toe grounding for supporting the robot 1 by grounding only the toes (respective finger parts 121A and 121B), or by contacting only the buttocks 122 as shown in FIGS. 31 and 32 29 when the heel grounding support for supporting the robot 1 is supported, and as in FIG. 29, the fingers 121A and 121B, the foot main body 120, and the heel 122 are all grounded to support the robot 1 in any state. The grounding point is appropriately secured and a sufficient supporting moment can be secured.

  For this reason, in the foot block 32, the front ends of the protrusions 121AX and 121BX, which are connected to the foot main body 120 provided at the rear end edges of the finger portions 121A and 121B, and the foot main body 120 Each protrusion formed by a connecting portion between the tip of each protrusion 120A formed by a connecting portion with each finger 121A, 121B provided at the front end edge and a heel portion 122 provided at the rear end edge of the foot main body 120. The front end portion of 120B and the front end portion of each projection 122A, which is a connecting portion between the foot main body portion 120 provided at the front end edge of the heel portion 122, are each formed with a smooth curved surface. When switching to the toe grounding support or the heel grounding support, it is possible to prevent an abrupt change in the state of the support state.

  Further, in the case of this robot 1, the outermost peripheral portions of the finger portions 121A and 121B, the foot main body portion 120, and the heel portion 122 are formed with a curvature, thereby preventing the catching due to the ground angle at the time of ground contact. It is made to be able to do.

(2-3) Effects obtained by the foot structure according to the present embodiment As the foot structure of the robot 1, the two fingers 121A and 121B and the buttocks 122 are rotated in the jumping up and down directions as described above. By adopting a freely provided structure, the following effects can be obtained.

(2-3-1) Motor performance and application performance With respect to the mechanism / control method for determining the sole position / posture, each finger 121A, 121B of the foot block 32 corresponding to the toe and the heel 122 corresponding to the heel By appropriately controlling each of these, the motion performance and application performance of the robot 1 can be improved.

(2-3-2) High-speed movement The robot 1 can move at high speed by installing the two finger portions 121A and 121B at the tip of the foot. This is because the range of motion that can be selected can be expanded in addition to the substantial lengthening of the legs. As shown in FIG. 33, it is possible to change the apparent ankle height by attaching toes (finger portions 121A, 121B) to the foot block 32, and the leg length or the distance between the moving surface and the ankle axis can be changed. Can be enlarged. Thereby, walking speed can be raised. In addition, by giving compliance as appropriate, an unprecedented leg trajectory is possible, and higher speed movement is possible.

  Usually, for walking, it is important to set the ankle shaft low from the viewpoint of reducing the total driving torque. However, this is not necessarily true when the unevenness of the moving surface topography becomes deep. Since there is interference between the leg mechanism and the moving terrain, this can be prevented by keeping the ankle height high.

(2-3-3) Ability to adapt to road surface irregularities The fact that the finger parts 121A and 121B operate downward and can be operated independently by two degrees of freedom, that is, a part, has a great effect.

  For example, as shown in FIG. 34, an arch can be formed by lowering the finger parts 121A and 121B and the heel part 122 from the foot main body part 120. Thus, it is possible to move without stepping on a small obstacle on the moving surface geometrically. That is, by using the finger parts 121A and 121B, the grounding part can be secured as much as possible even on rough terrain, and a high support moment can be generated.

  Further, as shown in FIG. 35, when an obstacle 150 that is not fixed to the ground exists on the moving surface, the obstacle 150 acts as a roller and is in a particularly dangerous state. According to the foot structure of the present embodiment, a arch is formed by giving a movable angle to the finger parts 121A, 121B and the buttocks part 122, and the risk of stepping on a small obstacle that is not fixed to the moving surface is eliminated. It is possible.

(2-3-4) Rough terrain mobility The ground surface is more or less uneven. Moreover, since they are scattered in a distributed manner, even a small one cannot be handled with a one-degree-of-freedom toe.

  Even if three support points can be formed on the foot, the support polygon tends to be small. Even if one degree of freedom is added to the toe, it is actually difficult to realize an ideal grounding state. This is because toe face grounding (four-point grounding) is not possible due to sensor errors, control errors, or other causes besides a slight unevenness. Furthermore, in uneven terrain, it is difficult to select the necessary ground plane.

  On the other hand, according to the foot structure according to the present embodiment, it is possible to improve the ground adaptability by equipping the two finger portions 121A and 121B that can operate independently. FIG. 36 shows a state in which the two finger portions 121A and 121B that can operate independently are adapted to an uneven road surface. As shown in the figure, when landing on rough terrain, if two degrees of freedom are used, reliable grounding can be realized, and a necessary supporting moment can be generated. This performance is a sole mechanism that can generate a support polygon that is effective even in normal walking and can generate a support moment sufficiently.

(2-3-5) Improvement of expressive power The robot is still in the progress stage and needs to show various performances. In the case of a humanoid robot, the expressive power of movement should also be considered as one of the performances. By giving compliance to the robot, the ground contact state of the sole can be selected even without precise force control. For example, when making a robot perform a butterfly motion, the degree of freedom of joints of the robot is generally insufficient for design reasons. This is because a normal leg does not have a redundant degree of freedom, and the degree of freedom may be substantially reduced depending on the posture.

  On the other hand, by adopting the foot structure according to the present embodiment, it becomes possible to perform a redundant operation, and the range of expressive power is expanded. Of course, it is a great merit that the shape of the foot itself can be freely changed by increasing the number of joints.

(2-3-6) Obstacle avoidance As the number of joints increases, the range of selection of the leg posture is expanded. As a result, the possibility of being able to deal with an obstacle placed on the moving surface increases. An increase in the ability to cross the steps of the stairs is also a manifestation of this effect.

(2-3-7) Buffering By installing the two finger parts 121A, 121B and the heel part 122 according to the present embodiment on the foot block 32, an impact absorbing effect can be obtained in most leg-type operations. Can do. In addition, unlike the case where the spring is simply attached, since it includes a characteristic that does not impair the controllability, the mechanism can be preferentially and selectively operated when it is necessary to protect the mechanism.

  Depending on the control (compliance setting) of the two finger parts 121A and 121B and the buttocks part 122, the state from the initial contact to the load can be controlled, and damage to the aircraft can be reduced. Only a passive or active control system can exhibit this effect in any of the combinations of these control elements.

(2-3-8) Ground contact sensing By configuring the compliance of the two finger parts 121A, 121B and the heel part 122 so as to be adjustable, it takes time to apply a load in earnest from the moment of toe grounding or heel grounding. Can do. As a result, a gait including discrimination of the ground contact state can be realized. For example, this operation is effective when the surface is relatively large and is not a large obstacle to avoid.

(2-3-9) Improvement of energy efficiency The foot structure provided with the two finger parts 121A and 121B and the heel part 122 according to the present embodiment includes an element that dissipates energy. Energy consumption does not always increase. This is because there are effects such as weight reduction and dispersion of peak output.

  Here, let us consider high-speed movement. Not all body motions are handled by the actuator, but the compliance element can temporarily receive it, or the damping element can handle the force in the direction of gravity, and the control operation itself can be changed. Therefore, the motor can be set small. As a result, the scale of the power source can be reduced, and a drive capacity can be afforded synergistically. And, a high-performance aircraft that is efficient in terms of energy can be realized. In addition, small actuators are generally very responsive and are very advantageous for achieving dynamic operation.

  By adopting the foot structure according to the present embodiment, the above-described effects can be obtained. Although it is different in nature from movement performance, the movement posture becomes beautiful and natural. Further, even if the gait leg stroke is the same, the substantial duty ratio is improved, so that more stable movement is easily realized.

(2-4) Walking control of robot 1 (2-4-1) Configuration of driving mechanism of each finger 121A, 121B and buttocks 122 Next, each fingertip 121A, 121B and buttocks in the foot block 32 A drive mechanism for driving 122 will be described.

  In the robot 1, the fingertip portions 121A and 121B and the heel portion 122 of the foot block 32 can be driven in conjunction with each other during walking or running. As means for that purpose, nonlinear pulleys, drive wires, and the like are provided. A driving mechanism (hereinafter referred to as a toe hook driving mechanism) is provided.

  In practice, as shown in FIG. 37, the toe-heel drive mechanism 130 includes a first nonlinear pulley 131 having an elliptical shape, for example, which is arranged with the pitch axis of the hip joint mechanism 36 and the central axis thereof aligned. In addition, one end side of the long diameter portion of the first nonlinear pulley 131 is connected to the tips of the finger portions 121 </ b> A and 121 </ b> B of the foot block 32 via the first drive wire 132. The first nonlinear pulley 131 is driven to rotate about the pitch axis integrally with the corresponding leg units 6A and 6B.

  Thereby, in the robot 1, when the leg units 6 </ b> A and 6 </ b> B are rotationally driven around the pitch axis, the first nonlinear pulley 131 rotates around the pitch axis integrally with the leg unit 6 </ b> A, 6 </ b> B. By pulling, the fingers 121A and 121B of the foot block 32 can be driven to rotate in the flip-up direction.

  In this case, the first drive wire 132 is formed by connecting the first and second wires 132A and 132B via the first buffer portion 133 made of a nonlinear spring or a damper mechanism, thereby The first buffer portion 133 can absorb external force in the downward direction acting on the finger portions 121A and 121B as necessary.

  In addition, a second nonlinear pulley 135 is arranged in the heel joint portion 134 so that the joint axis of the heel joint portion 134 coincides with the central axis thereof, and one end side of the long diameter portion of the second nonlinear pulley 135 is on the sole side. Are connected to the tips of the finger portions 121A and 121B of the foot block 32 via a second drive wire 136 disposed on the foot.

  As a result, in this robot 1, when the collar 122 is rotated in the jumping-up direction from the state in which the buttocks 122 are lowered in the lowering direction, the respective finger parts 121A and 121B are lowered in the lowering direction via the second drive wires 136. On the contrary, when each finger 121A, 121B is rotated from the raised state to the lowered direction, the second driving wire 136 is used to Rotational force in the downward direction can be applied to the part 122, thereby rotating the other in the downward direction in conjunction with the rotation of any one of the finger parts 121 </ b> A, 121 </ b> B and the collar part 122. It can be driven.

  In this case, similarly to the first drive wire 132, the second drive wire 136 connects the third and fourth wires 136A and 136B via the second buffer portion 137 formed of a nonlinear spring or a damper mechanism. Thus, the second buffer portion 137 can absorb the external force in the flip-up direction acting on the finger portions 121A and 121B and the collar portion 122 as necessary.

(2-4-2) Operation of each finger 121A, 121B and buttocks 122 during walking etc. Next, as shown in FIG. 38, for the operation of each finger 121A, 121B and buttocks 122 during walking, etc. The free leg period in the state where one leg unit 6A, 6B is lifted, the landing period immediately after landing the leg unit 6A, 6B, and the weight of the robot 1 after that, the two leg units 6A , 6B supported by both legs, and then the single leg supporting period in which the weight of the robot 1 is supported only by one leg unit 6A, 6B, and thereafter, the leg units 6A, 6B move in the direction in which the robot 1 moves. This will be described separately in the single-leg support period, in which the weight of the robot 1 is supported only by the leg units 6A and 6B.

(2-4-2-1) Free leg period In the free leg period, as shown in FIG. 39, the leg unit 6A, 6B rotates forward about the pitch axis of the hip joint mechanism part 36, so that the knee joint mechanism When the portion 38 moves upward and is pulled by the first drive wire 132 coupled to the first nonlinear pulley 131, a rotational force in the jumping direction is given to each finger portion 121A, 121B. Further, in conjunction with this, the second nonlinear pulley 135 is pulled by the second drive wire 136, whereby a downward rotational force is applied to the flange portion 122 via the second nonlinear pulley 135.

  As a result, in the robot 1, even when the finger parts 121A and 121B collide with unknown irregularities during the swinging legs, the impact force is applied to the passive characteristics of the direction of the finger parts 121A and 121B and the antagonistic drive by the first buffer part 133. Can be swept away. Accordingly, the movement of the leg units 6A and 6B as a whole can be prevented from being hindered, so that the robot 1 can be prevented from being overturned by rolling.

  Further, the torque output of the actuator 35 (FIG. 3) for the 36 pitch axis of the hip joint mechanism is allocated to drive the fingers 121A and 121B of the foot block 32 during the period when the torque is not required. Compared to the case where a new actuator is provided in the foot block 32 with many space constraints and the fingers 121A and 121B are directly driven by the output torque, the leg units 6A and 6B are significantly reduced in weight. In addition, more torque and speed can be applied to the finger joint 138.

(2-4-2-2) Landing period In the landing period, as shown in FIG. 40, the leg units 6A and 6B return slightly to the rear side, so that the pitch axis of the hip joint mechanism 36 is the first drive wire. It rotates in the direction of decreasing the tension of 132. Further, when the collar 122 rotates in the jumping direction by its own weight due to the subsequent landing, the second nonlinear pulley 135 rotates and pulls the second drive wire 136. Thereby, each finger part 121A, 121B rotates in the lowering direction, and accordingly, the corresponding tension is generated in the first drive wire 132. Therefore, at this time, the impact and vibration at the time of landing can be rapidly damped by the two-stage buffering action by the first and second buffer parts 133 and 137.

(2-4-2-3) Both-leg support period In the both-leg support period, as shown in FIG. 41, the heel part 122 of the leg unit 6A, 6B of the front leg and the finger part of the leg unit 6B, 6A of the rear leg A ZMP stable region is formed by 121A and 121B.

(2-4-2-4) Single Leg Support Period Early Period As shown in FIG. 42, the second non-linear pulley 135 is integrated with the flange 122 by the force of supporting its own weight, as shown in FIG. By rotating, the second drive wire 136 connected to the second nonlinear pulley 135 is pulled. As a result, the finger parts 121A and 121B are driven via the second buffer part 137, and the supporting force of the own weight and the tension of the second drive wire 136 antagonize, so that the finger joint 138 and the hip joint 134 are separated. To generate high rigidity and proper viscosity. This makes it possible to form a ZMP stable region between the distal end portion of the collar portion 122 and the distal end portions of the finger portions 121A and 121B without providing a special actuator.

(2-4-2-5) Late single leg support period In the late single leg support period, as shown in FIG. 43, when the ZMP moves forward and the torque applied to the buttocks 122 decreases, the second nonlinearity occurs. The tension of the second drive wire 136 that drives the finger parts 121A and 121B via the pulley 135 is weakened, and the rigidity of the finger joint 138 is lowered. As a result, the ZMP stable region is reduced, and the body of the robot 1 moves by an inverted pendulum system with the vicinity of the finger joint 138 as the center of rotation. When the finger joint 138 rotates above a certain level, the second drive wire 136 connected to the second nonlinear pulley 135 is pulled, the rigidity of the finger joint 138 is increased, and a ZMP stable region is formed.

(2-5) Walking Control Processing in Robot 1 Next, walking control of the robot 1 having the above-described foot structure will be described.

In this robot 1, the main control unit 50 described above with reference to FIG. 5 sets the trajectory plan of the leg units 6A and 6B for each step according to the walking control processing procedure RT1 shown in FIG. 44 and FIG. By driving and controlling the actuators A _{1 to} A ₁₇ , a walking motion can be performed.

  That is, when the main control unit 50 determines “walking” as the next action of the robot 1, the walking control processing procedure RT1 is started in step SP0, and in the subsequent step SP1, each leg unit 6A, 6B is set as a ground contact part. Set. Incidentally, when “advance by handstand” is determined as the next action of the robot 1, the arm units 5A and 5B are set as the ground contact parts. In addition, when the main control unit 50 determines “walking” as the next action, the main control unit 50 also determines the stride and the walking cycle (walking speed).

  Subsequently, the main control unit 50 proceeds to step SP2, and according to the stride and period at the time of “walking” determined at that time, the main control unit 50 maximizes within the ground contact part (leg unit 6A, 6B) set in step SP1. The contactable area that can form the support polygon (corresponding to the ZMP stable area in the both-leg support period shown in FIG. 41) is calculated.

  Therefore, in the case of “walking”, the contactable area where the maximum support polygon can be formed when the crotch is widened by a constant angle (step length) at a constant speed (cycle) is at the tip of the leg units 6A and 6B. It is the toe part (each finger part 121A, 121B), the foot main body part 120, and the heel part 122 that can be grounded in the sole, and the most prominent of these is the sole. Since it is a toe part, at the time of “walking”, the toe of the sole (each finger part 121A, 121B) is first calculated as a groundable area in this step SP2.

  In subsequent step SP3, the main control unit 50 sets the groundable area calculated in step SP2 as a part to be landed in the next one walking cycle (hereinafter referred to as a grounding part), and then proceeds to step SP4. Then, a trajectory plan for the whole robot 1 is made to operate the robot 1 so as to land the grounding part in the next one walking cycle with a predetermined stride and cycle.

  Then, the main control unit 50 thereafter processes step SP5 to step SP9 in parallel to determine whether or not such a trajectory plan can be realized.

  In practice, in step SP5, the main control unit 50 determines the acceleration to be generated in each joint mechanism unit to operate the robot 1 in accordance with the trajectory plan established in step SP4, depending on the robot hardware configuration such as actuator performance. Whether it is within the feasible range is determined based on the specification information stored in the external memory 66 (FIG. 5) in advance.

  If the main control unit 50 obtains a negative result in step SP5, it proceeds to step SP17. If it obtains a positive result, it proceeds to step SP10, and the acceleration of each joint mechanism unit in accordance with the trajectory plan. Is multiplied by a predetermined weighting factor a. The weighting factor a is determined according to the degree of importance that the acceleration to be generated in each joint mechanism unit in order to operate the robot 1 in accordance with the trajectory plan established in step SP4 is within a realizable range. It is a numerical value.

In step SP6, the main control unit 50 calculates an average torque of all the actuators A _{1 to} A ₁₇ within one walking cycle when the robot 1 is caused to walk according to the trajectory plan set in step SP4. It is determined based on the spec information or the like whether or not the average torque is within a predetermined allowable range.

If the main control unit 50 obtains a negative result in step SP6, it proceeds to step SP17. If it obtains an affirmative result, it proceeds to step SP11 and follows the trajectory plan within one walking cycle. Multiply the average torque of all the actuators A _{1 to} A ₁₇ by a predetermined weight coefficient b. The weight coefficient b is important because the average torque of all the actuators A _{1 to} A ₁₇ within one walking cycle when the robot 1 is operated according to the trajectory plan established in step SP4 is within an allowable range. It is a numerical value determined according to the degree.

  Further, in step SP7, the main control unit 50 realizes the speed to be generated in each joint mechanism unit to operate the robot 1 according to the trajectory plan set in step SP4, which is determined by the hardware configuration of the robot 1 such as actuator performance. Whether it is within the possible range is determined based on the spec information and the like.

  If the main control unit 50 obtains a negative result in step SP7, the process proceeds to step SP17. If an affirmative result is obtained, the main control unit 50 proceeds to step SP12, and the speed of each joint mechanism unit in accordance with the trajectory plan. Is multiplied by a predetermined weighting factor c. The weighting factor c is determined according to the degree of importance that the speed to be generated in each joint mechanism unit in order to operate the robot 1 in accordance with the trajectory plan established in step SP4 is within a realizable range. It is a numerical value.

On the other hand, in step SP8, the main control unit 50 causes the output torques of the actuators A _{1 to} A ₁₇ when the robot 1 is caused to walk according to the trajectory plan established in step SP4 to be within a predetermined allowable range. Is determined based on the spec information or the like.

When the main control unit 50 obtains a negative result at step SP8, the main control unit 50 proceeds to step SP17. When the main control unit 50 obtains an affirmative result, the main control unit 50 proceeds to step SP13, and the actuators A _{1 to} A when the trajectory plan is followed. The sum of the ₁₇ output torques is multiplied by a predetermined weight coefficient d. This weight coefficient d depends on the degree of importance that the sum of the output torques generated by the actuators A _{1 to} A ₁₇ in order to operate the robot 1 in accordance with the trajectory plan established in step SP4 is within an allowable range. It is a numerical value determined by

  On the other hand, in step SP9, the main control unit 50 has an operation range of each joint mechanism unit when operating the robot 1 in accordance with the trajectory plan established in step SP4 within a movable range determined by the hardware configuration of the robot 1. Is determined based on the corresponding information stored in the external memory 66 in advance.

  When the main control unit 50 obtains a negative result in step SP9, the main control unit 50 proceeds to step SP17. When the main control unit 50 obtains a positive result, the main control unit 50 proceeds to step SP14 and each joint mechanism unit required when following the trajectory plan. Is multiplied by a predetermined weight coefficient e. The weighting factor e is determined according to the degree of importance that the operating range of each joint mechanism is within the movable range when the robot 1 is operated according to the trajectory plan established in step SP4. It is a numerical value.

  Then, when the main control unit 50 finishes processing step SP5 to step SP9 and step SP10 to step SP14 in parallel as described above, the main control unit 50 proceeds to step SP15 and sets the numerical values obtained in step SP10 to step SP14 respectively to the predetermined values. By substituting into the equation, the evaluation value of the trajectory plan established in step SP4 is calculated.

  Further, the main control unit 50 thereafter proceeds to step SP16 to determine whether or not the calculated evaluation value is within a predetermined range, and if a negative result is obtained, the main control unit 50 proceeds to step SP17 and in step SP3. The set grounding portion is deleted from the target of the groundable area where the maximum support polygon can be formed, and then the process returns to step SP2, and thereafter the same processing is performed for step SP2 to step SP16 until an affirmative result is obtained in step SP16. repeat.

  As a result, for example, when the stride is large and the cycle (walking speed) is fast, the toe portion (finger portion 121A, 121B) is selected as the grounding portion, and when the stride is small and the cycle is slow, the foot body as the grounding portion When the part 120 is selected (so-called solid foot) and the stride and period are normal, the collar part 122 is selected as the grounding part, and the trajectory plan at that time is set.

When the main control unit 50 eventually obtains a positive result in step SP16, the main control unit 50 proceeds to step SP18 and finally sets the trajectory set up in step SP4 as a trajectory to be executed in the next one walking cycle, and thereafter, in step SP19 Proceeding and driving and controlling the actuators A _{1 to} A ₁₇ of the whole body of the robot 1 according to the set trajectory plan, the robot 1 is caused to perform a walking operation for one walking cycle. The main control unit 50 then returns to step SP1, and thereafter repeats step SP1 to step SP19 in the same manner.

  In this manner, the robot 1 can perform a walking motion under the control of the main control unit 50.

(3) Operation and effect of the present embodiment In the above configuration, in the robot 1, as described above with reference to FIGS. 38 to 43, the toe portion (finger Rotate the part 121A, 121B) in the jumping direction and rotate the collar part 122 in the downward direction, and rotate the toe part (finger part 121A, 121B) in the downward direction during the landing period of the leg unit 6A, 6B. Let

  Therefore, the robot 1 can avoid a fall due to the toe portion (finger portions 121A and 121B) colliding with an unknown unevenness during the swing phase. Further, during the landing period, the impact at the time of landing on the saddle is absorbed and alleviated by the first and second buffer parts 133 and 137 (FIG. 37) via the hook part 122 rotated in the lowering direction, and then the single leg. The impact at the time of transition to the first support period can be absorbed and alleviated by the second buffer portion 137 via the toe portions (finger portions 121A and 121B) rotated in the downward direction.

  Further, in this case, the robot 1 is constructed so that the toe portions (finger portions 121A and 121B) and the heel portion 122 can freely rotate in the jump-up direction and the lowering direction. The toe part (finger part 121A, 121B) and the buttock part 122 can be grounded in accordance with the unevenness of the grounding surface in the first stage, the first stage of the single leg support period, and the second stage of the single leg support period. Sufficient ground pressure and support moment can be obtained.

  According to the above configuration, during the swing leg period, the toe part (finger parts 121A, 121B) is rotated in the jumping direction and the buttock part 122 is rotated in the downward direction, so that the landing period of the leg unit 6A, 6B is reached. Since the toe part (finger part 121A, 121B) is rotated in the downward direction, the first and first impacts at the time of landing on the heel are transmitted via the toe part (finger part 121A, 121B) or the heel part 122. Can be absorbed and relaxed by the two buffer portions 137, and the toe portions (finger portions 121A and 121B) and the heel portion 122 are grounded by following the unevenness of the grounding surface, so that sufficient ground pressure and A robot that can obtain a supporting moment and thus improve walking performance can be realized.

(4) Other Embodiments In the above-described embodiment, the case where the present invention is applied to the biped robot 1 configured as shown in FIGS. 1 to 3 has been described. The present invention is not limited to this, and in short, any legged mobile robot apparatus having a plurality of movable legs can be widely applied to various other types of robot apparatuses.

  In the above-described embodiment, as described above with reference to FIGS. 25 to 27, the case where the heel has one degree of freedom has been described. However, the present invention is not limited thereto, and for example, FIGS. 25 to 27. As shown in FIG. 49 in which the corresponding parts are denoted by the same reference numerals, two heel portions 141A and 141B are arranged on the rear end edge of the foot main body portion 140 so as to have two degrees of freedom. May be.

  In this case, FIG. 50 shows the foot main body 141A, 141B with the joint axis inclined to the lower side in the roll direction and the forward side in the yaw direction inclined with respect to the pitch axis direction. This is an example of a foot structure attached to the rear edge of the portion 86, and the back surface (surface to be grounded) of the foot body portion 140 and the heel portions 141A and 141B is the same as that of the embodiment described above with reference to FIGS. Similar to the foot structure in the form, it is formed in a concave shape curved in an arch shape.

  Furthermore, in the above-described embodiment, the case where the toe has two degrees of freedom has been described. However, the present invention is not limited to this, and when the walking control method according to the present invention is employed, the toe has two degrees of freedom. You may make it give 1 or 3 or more degrees of freedom other than degree.

  Furthermore, in the above-described embodiment, the toe heel drive mechanism 130 that drives the toe portions (finger portions 121A and 121B) and the heel portion 122 is used as the first and second nonlinear pulleys 131 and 135, and the first and second drives. Although the case where the wires 132 and 136 and the first and second buffer portions 133 and 137 are configured has been described, the present invention is not limited thereto, and various other configurations can be widely applied.

  In this case, when the walking control method of the robot apparatus according to the present invention is adopted, the toe portions (finger portions 121A and 121B) and the heel portion 122 are rotated or driven in the jumping-up direction or the lowering direction independently or in conjunction with each other. For this purpose, the toe part (finger part 121A, 121B) and the heel part 122 are respectively driven as described above with reference to FIGS. Also good.

  The present invention can be applied to various types of legged mobile robot devices in addition to biped walking robot devices.

It is a rough-line perspective view which shows the external appearance structure of the robot by this Embodiment. It is a rough-line perspective view which shows the external appearance structure of the robot by this Embodiment. It is a conceptual diagram with which it uses for description of the freedom degree in each joint mechanism part etc. of a robot. It is a block diagram with which it uses for description of the internal structure of a robot. It is a block diagram with which it uses for description of the internal structure of a robot. It is the front view, top view, and side view which show an example of the foot structure which has two finger parts and a heel part. It is a side view with which it uses for description of the movable range of a finger | toe part and a collar part. It is the front view, top view, and side view which show an example of the foot structure which has two finger parts and a heel part. It is the front view, top view, and side view which show an example of the foot structure which has two finger parts and a heel part. It is a basic diagram which shows the mode of operation | movement at the time of setting the height of the joint axis of a finger high from the sole. It is a basic diagram which shows a mode that the joint axis | shaft of a finger | toe was set to the low position from the sole. It is a perspective view which shows the structural example about the shape of a joint-axis junction part vicinity. It is a side view with which it uses for description of the specific height position of a finger joint axis | shaft. It is an approximate line figure showing an example of arrangement of an actuator in a movable leg. It is the block diagram which showed the structural example of the linear motion mechanism which implement | achieved the concept of the mixed control system which consists of an active servo control system and a passive mechanism control system. It is the characteristic view which showed the elastic characteristic of the normal spring. It is a characteristic view showing an elastic characteristic of a spring when a limiter mechanism and a preload mechanism are applied. It is a characteristic view showing an elastic characteristic of a spring when a limiter mechanism and a preload mechanism are applied. It is the characteristic view which showed the characteristic of the damper used as a basic model. It is a characteristic figure showing the characteristic of the damper which gave the different characteristic by direction. FIG. 6 is a characteristic diagram showing a speed-dependent damper characteristic. It is a characteristic curve diagram showing a damper characteristic to which a nonlinear characteristic for preventing a steep change is given. It is a side view which takes a partial cross section and shows the nonlinear compliance mechanism which combined the spring, the damper, and the switching mechanism organically. It is the side view which showed the structural example of the mechanism unit which combined the spring, the damper, and the switching mechanism organically. It is a perspective view which shows the specific structure of the foot structure by this Embodiment. It is a perspective view which shows the specific structure of the foot structure by this Embodiment. It is a perspective view which shows the specific structure of the foot structure by this Embodiment. It is the front view, side view, and back view with which it uses for description of the bottom face shape of the foot structure by this Embodiment. It is the front view, side view, and back view with which it uses for description of the toe grounding support in the foot structure by this Embodiment. It is the front view, side view, and back view with which it uses for description of the toe grounding support in the foot structure by this Embodiment. It is the front view, side view, and rear view with which it uses for description of the heel grounding support in the foot structure by this Embodiment. It is the front view, side view, and rear view with which it uses for description of the heel grounding support in the foot structure by this Embodiment. It is a rough-line side view with which it uses for description of the effect acquired by the foot structure by this Embodiment. It is a rough-line side view with which it uses for description of the effect acquired by the foot structure by this Embodiment. It is a rough-line side view with which it uses for description of the effect acquired by the foot structure by this Embodiment. It is a rough-line side view with which it uses for description of the effect acquired by the foot structure by this Embodiment. It is a basic diagram which shows a toe hook drive mechanism. It is a basic diagram with which it uses for description of the flow of walking operation. It is a basic diagram with which it uses for description of the state of a toe and a buttocks in a free leg period. It is a basic diagram with which it uses for description of the state of a toe and a buttock in a landing period. It is a basic diagram with which it uses for description of the state of a toe and a buttock in both legs support period. It is a basic diagram with which it uses for description of the state of a toe and a buttocks in the single leg support period first half. It is a basic diagram with which it uses for description of the state of the toe in a late stage of a single leg support period, and a buttocks. It is a flowchart which shows a walk control processing procedure. It is a flowchart which shows a walk control processing procedure. It is a conceptual diagram which shows the walk state in the conventional foot structure. It is a conceptual diagram which shows the walking state in the foot structure by this Embodiment. It is a conceptual diagram which shows the driving | running | working state by the foot structure by this Embodiment. It is the front view, side view, and back view which show other embodiment. It is a side view with which it uses for description of the foot structure proposed conventionally. It is sectional drawing with which it uses for description of the example of the cage structure of the robot proposed conventionally.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Robot, 6A, 6B ... Leg unit, 36 ... Hip joint mechanism part, 38 ... Knee joint mechanism part, 41 ... Ankle mechanism part, 130 ... Toe heel drive mechanism, 131, 135 ... Non-linear Pulley, 132, 136 ... Drive wire, 133, 137 ... Buffer part, 50 ... Main control part, 120, 140 ... Foot main body part, 121A, 121B ... Finger part, 122, 141A, 141B ... 踵Department.

Claims (8)

In a legged mobile robot device having a plurality of movable legs,
A foot main body disposed at the lowermost end of the movable leg;
A toe part rotatably attached to the front end part of the foot main body part,
A heel portion rotatably attached to a rear end portion of the foot main body portion;
Driving means for rotationally driving the toe part and the heel part in the jumping-up direction or the lowering direction;
Buffer means for reducing impact in the direction of jumping or lowering applied from the outside applied to the toe portion or the heel portion,
The drive means is
During the free leg period of the movable leg, the toe portion is rotated in the jump-up direction and the buttocks portion is rotated in the lowering direction, and when the movable leg is landed, the toe portion is rotated in the lowering direction. A robotic device characterized by The drive unit is
The robot apparatus according to claim 1, wherein after the landing of the movable leg, the buttocks are rotated in the downward direction while the toe portion is positioned in the downward direction. The drive unit is
The robot apparatus according to claim 1, wherein the toe portion and the heel portion are interlocked to rotate in a jumping direction or a lowering direction. In a walking control method for a legged mobile robot apparatus having a movable leg,
The robot apparatus is
A foot main body disposed at the lowermost end of the movable leg;
A toe part rotatably attached to the front end part of the foot main body part,
A heel portion rotatably attached to a rear end portion of the foot main body portion;
Drive means for rotationally driving the toe part and the heel part in the jumping-up or lowering direction;
Buffer means for reducing the impact in the jumping direction or the lowering direction from the outside applied to the toe part or the hook part,
A first step of rotating the toe part in the jump-up direction and rotating the buttocks in the lowering direction during the swing leg period of the movable leg;
A walking control method for a robot apparatus, comprising: a second step of rotating the toe portion in a downward direction when the movable leg is landed. 5. The robot apparatus according to claim 4, further comprising a third step of rotating the hook part in the downward direction while the toe part is positioned in the downward direction after the landing of the movable leg. Walking control method. The walking control method for a robotic device according to claim 4, wherein the toe portion and the buttocks are rotated in a jumping direction or a lowering direction in conjunction with each other. In a legged mobile robot device having movable legs,
A foot main body disposed at the lowermost end of the movable leg;
A plurality of toe portions rotatably attached to the front end portion of the foot main body portion by joint axes in directions different from the pitch axis direction of the robot body;
A robot apparatus comprising: a heel portion rotatably attached to a rear end portion of the foot main body portion. The foot main body part, the toe part and the heel part are
The robot apparatus according to claim 7, wherein each back surface is formed in a concave shape.