JPH0994784A - Gait generating method for leg type walking robot and walking controller - Google Patents

Abstract

PROBLEM TO BE SOLVED: To quantitatively design an output of an actuator and an action speed and to facilitate generation of a gait by estimating a position of an upper body in a finish of kicking and deciding a lifting quantity of a heel part in a foot fin so as to position the upper body in the estimated position at least. SOLUTION: Revolute joints 10R, 10L for leg part rotation in a crotch, revolute joints 12R, 12L for the roll direction in the crotch, revolute joints 14R, 14L for the pitch direction in the crotch, revolute joints 16R, 16L for the roll direction in knee parts, revolute joints 18R, 18L for the roll direction in foot parts, and revolute joints 20R, 20L for the pitch direction in the foot parts are individually arranged in this order from the upper sides to the lower sides in respective leg part links 2 on the left side and the right side. Foot fins 22R, 22L are mounted in the foot parts, while a control unit 26 is stored inside an upper body 24 arranged in the uppermost part. By means of the control unit 26, the position of the upper body 24 in a finish of kicking is estimated, and a heel part lifting quantity for each of the foot fins 22R, 22L is decided so that the upper body 24 reach the estimated position at least, and on the basis of the lifting quantity, a gait is generated.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a gait generating method and a walking control device for a legged walking robot (hereinafter referred to as "a gait generating method for a legged walking robot"), and more specifically, a human (human). The present invention relates to a gait generating method and the like for optimally designing an actuator of a joint of a self-supporting bipedal walking robot having a similar size and effectively utilizing the robot such as mechanical strength and actuator ability of the robot.

[0002]

2. Description of the Related Art Most conventional bipedal robots are intended for control research rather than for practical use, and have a control computer, a hydraulic power source as a power source, and a drive battery outside. The things were mostly. Many of them are small both in terms of size and weight, and as a result, examples of detailed examination of gait design from a practical point of view, such as mitigating impact force from the floor surface (ground contact surface) and improving energy efficiency, Less physical walkability, eg ZMP
Although it satisfies the theoretical minimum condition such as being in the ground contact plane of the foot, many of them realize walking without designing the detailed conditions very strictly. Even with such a method, if the robot is small and lightweight, no big problem occurs.

Incidentally, ZMP (Zero Moment Point) is
“Pitch axis (axis in the frontal plane on the left and right sides) and roll axis (axis in the sagittal plane on the sagittal plane) on or inside the sides of the supporting polygon formed by the foot contact point and the road surface at every moment of walking. ) The point at which the moment becomes zero. "

As a conventional technique that is considered to include a practical point of view, there is a technique proposed in Japanese Patent Publication No. 6-88218. This is focused on the optimization of the torque generated by the actuator, but the optimization of the generated torque peculiar to a legged walking robot with a foot and the optimization considering the relaxation of the impact force from the floor surface are also examined. Is not described.

[0005]

However, the weight of an autonomous bipedal robot having a size comparable to that of a human is often much larger than that of a normal human,
It is important for the feasibility of a practical system, such as the mechanical strength being hindered unless the impact force from the floor surface (ground contact surface) is mitigated, and the need to improve energy efficiency in order to extend continuous operation time. New challenges arise.

The gait design or generation in consideration of such conditions is not enough to satisfy only the physical walking condition, for example, the condition that the ZMP is in the ground contact plane of the foot, and the output of the actuator. It is necessary to simultaneously satisfy the hardware conditions such as and the operation speed, and the conditions such that the impact force from the floor surface is kept below a certain level.

However, although there are countless gaits that can physically walk, all of them do not necessarily take into consideration the practical constraints described above. Therefore, it is not easy to select and design a gait satisfying such conditions from among them. Moreover, the bipedal robots used in various fields have different sizes and the walking speed is not uniform. It is more difficult to design an optimal gait for a bipedal walking robot that differs in such size and walking speed.

However, in designing the output of the actuator of the robot, the operation speed, etc., the gait at the time of actually walking must be taken into consideration. For this reason, it is theoretically possible to study the optimal gait in advance in consideration of all usage patterns, but in reality, it is often the case that optimization is advanced by repeating experiments using an actual machine, which is a great deal. It takes effort and time.

A practically valid gait design can be made offline or in real time, and when the application of the robot is determined, the specifications of the output and operating speed of the actuator that executes the robot are determined by the application. Will be decided to satisfy the most severe conditions. However, even in that case, it is not enough to use an actuator having a good performance and to increase the mechanical strength blindly in terms of physical and system costs. Therefore, it is natural that a walking form that is as comfortable as possible is desired.

Therefore, an object of the present invention is to realize a practical bipedal walking robot without the detailed design of complicated gaits, and to countless gaits satisfying only a physical walkable condition. Among them, we provide a gait generation method for legged walking robots that enables quantitative design of actuator output and motion speed, and limits gaits that are practically appropriate to facilitate gait generation. To do.

[0011]

According to the first aspect of the present invention,
At least an upper body and two leg links connected to the upper body via hip joints, the two leg links being respectively connected to the thigh links and the thigh links via knee joints. Walking of a bipedal legged walking robot, which comprises a lower leg link and a foot connected to the lower leg via an ankle joint, and the two leg links alternately kick the floor to move. In the volume generating method, the position of the upper body at the time of finishing the kick is estimated, the amount of raising the heel of the foot is determined so that the upper body reaches at least that position, and the gait is calculated based on the position. It was configured to generate.

According to a second aspect of the present invention, it is estimated that the upper body has moved a predetermined distance forward from the midpoint between the front foot and the rear foot at the time when the kick is finished, and the upper body reaches at least the estimated position. As described above, the amount of raising the heel of the foot is determined, and the gait is generated based on the amount.

According to a third aspect of the present invention, the raising amount of the heel portion of the foot is set to the lower limit value of the raising amount, and the hip joint, the ankle joint, and the tip portion of the foot are located on substantially the same line. The amount of increase of the heel portion is set as an upper limit value of the amount of increase, and the amount of increase of the heel portion of the foot is determined within the range of the lower limit value and the upper limit value.

According to a fourth aspect of the present invention, the amount of raising the heel of the foot is determined from the average value of the lower limit value and the upper limit value.

According to a fifth aspect of the present invention, the foot has a bottom surface formed in an arc shape near its tip, and / or the foot has a toe joint, and the arc bottom surface is grounded. And / or the heel of the foot is raised via the toe joint.

According to a sixth aspect of the present invention, there is provided an upper body and at least two leg links that are connected to the upper body via hip joints and are arranged in parallel, and the leg joints are provided at the ends of the leg links. In a legged walking robot having a foot that can be tilted in the direction of gravity connected to each other, the gait of the robot is basically composed of one leg (leg) supporting period and both legs (leg) supporting period,
Walking shall consist of continuing these two support periods alternately at a predetermined pitch, and
The transition from the two-leg supporting period to the one-leg supporting period is configured such that one of the ankle joints is rotated to start tilting of the foot.

According to a seventh aspect of the present invention, the control of the amount of tilting of the foot is performed by a relational expression of thH = f (PS, ST) (where thH: amount of tilting (angle), PS: both feet supporting period). The time ratio, ST: step length) is used.

In the eighth aspect, the control of the amount of tilt of the foot is performed by the following relational expression of thH = π-thA-thB-thC (where thH: tilt amount (angle), thA,
B, C: It is configured to be performed using a predetermined tilt amount (angle) determined from the posture of the robot.

[0019]

[Advantage] Since it is possible to determine in advance the amount of raising of the heel part of the kicking foot that can realize a gait with a high center of gravity trajectory, that is, a high energy efficiency, it is possible to narrow down the gait candidates and significantly design the gait. It can be simplified.

[0020]

BEST MODE FOR CARRYING OUT THE INVENTION First, a bipedal walking robot based on the gait generating method according to the present invention will be described.

FIG. 1 is an explanatory skeleton diagram showing the robot 1 as a whole.
Provided with individual joints (each joint is indicated by an electric motor driving it for the sake of convenience). The six joints are, in order from the top, joints 10R and 10L for rotating the legs of the crotch (R on the right side,
The left side is L. The same shall apply hereinafter), joint 12R in the crotch roll direction (around the y axis, movement within the sagittal plane of the sagittal plane),
12L, same pitch direction (around x axis. Left and right frontal pl
(motion in the ane) joints 14R, 14L, knee joints 16R, 16L in the roll direction, foot joints 18 in the roll direction
R, 18L and joints 20R, 20L in the same pitch direction.

Foots 22R and 22L are attached to the feet, and an upper body (base) 24 is provided at the top, and a control unit 26 including a microcomputer, which will be described later with reference to FIG. Is stored. In the above, the hip joint is joint 10R (L), 12R (L), 14R
From (L), the ankle joint is joint 18R (L), 20R (L)
Consists of Also, the hip and knee joints are thigh links 2
8R, 28L, lower leg link 30R for knee joint and ankle joint,
It is connected with 30L.

With the above structure, the leg link 2 is provided with six degrees of freedom for each of the left and right feet, and by driving these 6 × 2 = 12 joints at an appropriate angle during walking, A desired motion can be given to the whole, and the robot arbitrarily walks in a three-dimensional space.

FIG. 2 is a sectional view of the foot portion (cut along the sagittal plane). The output of the electric motor (not shown in FIG. 2) that drives the ankle joint 18 is a harmonic speed reducer (trade name. FIG. 2).
Is input to the input end of the fixed member 32 attached to the lower leg link 30 and the rotating member 34 below the axis 36 (the ankle joint 18R). The foot 22R (L) is relatively rotated (rotated) in the walking direction (in the sagittal plane) about the axis of (L), and the foot 22R (L) is tilted in that direction.

An electric motor 20M for driving the above-mentioned ankle joint 20 is arranged at a position orthogonal to the axis 36,
The output is input to the second harmonic speed reducer 40,
The fixed member 32 and the rotating member 34 are connected to each other by a second axis 42
The foot 22R (L) is tilted in that direction relative to the horizontal direction (in the frontal plane) orthogonal to the traveling direction about the same as the axis of the ankle joint 20R (L).
The details of the configuration of the robot including the ankle joint are previously proposed by the present applicant (Japanese Patent Application Laid-Open No. 3-184,782).
No.) etc., so further explanation is omitted.

A well-known 6-axis force sensor 44 is attached below the rotary member 34 to measure the three-direction components Fx, Fy, Fz of force and the three-direction components Mx, My, Mz of moment,
Detects the presence or absence of landing on the foot or ground load. 6
The foot 22R (L) described above is provided below the axial force sensor 44.
The flat plate-shaped frame 46 constituting the above is fixed. The lower surface (sole) of the frame 46 is formed to be substantially flat, and elastic bodies 50 and 52 that absorb the impact at the time of landing are attached to the toe portion 46a and the heel portion 46b.

Further, in the robot shown in FIG. 1, an inclination sensor 60 is installed on the body 24, and the inclination and its angular velocity with respect to the z axis (gravitational direction) in the left and right planes, as well as with respect to the z axis in the sagittal plane. The tilt and its angular velocity are detected. The electric motor of each joint is provided with a rotary encoder that detects the amount of rotation. Although not shown in FIG. 1, an origin switch 62 for correcting the output of the tilt sensor 60 and a limit switch 64 for fail protection are provided at appropriate positions of the robot 1. These outputs are sent to the control unit 26 in the body 24 described above.

FIG. 3 is a block diagram showing the details of the control unit 26, which comprises a microcomputer. There, the output of the tilt sensor 60 and the like is converted into a digital value by the A / D converter 70, and the output is output by the bus 72.
Is sent to the RAM 74 via. The output of the encoder arranged adjacent to each electric motor is input into the RAM 74 via the counter 76, and the output of the origin switch and the like is also stored in the RAM 74 via the waveform shaping circuit 78.

The control unit includes a first CPU,
Second computing devices 80 and 82 are provided, and the first computing device 80 reads gait parameters generated as described below and stored in the ROM 84 to calculate a target joint angle, and sends the target joint angle to the RAM 74. Further, the second arithmetic unit 82 reads out the target value and the detected actual value from the RAM 74 as will be described later, calculates the control value necessary for driving each joint, and outputs the control value via the D / A converter 86 and the servo amplifier. Output to an electric motor that drives each joint.

Based on the bipedal robot as described above, a gait generating (gait design) method and the like according to the present invention will be described below.

1) General Theory of Gait Design of Biped Robot In gait design, it is necessary to determine the walking speed, the stride, the way of walking and the way of moving the center of gravity.

As described above, it is well known that the ZMP must be present in the ground contact plane of the foot as a physically possible gait constraint.

Further, in order to develop a practical robot, the output of the actuator of the joint and the operation speed are appropriately set, the impact force from the floor surface is reduced, and the energy efficiency of the entire walking is increased. It is necessary to consider that

When the walking is performed with a low center of gravity, the knee will be bent, so that a large holding torque is always required. Therefore, it is advantageous in terms of energy efficiency that the trajectory of the center of gravity is as high as possible. That is, it is advantageous to move the support legs by moving in an arc around the support point like a stilt.

On the other hand, if the trajectory of the center of gravity is made too high, the center of gravity is inevitably lowered as the supporting legs fall, so that the center of gravity moves up and down. If the vertical movement is large,
Since the upper and lower components of the floor surface reaction force increase, the ground contact with the floor surface becomes poor and the impact force from the floor surface increases.

Furthermore, in order to reduce the size and weight of the actuator, it is necessary to minimize the output of the actuator. For that purpose, the gait should be set so that the maximum torque and the maximum angular velocity required for each joint are minimized. You have to design. Practical gait design involves finding a compromise while considering these conflicting conditions.

2) Actuator Specification Design Procedure In designing the actuator specifications, it is necessary to determine the maximum torque and the maximum rotation speed. If the maximum torque and the like are determined, the specifications including the maximum output and the necessity of the speed reducer can be determined based on the determined maximum torque.

The maximum torque is determined by the gait, and strictly speaking, it should be calculated dynamically. However, there is often no problem even if it is roughly calculated by performing static calculation from the gait and its own weight. The maximum angular velocity of the hip joint of the supporting leg is almost determined by the walking speed and the length of the one-leg supporting period because the next step must be taken during the one-leg supporting period.

For this purpose, it is necessary to determine the stride length and gait (walking cycle (pitch), stepping cycle) in walking. If the walking speed is determined, the walking speed = step length × step, so it is sufficient to decide one of the step and the step. The stride is closely related to the maximum torque of the hip joint, and the gait is closely related to the maximum angular velocity of the joint.

The maximum angular velocity of the knee joint largely depends on how much the free leg is lifted from the floor and how the foot 22R (L) of the free leg is carried, such as whether the leg is put out early or late. To do.

Similarly, the maximum angular velocities of the ankle joints 18 and 20R (L) greatly depend on how the foot of the idle leg is carried. Also,
The hip joints 10, 12, 14R (L) of the free leg are also affected by the movement of the knee. Therefore, it is necessary to consider how to carry the foot 22R (L), that is, to design and evaluate the gait.
It is desirable that the gait to be designed is the optimum one that maximizes the performance of the actuator.

If the gait can be determined, the maximum angular velocity of each joint is automatically determined from the gait, and the required torque and the maximum output of the actuator can be estimated as described above. Design becomes possible.

A detailed description will be given below.

3) Actuator Design Procedure Including Gait Design Procedure As described above, the final purpose of gait design is to determine the specifications of the joint actuator, and determine the mechanical strength and actuator capacity of the robot according to the gait. Although the robot can be effectively used to walk, in the present invention, the determination of the joint actuator specifications is as shown in FIG. Hereinafter, description will be given with reference to FIG.

In the gait design procedure described here, first, the walking speed and the stride are selected in S10.

In actual walking, the walking speed and the stride vary depending on the purpose of walking. There are also cases where the designer cannot arbitrarily select and must decide based on the situation. For example, there are cases where the walking speed cannot be increased in a place where walking is difficult, or where the foot landing position is limited and the stride is determined.

In determining the joint actuator specifications, the most severe one, that is, the maximum walking speed and the maximum stride, may be selected from those that may actually be used for walking. For example, the maximum walking speed is set to 5 km / h and the maximum stride is set to 750 mm.

Studies on human walking morphology (for example, “Influence of gait stride on gait parameters in normal gait”)
Iida et al., Orthopedic Biomechanics, Vol. 5,19
83. According to (), the ratio of both feet supporting period to one walking period is about 15%, which means that about 30% of the time required to take one step is the both feet supporting period.

However, this walking is related to the characteristic of human walking that the heel rises up to near the vertical when the kicking foot leaves the floor. That is, the biped robot does not necessarily have to take the same walking form as a human. Experiments have shown that about 20%, which is considerably smaller than that of humans, is practically desirable in consideration of the influence on the operation speed of the ankle joints 18 and 20R (L) described later.

Based on the above facts, in S12, the time ratio to the walking cycle in the both-leg supporting period is set.

Next, in S14, the foot 22R (L) of the free leg at the time of leaving the free leg in the two-leg supporting period is next determined from the step length thus determined and the time ratio to the walking cycle in the two-leg supporting period.
Is tilted in the direction of gravity, that is, the amount of lifting of the heel portion 46b of the foot 22R (L) is determined, and the process proceeds to step S16 to set conditions for carrying the foot such as the amount of lifting the foot.

The trajectory of the foot 22R (L) may be generated by a spline function or other means for generating a smooth curve within a range satisfying the trajectory condition, but in any case, In the trajectory of the flat 22R (L), the amount of raising the heel portion 46b at the time of leaving the supporting legs during the two-leg supporting period is also an important gait parameter.

The necessity and method of determining the amount of raising of the heel portion 46b will be described below. As a gait of a bipedal walking robot, in terms of energy, while sliding on the floor without raising the foot of the free leg. Walking is the most efficient way,
In such a gait, various disturbance shapes such as unevenness existing on the actual walking road surface are likely to trip. Therefore, although the amount varies depending on the situation, it is desirable to raise the free leg to at least the minimum required. The amount to be raised depends on the situation such as the unevenness of the road surface, so a value corresponding to it may be set appropriately.

If the free leg is raised high, the knee is also bent greatly, and the maximum angular velocity of the knee joint 16R (L) naturally increases. Also related hip joints 10, 12, 14R (L)
Also increases the angular velocity of. Therefore, it is practical to select the minimum height that is sufficient for the severest of the practical use conditions. For example, an orbital condition such as “raise at a height such that a block having a width of 100 mm and a height of 30 mm is straddled” may be set.

Like some stilt-like robots that do not have a foot, it is possible to walk mechanically without raising the heel. In addition, it is known that the human gait raises a considerable amount of heels as described in the above-mentioned research on human walking morphology.

When a robot takes the same walking form as a human,
This raises the heel of the foot to the vicinity of the vertical position, which increases the movable angle and the maximum joint angular velocity, resulting in an increase in the actuator output, which is not preferable in terms of design. In particular, in a fast walk in which the time for supporting both feet is shortened, the angular velocity of the ankle joint increases, which is a problem.

Considering only from such a point of view, it is qualitatively good to say that in the case of bipedal walking by a robot, it is qualitatively good to reduce the amount of raising the heel.

However, it can be easily imagined that if the heel part is not raised at all, the trajectory of the center of gravity of the upper body is low in the height in the direction of gravity and the knee has a bent walking form. At the same time, even in the operation of raising the foot, the foot portion is pulled up faster by the amount that the heel portion is not raised during the both-foot supporting period, and as a result, the bending speed of the knee of the free leg may be increased.

In the gait of a bipedal walking robot, the upper body 2
The amount of vertical movement of 4 (movement in the direction of gravity) is also an important gait parameter. Since the length of the leg link changes depending on the bending and stretching of the knee joint 16R (L), it cannot be uniquely determined only by the leg link length or the like. However, it is possible to determine by a constraint condition such as “select the highest geometrically possible body trajectory”.

If the body trajectory is lower than that, it can be geometrically set arbitrarily. Since the vertical movement of the upper body has a large effect on landing impact, etc., it is actually set to a trajectory with its amplitude reduced. Naturally, the reduction will be realized by lowering the body trajectory.

4) Determining the Appropriate Range of the Amount of Raising the Heel First, changing the amount of raising the heel while changing the amount of raising the heel under the constraint that the vertical G (displacement acceleration in the z direction (gravitational direction)) is kept below a certain value. FIG. 5 shows the relationship of how the height of the center of gravity trajectory of the upper body 24 changes. From this, it can be seen that the height of the track becomes extremely low when the amount of raising the heel portion becomes smaller than a certain angle. It can be understood that this is because the upper part of the body is low because the back of the foot cannot reach because the heel does not rise.

Based on these results, in the method according to the embodiment, based on the idea of raising the center of gravity of the upper body by an amount sufficient to maintain a high trajectory, the appropriate range of the amount of raising the heel portion is determined. Decided to decide. That is, the heel portion was raised so that the kicking foot, which is the hind foot during walking, can support the upper body while maintaining a high center of gravity trajectory.

In order to make a decision based on this idea, it is necessary to estimate in advance the position of the upper body at the time when the kick is finished. That is, the height of the center of gravity trajectory of the body 24 in the direction of gravity is set to a desired value, and the position of the body 24 at the time when the kick is finished is estimated.

As described above, since the walking form has an extremely large degree of freedom, the exact position of the upper body cannot be uniquely determined. However, according to the result of examining a large number of walking forms, the approximate calculation can be easily performed as follows.

5) Approximate calculation of the position of the upper body at the time of finishing the kick (the end of both legs supporting period) Here, in the robot described above, the thigh link 28 is used.
The length of R (L) and lower leg link 30R (L) is 400 mm.
And a walking speed of 1 to 4 km / h will be described. Since it is not possible to walk statically at this speed, it becomes a region of complete dynamic walking,
This is sufficient for studying the walking form in dynamic walking.

In order to briefly explain the problem, the bidirectional movement of the bipedal walking robot (three-dimensional movement) (frontal plan
motion in e) is not very relevant to this problem, so
I don't think here.

In addition, the bidirectional walking motion (sagitt
in the al plane), the weighted center of the foot (ZM
The way of moving P) also affects the walking form.

Although the weighted center of the foot in a normal human walking form varies irregularly, it is on the average near the center of the foot. The robot shown in the figure does not necessarily have to be the same, but the weighted center of the foot should be the foot 22R.
If the geometrical center of (L) is set near the geometrical center, there is an effect of increasing the generation margin of the restoring force against the disturbance, and therefore, in practice, it is considered to be an effective walking form even for a robot.

In this embodiment, in this way, basically, in the vicinity of the geometric center of the foot 22R (L),
More specifically, the weighted center of the foot (shown by P in FIG. 2) is set 30 mm closer to the traveling direction than the projection point of the intersection of the ankle joints 18 and 20R (L) (the intersection of the axes 36 and 42). still,
Of course, this is not the case when it is difficult to ground the entire foot on the road surface, such as walking on an uneven road surface or a narrow landing point.

As a result of examining various walking forms under such conditions, the following results were obtained.

The locus of movement of the center of gravity of the body 24 is accompanied by vertical movement in the z direction, as is sometimes modeled as an inverted pendulum. As a result, the moving speed of the body in the x and y directions also fluctuates. In normal walking, the height of the upper body is low during the period of supporting both legs, and the moving speed is higher than the average walking speed of the whole body. As shown in FIG. 6, the fluctuation of the moving speed is remarkable in the low speed walking and is small in the high speed walking. The size of the robot of this embodiment is about 2 at 2 km / h.
However, at 3 km / h, about 10% is faster than the average walking speed (constant speed walking).

As for the size of this robot, it is nearly 70% faster at 1 km / h, which can be said to be low-speed walking. However, extreme speed fluctuations during low-speed walking are close to static walking. In such a walking mode region, the amount of raising the heel can be designed in a static walking manner, and it hardly affects the vertical G and the energy efficiency. Therefore, it is necessary to apply the design method according to the present invention. There is no. Therefore, in a walking speed region where the speed fluctuation is not significantly large, the walking speed, which is the average moving speed of the upper body, can be approximated.

During the period of supporting both legs, the center of gravity moves from the hind legs to the front legs. In Figure 7, walking speed 1-4km
The position of the upper body 24 at the time of supporting both legs at / h is shown. At this time, the position of the upper body 24 in the supporting period for both feet is near the midpoint between the ankle joints 18 and 20R (L) of the forefoot and hindfoot, and it can be seen that the relationship with the walking speed is small.

Further detailed examination of the position of the upper body shows that the position of the weighted center of the foot during the supporting leg period and the angle at which the heel 46b of the foot of the hind foot, which is a kicking foot, are raised are slightly dependent. .

FIG. 8 shows a walking speed of 2 km / h and a stride of 500.
The change in the position of the center of gravity of the upper body when the weighted center of the foot 22R (L) is moved back and forth from the geometric center of the foot in mm is shown. The geometric center of the foot is, as described above, 30 mm ahead of the position of the ankle joint (point P in FIG. 2). If you move the weighted center forward,
It turns out that there is a tendency to move forward. It can be read that the amount of movement of the position of the upper body moves in proportion to the amount of movement of the weighted center forward. However, the amount of movement is not the same, and the position of the upper body is smaller.

When the weighted center is set to the above-mentioned point P, the upper body 24 moves 50 mm before the midpoint when landing and 70 mm ahead when leaving the floor. Upper body 2 for both legs
On the average, the position of 4 is shifted by about 10 mm before the middle point.

Also, at the angle of raising the heel portion 46b,
The position of the upper body 24 moves. Fig. 9 shows walking speed of 2 km /
The relationship between the angle of raising the heel portion 46b of the foot 22R (L) and the position of the upper body at a step length of 500 mm is shown. As a result of investigating from one in which the heel part is not raised to one in which the heel part is raised by 0.9 (rad), as the heel part is raised, the position of the upper body 24 in the both-leg supporting period tends to shift rearward, and its amount is It was found to be a little over 10 mm.

From these two findings, at the end of the both-leg supporting period, that is, at the time of leaving the hind legs, the upper body 24 moves from the midpoint of the front legs and the hind legs to the walking speed at that time for half of the both-leg supporting period. Even if it is approximated that it has reached the position where it has moved, it can be seen that the maximum error is within about 30 mm on the front side.

It is considered that this error can be corrected to some extent by parameters such as the size of the robot and walking speed. However, as described above, it is difficult to set the parameters by using a too complicated correction formula. As a result, it is often not useful for specifically designing the actuator.

From such a judgment, at the end of the two-leg supporting period, that is, at the time of leaving the hind legs, the upper body starts from the midpoint between the front legs and the hind legs and is half the time of the two-leg supporting period at the walking speed at that time. When it has reached the point where it has moved, it is practically appropriate to make a simple approximation.

6) Amount of raising the heel part The minimum amount of raising the heel part in order to reach the kicking foot with respect to the position of the center of gravity of the upper body which is estimated approximately in this way is as follows: The calculation was made as shown in 10.

Here, L: total thigh link length (L1) and crus link length (L2) (meaning a trajectory with the legs extended), ST: step length, PS: time ratio of both legs supporting period, L
T: the length from the projection point of the intersection of the ankle joints 18 and 20R (L) to the back of the foot to the toe 46b, LA: the length from the bottom of the sole of the foot to the intersection of the ankle joints 18 and 20R (L), LD: Distance traveled by the upper body during both legs, LH, LP, thA,
When thB and thC are shown in FIG. 10, and thHmin is the angle at which the heel portion 46b is raised to the minimum, the following is obtained.

LD = ST × PS LS = 0.5 × (ST-LD) LH = LA + √ (L × L-LS × LS) LP = ST-LS-LT thC = tan ⁻¹ (LH / LP) thA = Tan ⁻¹ (LA / LT) LB = √ (LT × LT + LA × LA) LX = √ (LP × LP + LH × LH) thB = cos ⁻¹ (LB × LB + LX × LX-L × L)
/2.0/LB/LX) thHmin = π-thA-thB-thC

In the above, the amount of raising of the heel portion 46b of the kicking foot required to reach the position (LS above) in which the center of gravity of the upper body 24 has moved by half the time during the two-leg supporting period at the walking speed at that time. With thHmin as the lower limit value, a gait is generated so as to have that posture.

To summarize the above, the control of the amount of tilt of the foot is performed by using the relational expression thH = f (PS, ST) or the relational expression thH = π-thA-thB-thC. You can

As in the robot according to the embodiment,
In the foot portion, the toe portion 46a is curved in an arc shape, and the radius of the arc is small and it is merely added to the tip portion, so that there is no problem even if that portion is neglected approximately as LT.

FIG. 11 shows the results of calculation of the above-mentioned steps and the ratios of the two-leg supporting period.

Considering the upper limit value thHmax for raising the heel portion 46b, from the viewpoint of increasing the gravity direction height of the center of gravity orbit of the upper body as much as possible, as shown in FIG. It can be said that there is no effect even if the heel portion 46b is raised beyond the posture in which the ankle joint 18R (L) and the hip joint 12R (L) are aligned. Therefore, thHmax = π-thA-thC.

Since the upper limit value and the lower limit value of the amount by which the heel portion 46b is raised are obtained, the design value of the amount by which the heel portion 46b is raised may be in principle between them. However, the lower limit value is the minimum amount required to set the center of gravity trajectory of the upper body to a desired height, and the height of the upper body can be increased to some extent by raising it to a higher level (FIG. 5). Therefore, if the driving capability of the actuator allows it, there is no big problem in raising it further.

In this embodiment, from the viewpoint of ensuring a margin from the lower limit value and not unnecessarily raising the heel portion, an average value which is an intermediate value between the two is a design value for raising the heel portion. And

7) Continuing the actuator design procedure After determining the amount of raising of the heel, the process proceeds to S18 and S20 to perform optimization by searching for a gait. That is, the gait is optimized so that the joint angular velocity and the displacement acceleration (in the x and y directions) of the body 24 are minimized in S20, and then the candidate of the optimal gait is determined in S22.

As a method of generating a mechanical gait prior to gait optimization, several methods have been used so far (eg, bipedal locomotion that compensates moments by body movements, Vol. 11, No. 3 of the Robotics Society). Robots ”) are open to the public and can be applied by applying them. In the literature of this example, a method of generating a gait from a free leg trajectory and a ZMP trajectory with the height of the center of gravity of the upper body kept constant is shown, but the upper body is raised so as not to bend the knee. There is no problem in setting a smooth trajectory. In the trajectory setting, the trajectory may be designated directly, or a constraint condition may be separately set and determined depending on the free leg trajectory or the ZMP trajectory.

The gait generated by the processing up to S22 still satisfies the conditions such as whether the landing impact force is within the allowable range in terms of the vertical acceleration (displacement acceleration in the z direction) of the upper body. Not necessarily. For this reason, S2 continues
Proceeding to step 4, the maximum value of the vertical acceleration of the upper body is set, and the vertical acceleration is suppressed within that range. This will be described below.

8) Relationship between walking speed and magnitude of vertical movement of upper body In the independent bipedal walking robot 1 shown in FIG.
The weight of 4 occupies a considerable part of the total weight of the robot, and since the magnitude of the vertical movement of the center of gravity of its upper body is directly related to the mechanical load on the legs and the impact force from the floor, In addition, the design of body weight center trajectory is an important item.

That is, the inertial force generated by the vertical movement is
Not only related to the output design of the leg link supporting the robot to the actuator, but also the strength of the leg link and the strength of the 6-axis force sensor 44 for directly measuring the floor reaction force, and the viewpoint of gait control. Is also related to the ground contact with the floor. From this point, it is desirable that the vertical movement is small, but if this is done, a walking form with a low body weight trajectory and low energy efficiency will result. Therefore, a certain amount of vertical movement is involved.

Also in this case, in order to keep the load on the legs below a certain level, it is necessary to limit the vertical movement. In principle, if the vertical acceleration does not exceed 1G, the robot should not bounce off the floor, but in reality, due to the unevenness of the floor, control errors, etc., even if the vertical acceleration is much smaller than that, the ground contact is poor. May be. In the illustrated robot structure,
Experimentally, it has been found that there is no practical problem if it is suppressed to about 0.2G.

However, keeping the vertical acceleration low at this level is a great limitation in the walking form. Therefore, except for low-speed walking where vertical acceleration is not a problem,
The optimum walking form is often the one that is used up to the maximum vertical acceleration limit, which is mainly determined by mechanical strength.
As described above, in order to realize a walking mode in which the vertical acceleration is fully used regardless of the walking speed, it is necessary to clarify the properties of the bipedal walking robot regarding the walking speed and the vertical acceleration.

In general, if the trajectory shape of the vertical movement accompanying walking is the same, the speed of the vertical movement increases in proportion to the walking speed, and the vertical acceleration increases in proportion to the square of the walking speed. For the sake of simplicity, assuming that the vertical trajectory is a sine wave with an amplitude A and a wavelength L, the height H of the upper body when moving at a walking speed V
Becomes H = Asin (2πV / Lt) (t: time).

At this time, the vertical acceleration DDH is differentiated twice with respect to t to obtain DDH = -A (2πV / L) ² sin (2πV / Lt). According to this equation, it can be seen that the vertical acceleration DDH is proportional to the square of the walking speed V.

Of course, the actual vertical trajectory is not a sine wave, but since it is a periodic function, Fourier series expansion is possible, and the acceleration of each term is also proportional to the square of the walking speed V. As is assumed, if the trajectory shape of the vertical movement is completely the same regardless of the walking speed, the magnitude of the vertical acceleration accompanying the vertical movement is obviously proportional to the square of the walking speed V.

However, the motion of the bipedal walking robot is greatly restricted by the equation of motion, and it is difficult to change the walking speed and still keep the vertical motion trajectory of the upper body completely the same. Therefore, strictly speaking, it cannot be said to be proportional to the square of the walking speed V.

In order to use this property in engineering, it is necessary to evaluate how much the error is. As a result of designing and analyzing a gait under many conditions, it was found that it can be qualitatively approximated to be proportional to the square of the walking speed V.

In the robot of FIG. 1, the thigh link 28R
(L) and lower leg link 30R (L) 400 mm, ankle joint height (LA) 140 mm, own weight 100 kg,
Walking speed 2km / h, 3km / h, 4km / h, 5km
/ H, the amplitude of vertical movement of the center of gravity of the upper body 24 is constant (22 m
13 to 16 show simulation data of the center-of-gravity trajectory and vertical acceleration of the body 24 when held at m).

It can be understood that, as the walking speed increases, the trajectory shape of the vertical movement changes in time axis depending on the walking speed, but the appearance is quite similar, and the vertical acceleration increases as expected. The relationship between the walking speed and the maximum vertical acceleration is shown by the solid line in FIG. In the figure, the broken line is 2 km / h.
The calculated value is shown in the case of being proportional to the square of the walking speed with reference to the acceleration at, but from the comparison with the broken line, it is not strictly proportional to the square of the walking speed. You can see that is true of.

As described above, in the biped robot capable of changing the walking speed, the requirement for the vertical acceleration of the upper body is
It depends on the strength of the legs and the strength of the sensor, and does not depend much on walking speed.

From this, in order to realize a walking mode in which the vertical acceleration is used up to the maximum limit, the vertical motion amplitude is kept constant regardless of the walking speed and the vertical motion amplitude is full. It will be good to decide.

That is, the actual vertical acceleration can be made substantially constant by reducing the amplitude of the vertical movement in inverse proportion to the square of the walking speed V. As a result, at all walking speeds, it is possible to realize a walking pattern that maximizes mechanical performance such as the leg strength and sensor strength of the robot.

If the above is applied more specifically, based on the above-mentioned method, the thigh link, the lower leg link 400 mm,
A biped robot with an ankle height of 140 mm and a weight of 100 kg, has a walking speed of 2 km / h and a vertical acceleration of 2000 m.
A gait for walking under a condition of m / sec ² or less can be obtained. The amplitude of the vertical movement at this time is 22 mm.

Based on this gait, the gait is designed by changing the amplitude of the vertical movement of the upper body so as to be inversely proportional to the square of the walking speed. The vertical acceleration of the entire robot with a walking speed of 3 km / h and an amplitude of vertical motion of 22 mm is set to 9.7 mm, which is (2/3) ² = 4/9 times the gait of walking speed of 2 km / h. Shown in 18. Similarly, FIG. 19 shows the vertical acceleration of the entire robot when the amplitude of vertical movement at a walking speed of 5 km / h is 3.5 mm.

At 3 km / h, the vertical acceleration is suppressed as expected, but at 5 km / h, the vertical acceleration is slightly increased although it is suppressed. In order to see this point in detail, the vertical acceleration of only the upper body 24 is shown in FIG. Looking at the vertical acceleration of only the upper body, it can be seen that the vertical acceleration is well suppressed. That is, at a walking speed of about 5 km / h, it means that the vertical acceleration due to the inertial force (Z-axis component) of the leg link (free leg) that could be ignored during low-speed walking cannot be ignored.

However, of course, it does not matter if the vertical movement of the upper body 24 is reduced. Therefore, when high-speed walking is required, such a factor of increase in vertical acceleration should be taken into consideration. It will be necessary to use measures such as giving a margin to the design strength beforehand. In this way, by making the amplitude of the vertical movement of the upper body inversely proportional to the square of the walking speed, it is possible to suppress the main component of the impact force from the floor surface due to the vertical acceleration of the entire robot within a wide walking speed range. It is possible to reduce the extreme impact force from the floor.

More specifically, in setting the maximum vertical acceleration of the upper body, it is necessary to consider the mechanical strength of the robot, the rigidity of the walking road surface, and the ground contact property with the road surface. Depending on the conditions, experimentally ± 2000 mm / sec ²
As far as the degree is concerned, the ground contact with the road surface does not matter.

Returning to the explanation of the flow chart of FIG. 4, S
26, S28, S20, S22, and S24 are looped, and the amplitude of the vertical motion of the body trajectory is reduced as described above, and the gait is generated again for the trajectory to optimize. If the vertical movement as a result of this is approximately the same as the set value, that is the optimal gait that allows walking realistically (S30).

In this way, the optimum gait is determined, then the process proceeds to S32, the angular velocity of each joint is directly determined as the motion of the joint in the gait, and the process proceeds to S34 to determine the joint torque T. From the position vector Z of the ZMP at the time point, the position vector P of the joint actuator, and the force vector F from the floor surface determined by the motion of the center of gravity, T = (P−Z) × F is determined.

Since the specifications of the maximum angular velocity and the maximum torque required for the joint actuator have been determined in this way, the process proceeds to S36, and the specifications of the joint actuator satisfying the maximum values are determined.

[0116]

Embodiments of the present invention will be described below.

The parameters of the robot 1 shown in FIG. 1 are as shown in FIG. Also, the weight of 100 kg, the ratio of both legs supporting period in the time to take a step (PS mentioned above)
Is 20%. This robot has a moving speed of 2 km / h,
The method of determining the amount of raising the heel portion 46b when walking with a stride of 500 mm and the method of determining actuator specifications based on the optimal gait based on the determination will be described more specifically.

These data are described above in the heel part 46.
Applying to the relational expression for obtaining the amount to increase b, the procedure as shown in S100 to S110 of the flowchart in FIG.
Angle upper limit thHmax (56.29 degrees) and lower limit t
hHmin (9.64 degrees) is obtained, and 32.96 degrees as the average value thH thereof is set as the design value of the amount by which the heel portion 46b is raised.

Next, a free leg trajectory is generated under the condition that the angle of raising the heel portion 46b at the time of leaving the bed is 32.96 degrees.
This trajectory should be such that it smoothly steps forward during the walking cycle. At this time, in order to ensure robustness against unevenness of the road surface, a trajectory is selected such that the foot 22R (L) of the free leg is raised to some extent, for example, 40 mm. The ZMP trajectory is also determined in consideration of mechanical conditions.

Specifically, in order to secure the stability of walking, basically, the ZMP is set to be near the center of the foot 22R (L), and the front foot is smoothed in the both foot support period. Set to move to. Before and after the transition between the two-leg support period and the one-leg support period, a transition period is provided so that the ZMP foot moves smoothly from the end point to the center. The upper body center trajectory is determined to be kept as high as possible depending on the restraint conditions.

Under this condition, a gait with a joint angular velocity that is as small as possible in actuator design is first obtained by searching for a free leg trajectory. The gait thus obtained does not yet satisfy the condition of vertical acceleration, which is a condition related to the impact force of the floor surface and the like. Here, the condition of vertical acceleration is ± 2000 mm / sec ² . The gait, the vertical motion trajectory of the upper body, and the vertical acceleration are shown in FIGS. 23, 24, and 25.

As can be seen from FIG. 25, since the condition of vertical acceleration is exceeded, the amplitude of vertical movement of the upper body that exerts a dominant influence on the vertical acceleration on this gait is reduced. Correct the body trajectory and generate the gait again. The results are shown in FIGS. 26, 27 and 28. FIG.
According to this, it is understood that this gait satisfies the condition of vertical acceleration.

In this way, the gait considering the vertical acceleration can be obtained.

The angular velocity of each joint can be directly obtained from this gait. The angular velocity of the ankle joint 18R (L) in this gait is shown in FIG. 29, the knee joint 16R (L) in FIG. 30, and the hip joint 12R (L) in FIG. Next, the joint torque is obtained by the simple calculation described above.

FIG. 32 shows the torque of the ankle joint 18R (L) and FIG. 33 shows the torque of the knee joint 16R (L) in this gait.
FIG. 34 shows the torque of the hip joint 12R (L). However, the torque for driving the free leg is not included in the calculation. still,
Since the calculated value is standardized with the robot's own weight being 1 kg, 100 times the torque is required when the own weight is 100 kg.

From these results, the maximum angular velocity and maximum torque required for the actuator are determined by the ankle joint 18R.
(L) 24 rpm, 15 kgm, knee joint 16R
(L) 32 rpm, 17 kgm, hip joint 12R
It can be seen that (L) requires 32 rpm and 11 kgm.

Next, a case where the walking of the illustrated robot is controlled based on the gait generated as described above will be briefly described with reference to the flow chart of FIG.

First, in S200, each part of the apparatus is initialized, and the process proceeds to S202, in which gait parameters are loaded from the ROM 84 described above. The gait of the robot generated as described above is stored in the ROM 84 as a gait parameter.
Is stored. Then, the process proceeds to S204 to wait for the start signal, and when the start signal is generated, the process proceeds to S206 to increment the counter.

Then, the process proceeds to S208 to calculate the posture parameter. This is because some of the above-mentioned gait parameters require an interpolation calculation, so that it is obtained here, and S
The posture parameter at the time designated by the counter 206 is calculated. Next, in S210, the angles of twelve joints are calculated, and in S212, the synchronization signal is waited for. When the synchronization is achieved, the process proceeds to S214 to output the calculated joint angle, and the process proceeds to S216 to end walking. Unless it is determined that the same operation is repeated, when it is determined to end, the process proceeds to S218, resets the counter, and ends.

Further, based on the joint angle output in S214, the second arithmetic unit 82 described above performs the servo control in parallel according to the flow chart shown in FIG. 36 so that the joint angle becomes the joint angle. Since the work is publicly known, a description thereof will be omitted.

Since this embodiment is configured as described above, it is possible to determine in advance an appropriate amount of raising the heel portion of the kicking foot that can realize a gait with a high center of gravity trajectory, that is, a high energy efficiency. The gait candidates can be narrowed down accordingly, from among the gaits that satisfy only the typical walkable condition, and the gait design can be greatly simplified.

Further, the gait design can be further simplified by obtaining the standard of the vertical motion acceleration related to the impact force from the floor surface from the amplitude of the vertical motion of the center of gravity.

FIGS. 37 to 39 show a second embodiment of the present invention.
It is explanatory drawing which shows another example of the foot part of the biped walking robot which the gait generation method etc. of a legged walking robot presuppose.

As shown in the figure, the foot portion is composed of a member 220 having a substantially C-shaped plane and a member 222 having a substantially rectangular shape in a plane.
The character member 220 and the rectangular member 222 are connected via a rod 224. That is, the rod 224 is the C-shaped member 2
20 and the rectangular member 222 are inserted into the holes formed,
The C-shaped member 220 is fixed by a pin 226 to form a toe joint.

On the other hand, the inner diameter of the hole formed in the rectangular member 222 is formed to be relatively larger than the outer diameter of the rod 224, and the rectangular member 222 is rotatable about the rod 224. A large-diameter annular hole 228 is formed near the central position of the rectangular member 222, and the six-axis force sensor 44 is formed therein.
(Not shown) is installed, and the ankle joints 18 and 20 are provided on the upper part thereof.
R (L) is connected.

Then, the rod 224 and the rectangular member 222
A spring 229 is mounted between and to urge the rectangular member 222 upward (in the direction of gravity). Therefore, the ankle joint 18R
When (L) rotates via the axis 36, the rectangular member follows it and ascends from the road surface. In this state, the sole of the foot is
Only the C-shaped member 220 contacts the road surface.

Although the foot structure shown in FIG. 2 has been described above, even in the case of the foot structure shown in FIG. 37 and the like, it may be difficult to accurately determine the length LT to the toe. Conceivable. However, even in that case, in principle, the heel part 4
There is no problem if the length equivalent to the fulcrum is used when raising 6b.

In the example shown in FIG. 37 and the like, since the rotation axis of the toe joint is slightly above the floor surface of the mechanism, the distance to this axis is not accurate as LT, but it is approximately sufficient. Except for extreme cases, the distance to the rotation axis excluding the toe portion may be LT and there is no problem.

Regarding the foot shape for which such an approximation does not hold, there are various possibilities of the shape, so it is not possible to express a concrete method in all cases, but the principle described in the beginning is used. Based on the calculation of the amount of the heel part to be raised when leaving the bed during the two-leg supporting period based on the convergence calculation or the like, the appropriate amount of raising the heel part can be calculated in the same manner.

Although the center of gravity of the robot is approximated to that of the upper body in the above, it is needless to say that the center of gravity of the entire robot may be obtained.

In the above, only the example of the bipedal legged mobile robot is shown, but the above configuration is also applicable in principle to a legged mobile robot having three or more legs.

[0142]

[Effects of the Invention] Since it is possible to predetermine an appropriate amount of raising of the heel part of a kicking foot that can realize a gait having a high center of gravity trajectory, that is, a high energy efficiency, gait candidates can be narrowed down accordingly to design gait. It can be greatly simplified.

[Brief description of drawings]

FIG. 1 is an overall skeleton diagram showing a bipedal walking robot, which is premised on a gait generating method for a legged walking robot according to the present invention.

FIG. 2 is an explanatory sectional view showing in detail a structure of a foot of the bipedal walking robot shown in FIG.

3 is a block diagram showing details of a control unit of the bipedal walking robot shown in FIG. 1. FIG.

FIG. 4 is a flow chart showing a gait generating method and the like according to the present invention on the premise of the bipedal walking robot shown in FIG.

FIG. 5 is a diagram showing a condition that the robot shown in FIG.
It is a data figure which shows the relationship of how the height of the center of gravity trajectory of an upper body changes.

6 is a data diagram showing the amount of movement of the upper body of the robot shown in FIG.

7 is a data diagram showing the position of the upper body of the robot shown in FIG.

8 is a center of weight of the foot of the robot shown in FIG. 1 (ZM
It is a data figure which shows the change of the position of the upper body when P) is moved back and forth from the geometric center of a foot.

9 is a data diagram showing the relationship between the position of the upper body and the amount of lifting of the heel of the robot shown in FIG.

10 is an explanatory diagram showing calculation of a lower limit value of a raising amount of a heel portion of the robot shown in FIG.

FIG. 11 is a data diagram showing a relationship between a step length and a heel amount of the robot shown in FIG.

12 is an explanatory diagram showing calculation of an upper limit value of a lift amount of a heel portion of the robot shown in FIG.

13 is a walking speed of 2 km for the robot shown in FIG.
It is a data figure which shows the gravity center trajectory of a body and the relationship of a vertical acceleration when the amplitude of the up-and-down movement of the body of / h is kept constant.

14 is a walking speed of 3 km for the robot shown in FIG.
It is a data figure which shows the gravity center trajectory of a body and the relationship of a vertical acceleration when the amplitude of the up-and-down movement of the body of / h is kept constant.

15 is a walking speed of 4 km for the robot shown in FIG.
It is a data figure which shows the gravity center trajectory of a body and the relationship of a vertical acceleration when the amplitude of the up-and-down movement of the body of / h is kept constant.

16 is a walking speed of 5 km for the robot shown in FIG.
It is a data figure which shows the gravity center trajectory of a body and the relationship of a vertical acceleration when the amplitude of the up-and-down movement of the body of / h is kept constant.

FIG. 17 is a data diagram showing the relationship between the walking speed and the vertical acceleration of the robot shown in FIG.

18 is a walking speed of 3 km for the robot shown in FIG.
It is a data figure which shows the up-and-down acceleration when changing the amplitude of the up-and-down motion at / h in inverse proportion to the square of walking speed.

FIG. 19 is a data diagram similar to FIG. 18, showing a walking speed of 5 km /
It is a data figure which shows the vertical acceleration at the time of h.

FIG. 20 is a data diagram similar to FIG. 18, showing a walking speed of 5 km /
It is a data figure which shows the acceleration only of the upper body at the time of h.

FIG. 21 is an explanatory diagram of parameters of the robot of FIG. 1 used in the embodiment of the present invention.

FIG. 22 is a flow chart showing calculation of the raising angle of the heel portion in the embodiment of the present invention.

FIG. 23 is a simulation data diagram showing a gait calculated by the flow chart of FIG. 22.

24 is a data diagram showing a locus of vertical movement of the upper body of the gait shown in FIG. 23.

25 is a data diagram showing vertical acceleration of the upper body of the gait shown in FIG. 23. FIG.

FIG. 26 is a simulation data diagram showing an example of correcting the gait shown in FIG. 23.

27 is a data diagram showing a locus of vertical movement of the upper body of the gait shown in FIG. 26. FIG.

28 is a simulation data diagram showing vertical acceleration of the upper body of the gait shown in FIG. 26. FIG.

FIG. 29 is a simulation data diagram showing the angular velocity of the ankle joint obtained from the gait of FIG. 26.

30 is a simulation data diagram showing the angular velocity of the knee joint obtained from the gait of FIG.

FIG. 31 is a simulation data diagram showing the angular velocity of the hip joint obtained from the gait of FIG. 26.

32 is a simulation data diagram showing the joint torque of the ankle joint obtained from the gait of FIG. 26. FIG.

FIG. 33 is a simulation data diagram showing the joint torque of the knee joint obtained from the gait of FIG. 26.

FIG. 34 is a simulation data diagram showing the joint torque of the hip joint obtained from the gait of FIG. 26.

FIG. 35 is a flow chart showing an example of controlling the walking of the robot of FIG. 1 based on a gait generated by the gait generating method according to this embodiment.

36 is a flow chart showing an example of performing servo control of a joint angle in the control of FIG. 35.

FIG. 37 is an explanatory cross-sectional view of a foot showing another example of a bipedal walking robot which is premised on the gait generating method and the like according to the present invention.

38 is a sectional view taken along line XXXVIII-XXXVIII of FIG. 37. FIG.

39 is an explanatory side view showing a case where the heel portion of the foot shown in FIG. 37 is raised.

[Explanation of symbols]

 1 Robot 2 Leg links 10, 12, 14R, L Hip joint 16R, L Knee joint 18, 20R, L Ankle joint 22R, L Foot 46b Heel part

Claims (8)

[Claims] 1. At least an upper body, and two leg links connected to the upper body via hip joints, wherein the two leg links are respectively thigh links, and the thigh links are knee joints. A leg for bipedal walking consisting of a lower leg link connected via a leg and a foot connected to the lower leg via an ankle joint, and the two leg links alternately move and kick the floor. In a gait generating method for a walking robot, the position of the upper body at the time of kicking is estimated, and the amount of lifting of the heel of the foot is determined so that the upper body reaches at least that position. A gait generation method for a legged walking robot, characterized in that the gait is generated based on the gait. 2. It is estimated that the upper body is moving forward a predetermined distance from the midpoint between the front foot and the rear foot at the time when the kick is finished, and the upper body reaches at least the estimated position. 2. The gait generating method for a legged walking robot according to claim 1, wherein the gait generating method is configured to determine a raised amount of a flat heel and generate a gait based on the determined amount. 3. The raising of the heel portion, wherein the raising amount of the heel portion of the foot is set to a lower limit value of the raising amount, and the hip joint, the ankle joint, and the tip portion of the foot are located on substantially the same line. 3. The legged walking robot according to claim 2, wherein the amount is set as an upper limit value of the raising amount, and the raising amount of the heel portion of the foot is determined within the range between the lower limit value and the upper limit value. Gait generation method. 4. The gait generating method for a legged walking robot according to claim 3, wherein the raising amount of the heel portion of the foot is determined from the average value of the lower limit value and the upper limit value. 5. The foot has a bottom surface formed in an arc shape near the tip thereof, and / or the foot includes a toe joint, and the arc bottom surface is grounded, and / or
Alternatively, the heel portion of the foot is raised via the toe joint.
A gait generating method for a legged walking robot according to any one of items. 6. An upper body and at least two parallel leg links connected to the upper body via a hip joint,
In a legged walking robot having a foot which is connected to an end of the leg link via an ankle joint and can tilt in the direction of gravity, the gait of the robot is basically divided into a one-leg supporting period and a two-leg supporting period. It is assumed that walking consists of continuously and alternately maintaining these two support periods at a predetermined pitch, and the transition from the two-foot support period to the one-foot support period rotates one ankle joint. A walking control device for a legged walking robot, which is characterized by performing tilting of the foot. 7. The control of the amount of tilt of the foot is performed by a relational expression of thH = f (PS, ST) (where, thH: amount of tilt (angle), PS: time ratio of both feet supporting period, ST: stride). 7. The walking control device for a legged walking robot according to claim 6, wherein the walking control device is configured to be performed by using. 8. The control of the tilting amount of the foot is performed by a relational expression of thH = π-thA-thB-thC (where thH: tilting amount (angle), thA,
B, C: The walking control device for a legged walking robot according to claim 6, wherein the walking control device is configured to perform using a predetermined tilt amount (angle) determined from the posture of the robot.