Control method for legged walking

Abstract

To provide a control method of leg walking that allows walking to be efficiently controlled.SOLUTION: A leg walking device includes a spring mechanism, a first driving source for changing a leg length, and a second driving source for rotating a leg. A control method of leg walking includes: a first step in which a support leg makes the first driving source for changing a support leg length immobile during single leg support, and a second step in which a ground-contact impact after the single leg support and immediately before both leg support is absorbed using the spring mechanism, in the leg walking device.SELECTED DRAWING: Figure 12

Description

[Technical Field]

[0001] The present invention relates to a method for controlling legged locomotion. [Background technology]

[0002] In recent years, research into bipedal robots has been progressing. For a robot to be able to walk like a human, it needs to have sufficient functionality, especially in terms of fault tolerance and efficiency. In bipedal robots, adding flexibility to the robot's structure improves its fault tolerance, and robots capable of torque control of the joints are also being developed.

[0003] For example, one method for controlling the walking of a legged robot with elasticity is to realize walking by decomposing the walking motion into three elements (body posture control, kick-off control, and ground contact position control) that focus on the macroscopic motion of walking. For example, in Non-Patent Document 1, as shown in Figure 20, the elemental decomposition of walking is decomposed into macroscopic behavior that focuses on human walking motion. The elements are three active components: kick, swing leg / stride, and state posture maintenance. This method is a running control method proposed by Raibert et al. Figure 20 is a diagram for explaining the running control method proposed by Raibert et al. Hodgins also extended this method to walking (see Non-Patent Document 2). [Prior art documents] [Non-patent literature]

[0004] [Non-Patent Document 1] Marc H. Raibert, “Legged Robots That Balance (Artificial Intelligence)”, MIT Press, 1986 [Non-patent document 2] JK Hodgins, “Biped gait transitions,” in Proc. IEEE Int. Conf. Robot. Autom, 1991, pp. 2092-2097. Summary of the Invention [Problem to be solved by the invention]

[0005] However, although the technology described in Non-Patent Document 1 breaks down the walking elements, no analysis has been conducted into which specific movements during walking achieve "how efficient movements" and therefore there has been little progress in understanding.

[0006] The present invention has been made in consideration of the above problems, and an object of the present invention is to provide an efficient method for controlling leg walking. [Means for solving the problem]

[0007] (1) In order to achieve the above object, one embodiment of the present invention provides a method for controlling leg walking in a leg walking device that includes a spring mechanism, a first drive source that changes the leg length, and a second drive source that rotates the leg, the method comprising: a first step of immobilizing the first drive source that changes the supporting leg length during single leg support; and a second step of absorbing the ground contact impact after the single leg support and immediately before double leg support using the spring mechanism.

[0008] (2) In addition, in the method for controlling leg walking according to one aspect of the present invention, in the first step, the support movement may be left to passive movement, and in the second step, the support movement may be made to work so as to regenerate in the later stage of both legs supporting.

[0009] (3) Furthermore, in a method for controlling leg walking according to one aspect of the present invention, the contact impact of the front leg immediately before the double leg support may be stored in the spring mechanism at the beginning of the double leg support, and the energy stored in the spring mechanism may be retained during the single leg support and then released by the rear leg at the next double leg support.

[0010] (4) Furthermore, in a method for controlling leg walking according to one aspect of the present invention, leg walking may be controlled by controlling the spring mechanism, a first actuator module including the first drive source, and a second actuator module including the spring mechanism and the second drive source.

[0011] (5) Furthermore, in a method for controlling leg walking according to one aspect of the present invention, the spring mechanism may include an outer shaft, an inner shaft, and a plurality of urethane rubber springs evenly arranged concentrically between the outer shaft and the inner shaft, the outer shaft being connected to a drive gear as an input, and the inner shaft being connected to a predetermined link as an output. [Effects of the Invention]

[0012] According to (1) to (5), by utilizing the passive dynamic characteristics of the body, it is possible to realize efficient supporting leg exercise without performing work. [Brief explanation of the drawings]

[0013] [Figure 1] 1 is a diagram for explaining the difference between the control method of the embodiment and the conventional technique. FIG. [Figure 2] FIG. 1 is a diagram for explaining element decomposition and the advantages of decomposition. [Figure 3] 10A and 10B are diagrams illustrating an example of the structure of a right leg according to an embodiment. [Figure 4] FIG. 4 is a diagram showing an example of an actuator system in the case of the mechanism example of FIG. 3. [Figure 5] 1A and 1B are diagrams illustrating an example of the configuration of a spring module. [Figure 6] 10A and 10B are diagrams illustrating an example of the relationship between spring torque and displacement angle. [Figure 7] 10A and 10B are diagrams for explaining the disadvantages of not having a spring in the leg portion. [Figure 8] 10A and 10B are diagrams for explaining the advantages of providing the legs with springs. [Figure 9] FIG. 1 is a diagram for explaining decomposed elements in walking control. [Figure 10] This shows the coordinates of the motion space and task space for the example mechanism in Figure 3. [Figure 11] FIG. 10 is a diagram for explaining definitions of names of legs. [Figure 12] 10A and 10B are diagrams for explaining an example of a method for controlling leg walking according to an embodiment. [Figure 13] FIG. 4 is a diagram showing the external shape of the robot in the mechanical example of FIG. 3. [Figure 14] FIG. 10 is a diagram for explaining confirmation items. [Figure 15] FIG. 10 shows the results of confirming that the front legs are in the DS phase and the supporting legs are in the SS phase. [Figure 16] This figure shows the results of confirming that the hind legs are in the DS phase and the swing legs are in the SS phase. [Figure 17] FIG. 10 is a diagram showing the displacement of leg length angular velocity. [Figure 18] FIG. 10 is a diagram showing power in each task space during walking. [Figure 19] FIG. 10 is a diagram showing power in each task space during walking. [Figure 20] FIG. 1 is a diagram for explaining the cruise control method proposed by Raibert et al. DETAILED DESCRIPTION OF THE INVENTION

[0014] Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the drawings used in the following description, the scale of each component is appropriately changed so that each component can be recognized. In all the drawings for explaining the embodiments, the same reference numerals are used for components having the same functions, and repeated explanations will be omitted. Furthermore, in this application, "based on XX" means "based on at least XX," and includes cases where it is based on other elements in addition to XX. Furthermore, "based on XX" is not limited to cases where XX is used directly, but also includes cases where it is based on XX that has been calculated or processed. "XX" is any element (for example, any information).

[0015] <Summary> First, the difference between the control method of this embodiment and the prior art will be described. FIG. 1 is a diagram illustrating the difference between the control method of this embodiment and the prior art. The diagram of the area indicated by the symbol g10 is an image of control using a linear inverted pendulum (LIP) and a zero moment point (ZMP) trajectory according to the prior art. LIP stands for linear inverted pendulum motion. The ZMP is a ZMP standard that, for stable walking, the ZMP point on the floor where the moment due to the floor reaction force is zero must exist within the range of the foot of the supporting leg. The diagram indicated by the symbol g11 is an image of the configuration of the robot's legs, and the diagram indicated by the symbol g12 is an image of the application of force (or torque) (double arrow g31) and velocity (or angular velocity) (double arrow g32) during walking. In the diagram indicated by the symbol g11, the legs are connected by elastic actuators g13. The dotted symbol g33 in the white circle represents a joint. The dotted symbol g34 in the white circle represents the joint between the leg and the torso. In conventional control methods, as shown in the diagram with reference symbol g12, force or torque and velocity act on the contact point with the floor, the joints of the legs, and the joints between the body and the legs. In conventional control methods, the trajectory of the robot is controlled, and the motion is determined by the trajectory.

[0016] The diagram of the area indicated by reference symbol g20 is an image diagram of control utilizing passive characteristics according to this embodiment. The diagram indicated by reference symbol g21 is an image diagram of the configuration of the robot's legs, and the diagram indicated by reference symbol g22 is an image diagram of how force (or torque) (double arrow g31) and velocity (or angular velocity) (double arrow g32) are applied when walking. In the control method of this embodiment, as shown in the diagram indicated by reference symbol 2, velocity (or angular velocity) acts at the contact point with the floor, force (or torque) acts on the joints of the legs, and force (or torque) and velocity (or angular velocity) act on the joints between the torso and the legs. In the control method of this embodiment, the dynamics of the robot are controlled, and as a result, motion is generated.

[0017] As shown in the diagrams g11 and g21, in conventional methods, force (or torque) and velocity (or angular velocity) act on the contact surface with the floor, the joints of the legs, and the joints between the trunk and legs, whereas in the control method of this embodiment, only one of force (or torque) or velocity (or angular velocity) acts on the contact surface with the floor and the joints of the legs. In this way, in the control method of this embodiment, for example, the contact surface with the floor and the joints of the legs do not perform work partially (= no force or torque acts, or no velocity acts), and therefore control can be performed more efficiently than in conventional methods.

[0018] <Functional decomposition> First, we will explain the decomposition of elements and the benefits of decomposition. Figure 2 is a diagram to explain the decomposition of elements and the benefits of decomposition. Diagrams g110, g130, and g150 show examples of transition states during walking. Diagrams g110, g130, and g150 show examples of transition states during walking. As shown in diagrams g110, g130, and g150, the control functions are decomposed into the following six ((a) to (f)).

[0019] The diagram with reference numeral g110 is a diagram showing an example of elements for supporting with both legs. The diagram with reference numeral g111 is an image diagram of the movement of the legs and torso. The diagram with reference numeral g112 is a diagram showing the elements acting on each part. As shown in the diagram with reference numeral g112, when supporting with both legs, the elements acting on each part can be broken down into the following four: (a) A force or torque acts on the front leg joint. (b) Force or torque and velocity are generated at the hind leg joints. (c) Velocity is generated at the contact points of the front and back legs with the floor. (e) Forces or torques and velocities are generated at the joints of the torso and legs. In the case of double leg support, energy is dissipated by the front leg touching the ground and energy is applied by the rear leg kicking.

[0020] The diagram with reference numeral g130 is a diagram showing an example of the movement and elements of the supporting leg during the single-leg support phase. The diagram with reference numeral g131 is an image diagram of the movement of the leg and torso. The diagram with reference numeral g132 is a diagram showing the elements acting on each part. As shown in the diagram with reference numeral g132, in the case of a supporting leg in single-leg support, the elements acting on each part can be broken down into the following three: (c) Velocity is generated at the contact points of the front and back legs with the floor. (d) Forces or torques are generated at the joints of the legs. (e) Forces or torques and velocities are generated at the joints of the torso and legs. In the case of a single support leg, work can be reduced to zero by leaving it to passive motion.

[0021] The diagram with reference numeral g150 is a diagram showing an example of the movement and elements of the swing leg during the single support phase. Reference numeral g151 is an image diagram of the movement of the leg and torso. The diagram with reference numeral g152 is a diagram showing the elements acting on each part. As shown in the diagram with reference numeral g152, in the case of the swing leg during the single support phase, the elements acting on each part can be broken down into the following three: (f) Force or torque and velocity are generated in the swing leg. In the case of the swing leg during the single-leg support phase, energy is consumed by the swing leg, but the energy consumed by the lighter leg is small.

[0022] As shown in Figure 2, in the movements (a), (c), and (d), only force (or torque) or velocity is generated, allowing for efficient movement control without passive work.

[0023] <Example of leg structure> Next, an example of the structure of the leg portion of this embodiment will be described. FIG. 3 is a diagram showing an example of the structure of the right leg according to this embodiment. The diagram with reference numeral g200 is a diagram showing an example of a modeled structure. The diagram with reference numeral g210 is a diagram showing an example of the structure as viewed from the side. The diagram with reference numeral g220 is a diagram showing an example of the configuration as viewed from the front. Note that the configuration shown in FIG. 3 is an example and is not limited to this.

[0024] As shown in the diagrams of reference numerals g200 and g210, the robot 1 includes at least a body 11 and legs 12. The legs 12 are connected to links 1, which are, for example, lever links. lever , Thigh Link thigh , the link l which is the rod Link rod , the link l which is the knee link knee , Shin Link is the link lshin , and Foot. knee and shin Angle θ with knee is a fixed angle of the rigid link having a triangular shape. Each part is connected by a free connection (symbol represented by a dot in a white solid circle). The body 11 and the leg 12 are connected by an elastic actuator (symbol represented by a dot in a white chain circle). The body 11 and the link l lever Between them is the spring constant k roc The fuselage 11 and the link 1 thigh Between them is the spring constant k thigh It has.

[0025] As shown in the diagram of reference symbol g220, the robot 1 includes, for example, a force sensor or torque sensor 101, actuator modules 102 (102-1, 102-2), actuator modules 103 (103-1, 103-2), a motor control unit 104, an IMU 105, and a control unit 106. The IMU 105 is an inertial measurement unit that detects inertial momentum using, for example, a three-axis gyro sensor that detects changes in rotation and direction, and a three-axis acceleration sensor that detects changes in axial velocity. The motors included in the actuator modules 102 (102-1, 102-2) correspond to, for example, a first drive source that changes the leg length. The motors included in the actuator modules 103 (103-1, 103-2) correspond to a second drive source that rotates the legs.

[0026] FIG. 4 is a schematic diagram of an actuator system according to this embodiment. FIG. 4 shows the link structure between the Lever module provided at the hip joints of both legs and the Thigh module. The Lever and Thigh links are both connected to the Shin link in opposite directions. As shown in FIG. 4, each module includes, for example, a motor (including an encoder) 201, a first drive gear 202, a second drive gear 203, a spring 204 (spring module), and a Lever link (including an encoder) 205. The Lever module and Thigh module are connected by a Shin link 206. The Thigh module includes, for example, a Thigh link 207, a spring 208 (spring module), a second drive gear 209, a first drive gear 210, and a motor (including an encoder) 211. Thus, in this embodiment, the actuator modules 102 (102-1, 102-2) and the actuator modules 103 (103-1, 103-2) are provided with springs (spring mechanisms) 204, 208. Note that the configuration shown in Fig. 4 is an example and is not limited to this.

[0027] The configuration of the legs and the like in this embodiment uses the design method described in Reference 1. In addition, each actuator module 102 of the robot uses an infinitely rotatable Neidhart rubber spring as an elastic body to provide fault tolerance and reduce weight. In addition, the controller proposed in Non-Patent Document 1 is applied to the control unit 106.

[0028] Reference 1; H. Shin, T. Ishikawa, T. Kamioka, K. Hosoda, and T. Yoshiike, “Mechanistic Properties of Five-bar Parallel Mechanism for Leg Structure Based on Spring Loaded Inverted Pendulum,” in Proc. IEEE-RAS Int. Conf. Humanoid Robots, 2019, pp. 320-327.

[0029] In Reference 1, a design method for a parallel mechanism for constructing a virtual linear spring in the leg from the foot to the hip joint was established. In the following explanation, this mechanism will be referred to as the SLIP mechanism. In Reference 1, the condition for realizing a virtual linear spring from the foot to the hip joint is the following equation (1).

[0030]

number

[0031] In equation (1), k i and i (i is one of thigh, lever, knee, rod, or shin) is the spring constant of each joint and the link length of each link. Here, we assume that actuation joints, consisting of series-elastic actuators (SEAs) (see Reference 2), are arranged coaxially.

[0032] Reference 2; G. Pratt and M. Williamson, “Series elastic actuators”, Proc. IEEE / RSJ Int. Conf. Intel. Robots Syst., vol. 1, pp. 399-406, 1995.

[0033] Here, the length of the link is l knee =l lever and rod =l thigh =l shin , and angle θ knee There are no restrictions on. Therefore, this condition indicates that it is possible to use this condition to design legs of various shapes.

[0034] The robot 1 shown in the diagram of reference symbol g220 in Figure 3 is designed based on the conditions of formula (1), and its dimensions are, for example, 0.7 m in height, 0.53 m in width, and 0.12 m in depth. The length of the link is, for example, l lever =l knee =0.075m, l rod =l thigh =l knee = 0.25m. The total weight is, for example, 10 kg, the weight of each leg is, for example, 1 kg, and the weight ratio is, for example, 1:10.

[0035] <Spring> Next, we will explain design examples of leg springs and the benefits of including springs. The advantages of applying compliance as an energy storage to bipedal robots include I. large displacement and high rigidity, II. lightweight and compact design, and III. fault tolerance.

[0036] First, an example of a spring design will be described. The springs used in the legs must have high rigidity to support the body weight. They also need to reduce the impact force to protect the robot 1 body when the legs land. Since the impact force can be two to three times the force of gravity due to the robot 1's own weight, it is desirable for the springs to have a large displacement.

[0037] Based on Reference 1, the spring stiffness requirement is k thigh =k lever = 150 Nm / rad, and the displacement angle was set to 15°. Figure 5 is a diagram showing an example of the configuration of a spring module (spring mechanism). The diagram with reference numeral g251 is a diagram showing an example of the structure before the spring is deformed, and the diagram with reference numeral g252 is a diagram showing the state after the spring is deformed.

[0038] As shown in the diagram of reference symbol g251, the spring module 250 is composed of, for example, an outer shaft 252, an inner shaft 253, and eight urethane rubber springs 251. In the spring module, the outer shaft 252 is connected to a drive gear as an input, and the inner shaft 253 is connected to a lever link (a predetermined link) as an output. The eight rubber springs are equally spaced concentrically at 45° between the outer shaft 252 and the inner shaft 253. As shown in the diagram of reference symbol g252, when the inner shaft 253 rotates, the rubber spring 251 is compressed, generating a restoring torque. The spring has an outer diameter of, for example, 0.067 m, a thickness of 0.025 m, a weight of 0.15 kg, and a maximum torque of 50 Nm.

[0039] Figure 6 shows an example of the relationship between spring torque and displacement angle. The horizontal axis is angle (deg), and the vertical axis is torque (Nm). Line g271 is the theoretical value, dashed line g272 is the first test result, solid line g273 is the second test result, and dashed line g274 is the third test result.

[0040] Figure 7 is a diagram to explain the disadvantages of not having springs in the legs. The diagram with reference numeral g421 is a model of walking when the legs are not equipped with springs. The diagram with reference numeral g422 is an image diagram showing the state of energy when walking. In Figure 7, the unhatched double arrow g425 represents force or torque, and the hatched double arrow g426 represents velocity. As shown in Figure 7, if the legs do not have springs, for example, a force or torque is generated at the joint of the front legs, and a force or torque and velocity are generated at the joint of the rear legs. As a result, if the legs do not have springs, energy is first dissipated through inelastic collisions, and then the dissipated energy is applied. If the legs do not have springs, both the energy dissipated and the energy applied are large.

[0041] Fig. 8 is a diagram for explaining the advantages of providing the legs with springs. In Fig. 8, the unhatched double arrow g425 represents force or torque, and the hatched double arrow g426 represents velocity. As shown in Figure 8, when the legs are equipped with springs, for example, a force or torque is generated at the joint of the front leg, and a force or torque is generated at the joint of the back leg. At this time, no velocity is generated. As a result, in the case of a perfectly elastic collision, there is no dissipation or addition of energy. Also, in the case of an elastic collision, the dissipation and addition of energy are small. In this manner, in this embodiment, the legs are provided with springs, and the passive dynamic characteristics of the springs are utilized, allowing for efficient control.

[0042] <Control method> Next, an example of a control method will be described. The walking control method proposed by Raibert, Hodgins, et al. is often applied as a controller for robots that can be modeled using SLIP (spring-loaded inverted pendulum) characteristics. This control method has also been extended to walking, so in this embodiment, we assumed that it can achieve efficient walking motion.

[0043] First, the concept of the control algorithm will be explained. Figure 9 is a diagram for explaining the decomposed elements in walking control. As shown in Figure 9, the walking motion generated by the walking algorithm can be decomposed into the following five motions. Movement (i) The movement of kicking the floor with the rear leg to switch from double support (DS) to single support (SS) Movement (ii) Movement that controls horizontal velocity during the SS phase by adjusting stride length Movement (iii) Movement that controls body rotation in both the DS and SS phases Movement (iv) Absorbing the impact of the swing leg at the start of the DS phase Movement (v) Movement to support the body with the legs during the SS period Of these actions, actions (i), (ii), and (iii) are active actions, while actions (iv) and (v) are passive actions. Passive actions do not consume energy.

[0044] Movement (i) is active, but by utilizing the spring energy stored in the support leg, movement (i) can be performed without consuming energy. First, in the DS stage, the front leg stores energy through the impact of the sole of the foot and the movement of supporting the body. Then, that energy is retained in the SS stage. Finally, in the DS stage, the hind leg releases the stored energy. Therefore, if the legs do not have damping elements, kicking motion (i) can be generated by a spring rather than a motor. In this way, when motions (i), (iv), and (v) are taken into consideration, the control unit 106 and the SLIP leg structure may be able to achieve highly efficient walking without generating electricity with a motor.

[0045] Figure 10 shows the coordinates of the motion space and task space for the example mechanism in Figure 3. Lines g451 and g452 are coordinates of the motion space. Lines g461 and g462, angle θ roc , θ thigt are coordinates in the task space. The angles ψ, Ψ, and length l are coordinates in the task space. The motor space is defined as θ M ={θ thigh ,θ lever} T Then, the task space x={l i ,Ψ i} T The foot position is defined as follows:

[0046]

number

[0047]

number

[0048] In equation (2), ψ i is expressed as the following equation (4).

[0049]

number

[0050] In equations (2) to (4), the subscript i indicates the front leg or rear leg in the DS phase (g461), and the supporting leg or swing leg in the SS phase (g471), as shown in FIG. 11. FIG. 11 is a diagram for explaining the definition of the names of the legs. link indicates the thigh link length, and φ and ψ indicate the angle related to the link length and leg rotation angle, respectively.

[0051] The motor command is given using the Jacobian matrix J as shown in the following equation (5).

[0052]

number

[0053] In equation (5), (x · ) cmd represents the command velocity in task space coordinates.

[0054] (DS stage) In an actual robot, kinetic energy is dissipated when the swing leg hits the ground. Therefore, during the DS phase, additional energy must be supplied by kicking off the floor with the rear leg. The target height of the waist in an upright position is h ref , the forward speed of the center of motion is v ref Then, the target velocity of the hind leg length during the DS period is defined as follows:

[0055]

number

[0056] In equation (6), E(h, v) is given by the following equation (7).

[0057]

number

[0058] In equations (6) and (7), K kickis the gain, g is the gravitational acceleration constant, and m is the total weight of the robot. In order to maintain the length of the front legs, the target speed of the front legs is defined as follows:

[0059]

number

[0060] In equation (8), K sup is the gain of the support leg, and l ref sup is the target leg length. In addition, to control the rotation of the body, the angular velocity commands for the front and rear legs are defined as follows:

[0061]

number

[0062] In equation (9), K body,p is the proportional gain and K body,d is the differential gain, and γ represents the rotation angle of the body.

[0063] (SS period) During the SS phase, the hind legs must maintain the posture of the upper body while keeping the leg length constant. Therefore, the target values ​​are defined as follows:

[0064]

number

[0065]

number

[0066] After swinging up, the swinging leg is ψ ref swg and ref swg The target value of the free leg is defined as follows:

[0067]

number

[0068]

number

[0069] In equations (12) and (13), K ψswg is the proportional gain and K lswg is the differential gain. Note that the reference value ψ ref swg and ref swg Several intermediate positions were used.

[0070] <Leg walking control method> Next, an example of a method for controlling leg walking will be described. Figure 12 is a diagram for explaining an example of a method for controlling leg walking according to this embodiment.

[0071] (a) Front leg in DS (Step S1) Using the detection value of the force sensor or torque sensor 101, the control unit 106 controls the motor position related to the adjustment of the supporting leg length not to change during single leg support, leaving the supporting motion to passive motion, and causing the spring to absorb the impact of the front leg touching the ground just before double leg support. That is, in this embodiment, control is performed so that the supporting leg length does not change, leaving the motion to itself and causing the spring to absorb the impact.

[0072] (b) Support leg in SS (Step S2) The kicking is performed using only the spring energy stored in the spring in step S1. As a result, according to this embodiment, the kicking operation can be performed by controlling the motor so as not to move its position.

[0073] (c) Ankle exercise (Step S3) The control unit 106 controls the motor so that no ankle torque is generated, thereby causing the ankle to passively rotate.

[0074] (d) Swing leg movement (Swing leg in SS) (Step S4) Using the detection value of the force sensor or torque sensor 101, the control unit 106 controls the actuator module 102, the actuator module 103, and the motor control unit 104 so that the foot reaches a specific position at the end of the swing leg.

[0075] The movements of steps S1 and S2, or steps S3 and S4, are controlled by breaking them down into the five movements explained using Figure 9 (movement (i) kicking the floor with the rear leg to switch from double support (DS) phase to single support (SS) phase, movement (ii) adjusting the stride length to control the horizontal speed in the SS phase, movement (iii) controlling the rotation of the body in both the DS phase and the SS phase, movement (iv) absorbing the contact impact of the swing leg at the start of the DS phase, and movement (v) supporting the body with the leg in the SS phase).

[0076] <Experiment, Evaluation> Next, we will explain the results of a walking experiment conducted on a robot to which the legged walking control method of this embodiment is applied. The experiment and evaluation were conducted using the robot 1 shown in Fig. 13. Fig. 13 is a diagram showing the external appearance of the robot to which the legged walking control method of this embodiment is applied.

[0077] Each motor of the robot was operated under speed control. When the robot received a start signal, it began walking, starting with a kicking motion in the DS period. The encoders measured the angles and angular velocities of the motors and joints, but not the motor torque.

[0078] By conducting walking experiments, we tuned the control parameters to achieve stable walking. The obtained control parameters were,K kick =0.04, K sup =25,K lswg =25,K body =20, K ψswg= 10. The reference leg length was l ref swg =0.485m, l ref swg =0.430m. The reference leg is ψ ref swg = 28°. The reference speed was v ref ={0.5,0} T m / s, and the base peak height is h ref =0.485m. These parameter settings enabled stable continuous walking.

[0079] The robot walked stably for 110 m until it received a stop signal. To test the robot's robustness, 0.005 m thick urethane boards were placed as obstacles. When two boards were stacked (0.01 m thick), the robot was able to walk without falling even when it tripped.

[0080] (Analysis results) To verify the control concept, we will explain the results of analyzing the spring deformation and power consumption in walking experiments. FIG. 14 is a diagram for explaining the items to be checked. The diagram with reference symbol g501 shows (a) front leg ground contact movement and (b) rear leg push-off movement. In (a) front leg ground contact movement, shock absorption by the front leg ground contact is checked. In (b) rear leg push-off movement, energy application by the rear leg push-off is checked. The diagram with reference symbol g502 shows (c) ankle movement and (d) swing leg movement. In (c) ankle movement, zero ankle torque is checked with a point foot. In (d) swing leg movement, passive movement with zero leg support velocity is checked. In FIG. 14, the unhatched double arrow g505 represents force or torque, and the hatched double arrow g506 represents velocity.

[0081] First, the results of confirming (a) front leg ground contact movement and (b) rear leg push-off movement will be explained using Figs. 15 and 16. Fig. 15 is a diagram showing the results of confirming the front leg's DS period and the supporting leg's SS period. Fig. 16 is a diagram showing the results of confirming the rear leg's DS period and the swing leg's SS period. Note that Figs. 15 and 16 focus on the right leg and divide it into four movement states (front leg's DS period, supporting leg's SS period, rear leg's DS period, swing leg's SS period). Fig. 17 is a diagram showing the displacement of leg length angular velocity.

[0082] The diagram with reference symbol g511 in Figure 15 is a model diagram when the front leg is in the DS period, the diagram with reference symbol g521 is a model diagram when the supporting leg is in the SS period, the diagram with reference symbol g531 in Figure 16 is a model diagram when the rear leg is in the DS period, and the diagram with reference symbol g541 is a model diagram when the swing leg is in the SS period.

[0083] In Figures 15 and 16, the graphs denoted by g512, g522, g532, and g542 are graphs showing the spring displacement φ, and the graphs denoted by g513, g523, g533, and g543 are graphs showing the spring displacement ψ. In the graphs denoted by g512, g513, g521, g522, g532, g533, g542, and g543, the horizontal axis represents time (seconds) and the vertical axis represents angle (degrees). In Figure 20, the graph denoted by g551 represents (a) the change in angular velocity during the front leg ground contact movement, the graph denoted by g552 represents (b) the change in angular velocity during the rear leg push-off movement, and the graph denoted by g553 represents (d) the change in angular velocity during the swing leg movement. In addition, in the graphs denoted by g551 to g553, line g554 represents the angular velocity of the joint, and line g555 represents the angular velocity of the motor. Each curve in the figure shows the trajectory of one step in a stable walking experiment without obstacles, and is aligned at the start of each phase (SS or DS phase). In order to analyze stable, periodic walking, the first and last few steps were excluded from the figure.

[0084] The angles in the task space are chosen as φ and ψ defined by equations (4) and (3), and are the angles related to prismatic motion and rotational motion, respectively. The deformation of the spring in the task space is defined as follows: (14) and (15).

[0085]

number

[0086]

number

[0087] In equations (14) and (15), △θ i ,i∈{in,out} denotes the deformation of the spring that can be calculated using the two encoders.

[0088] As shown in the graph g512 in Figure 15, the square columnar spring deformation Δφ of the front legs during the DS period gradually decreased. This movement corresponds to (a) front leg contact movement, and it was confirmed that the front legs were absorbing shock, i.e., absorbing energy, by contacting the ground. It was also confirmed that the motor speed was zero during this movement, as shown in the graph g551 in Figure 17.

[0089] As shown in the diagrams denoted by symbols g522 and g523 in Figure 15, the deformation of the spring was generally maintained, and the deformation of the stance leg during the SS period was kept constant.

[0090] As shown in the graphs g532 and g523 in Figure 16, the deformation of the hind legs during the DS period then gradually decreases. This movement corresponds to the (b) hind leg kicking movement, and it was confirmed that energy is being regenerated by the energy applied by the hind leg kicking. It was also confirmed that during this movement, the motor speed is almost zero, as shown in the graph g552 in Figure 17.

[0091] As shown in the diagrams g542 and g543 in Figure 16, the deformation of the free leg eventually approached almost zero.

[0092] Furthermore, as shown in the graph of symbol g553 in Figure 17, it was confirmed that the motor speed was almost zero during single-leg support (d) swing leg motion. This confirmed the "passive motion" caused by zero leg support speed.

[0093] These results confirmed that the impact of the front legs on contact is stored in the spring at the beginning of the DS period, and that the stored energy is maintained during the intermediate SS period before being released by the rear legs in the next step, the DS period. Furthermore, it was confirmed that the deformation of the rotational spring is almost zero during the DS period, but oscillates during the SS period, correcting the body tilt of the supporting leg during the SS period and moving the swing leg.

[0094] (Task space power during walking) 18 and 19 are diagrams showing the power in each task space during walking. Task space powers defined by the following equations (16) and (17) are shown in Figures 18 and 19. In Figures 18 and 19, the horizontal axis represents time (seconds) and the vertical axis represents P φ (W), P ψ (W).

[0095]

number

[0096]

number

[0097] In equations (16) and (17), the motor torque τ in , τ out is τ=αrK M i M where α is the gear efficiency, r is the gear reduction ratio, and K M is the torque constant of the motor, and i M is the motor current.

[0098] The figure of g561 in Figure 18 shows the front legs of the P φ (W) shows the change in the front legs. ψ (W) shows the change in the P of the front legs during the DS period, as shown by symbols g561 and g562. φwas initially high at the foot impact, then rapidly decreased to below 25W.

[0099] The diagram of symbol g571 in Figure 18 shows the P φ The figure of symbol g572 shows the change in (W) and the figure of symbol g572 shows the change in the P ψ The change in (W) is shown. As shown by symbols g571 and g572, the P φ was a relatively small value of less than 15W because the speed was almost zero.

[0100] The figure of symbol g581 in Figure 19 shows the P in the DS stage of the hind legs. φ The figure of symbol g582 shows the change in the hind legs during the DS stage. ψ As shown by symbols g581 and g582, in the push-off movement, a relatively large P is generated in the first half of the DS phase to accelerate the trunk. φ needed.

[0101] As shown in Figure 18, symbols g571 and g572, P ψ was vibrating violently. Therefore, the vibrations of symbols g522 and g523 in Fig. 15 are not passive due to the spring, but active due to the motor. On the other hand, the P of DS shown by symbols g561 and g562 in Fig. 18 and symbols g581 and g582 in Fig. 19 ψ decreased to a small value over time.

[0102] These results suggest that the power consumption required to support the weight of Robot 1 is much smaller than that required for the free legs or for correcting body rotation, since energy is supplied by the rectangular springs.

[0103] As described above, in this embodiment, walking experiments were conducted and evaluated using a 2DoF planar bipedal robot that uses a parallel mechanism consisting of a Naruto-Harut rubber spring and five bars to realize a virtual linear spring from the foot to the hip joint. The control method was decomposed into elements as described above, and used to achieve stable walking on uneven ground. As described above, the evaluation results showed that the robot walked by utilizing the passive dynamics of the prismatic springs in the supporting legs in both the DS and SS periods.

[0104] As described above, in this embodiment, by utilizing the passive dynamic characteristics of the body, supporting leg motion can be achieved without performing work. In this embodiment, during single leg support, the motor related to supporting leg length adjustment is not operated, and the supporting motion is left to passive motion. The front leg ground contact shock immediately before double leg support is absorbed by a spring, and is then operated to regenerate in the later stage of double leg support. This allows walking ideally without generating ankle torque, and the work related to leg length control is zero. In this embodiment, by leaving the motion of the robot 1 to passive motion as described above, the energy required for leg support motion during walking is zero.

[0105] In this embodiment, the walking motion is functionally broken down into elements, thereby improving the efficiency of the walking motion from the viewpoint of work. Furthermore, this embodiment provides an index for what mechanism should be adopted and how to control it in order to walk efficiently.

[0106] The above-described control method can also be applied to other robots that have virtual linear springs from the feet to the hip joints.

[0107] A program for implementing all or part of the functions of control unit 106 in the present invention may be recorded on a computer-readable recording medium, and the program recorded on the recording medium may be loaded into a computer system and executed to perform all or part of the processing performed by control unit 106. Note that the term "computer system" as used herein includes hardware such as an OS and peripheral devices. The term "computer system" also includes a WWW system equipped with a homepage provision environment (or display environment). The term "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, ROMs, and CD-ROMs, as well as storage devices such as hard disks built into computer systems. The term "computer-readable recording medium" also includes devices that retain a program for a certain period of time, such as volatile memory (RAM) within a computer system that acts as a server or client when the program is transmitted via a network such as the Internet or a communication line such as a telephone line.

[0108] The program may also be transmitted from a computer system storing the program in a storage device or the like to another computer system via a transmission medium or by transmission waves in the transmission medium. Here, the "transmission medium" that transmits the program refers to a medium that has the function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line. The program may also be a program that realizes part of the above-mentioned functions. Furthermore, the program may be a so-called differential file (differential program) that can realize the above-mentioned functions in combination with a program already recorded in the computer system.

[0109] The above describes the form for carrying out the present invention using an embodiment, but the present invention is not limited to such an embodiment, and various modifications and substitutions can be made within the scope that does not deviate from the gist of the present invention. [Explanation of symbols]

[0110] 1...robot, 101...force sensor or torque sensor, 102, 102-1, 102-2...actuator module, 103, 103-1, 103-2...actuator module, 104...motor control unit, 105...IMU, 106...control unit, 201...motor (including encoder), 202...first drive gear, 203...second drive gear, 204...spring, 205...lever link (including encoder), 206...shin link, 207...thigh link, 208...spring, 209...second drive gear, 210...first drive gear, 211...motor (including encoder)

Claims

1. A legged walking device including a spring mechanism, a first drive source for changing leg length, and a second drive source for rotating the legs, a first step of immobilizing the first drive source that changes the length of the support leg during single-leg support; a second step of absorbing a ground contact shock after the single leg support and immediately before the double leg support by the spring mechanism; and the legs of the legged walking device each include a first module and a second module provided at a hip joint; the spring mechanism includes a first spring mechanism provided in the first module and a second spring mechanism provided in the second module, the first spring mechanism and the second spring mechanism being connected by a link; a contact impact of the front leg immediately before the double support is stored in the first spring mechanism and the second spring mechanism at the beginning of the double support; the energy stored in the first spring mechanism and the second spring mechanism is maintained during the single stance and then released by the rear leg in the next double stance; A method for controlling legged walking.

2. In the first step, the supporting movement is left to passive movement; In the second step, the actuator is operated so as to regenerate during the latter stage of double-leg support. The method for controlling legged walking according to claim 1 .

3. a first actuator module including the spring mechanism and the first drive source; a second actuator module including the spring mechanism and the second drive source; By controlling the 3. A method for controlling leg walking according to claim 1 or 2.

4. the spring mechanism includes an outer shaft, an inner shaft, and a plurality of urethane rubber springs evenly and concentrically arranged between the outer shaft and the inner shaft; The outer shaft is connected as an input to a drive gear; The inner shaft is connected to a predetermined link as an output. The method for controlling leg walking according to any one of claims 1 to 3.

Images