KR101421230B1 - Control System for Compliant Legged Robot - Google Patents

Abstract

The present invention relates to a control system for a robot having a flexible leg structure, the robot leg driving unit driving a leg of the robot; A controller for controlling the robot leg driving unit; And a travel path planning device for providing a control target value for running the robot to the controller, wherein the travel path planning device provides the control target value at a time interval of ts / scale, wherein ts is feedback Is the control loop sampling time, and scale? 1. The control system of the flexible leg robot is capable of standing in a stopped state, avoiding the saturation of the actuator and reaching a speed in a self-stable region at a low speed, To be able to do so.

Description

{Control System for Compliant Legged Robot}

BACKGROUND OF THE INVENTION 1. Field of the Invention [0002] The present invention relates to a flexible leg robot, and more particularly, to a traveling path control system for a robot having a flexible leg structure.

Generally, a robot having a flexible leg structure must calculate a leg angular range, a leg angular velocity, a leg angle offset, and the like with respect to a moving speed of the robot in a traveling route planning process. In the prior art, the leg angle range, the leg angular velocity, the leg angle offset, and the like were determined by tuning each of the moving speeds of the walking robot.

Korean Patent Application No. 2011-0120394 filed by the applicant of the present application has disclosed an apparatus and method for planning a traveling path of a walking robot. The present invention relates to a traveling path planning apparatus and method for planning a traveling path in a numerical analytical manner as a dynamic equation of a spring-mass model for stable running of a robot having a flexible leg structure.

The prior art of the walking robot 1) forms a landing angle within a self-stable region with respect to the traveling speed, 2) generates a desired path of the robot from a spring-mass model, which is a simple model of the robot having a flexible leg structure, Create and apply the desired path of the robot leg. If the desired path of the robot leg is generated and applied by the above control method, energy consumption for maintaining the stability of the robot is not required and energy consumption for movement is not required when the robot travels at the constant speed and there is no energy loss. That is, it only requires energy supplement for energy loss and energy consumption in acceleration / deceleration traveling

 The present invention allows a robot having a flexible leg structure to stand in a stopped state, avoiding the saturation state of the actuator and reaching a speed within a self-stable range at a low speed, So that the user can drive the vehicle.

According to an aspect of the present invention, there is provided a robot control apparatus comprising: a robot leg driving unit for driving a leg of a robot; A controller for controlling the robot leg driving unit; And a travel path planning device for providing a control target value for running the robot to the controller, wherein the travel path planning device samples the controller control loop at a time interval of ts / scale, wherein ts is feedback Control loop sampling time, and a scale > = 1.

According to the present invention, since the landing angle of the robot leg is calculated on the basis of the self-stabilizing region, the robot with the flexible leg structure can maintain the stability, and the control input value can be minimized by the travel route planning, . Particularly, according to the present invention, there is an advantage that the traveling path planning can be obtained numerically from the kinetic equation of a dynamic model of a robot (spring-mass model) without tuning.

 In addition, the present invention allows a robot with a flexible leg structure to stand in a stopped state, avoiding the saturation of the actuator and reaching a speed within a self-stable range at low speed, So that the vehicle can be driven at a high speed.

Fig. 1 shows a spring-mass model for a walking state of a robot which is standing on two legs,
Fig. 2 shows a spring-mass model for a running state of a robot which is standing on one leg,
FIG. 3 is a view showing a traveling position model for two legs in a state in which the robot shown by the spring-mass model of FIG. 1 is walking,
Fig. 4 shows a traveling position model for one leg in a state where the robot shown in Fig. 2 is shown as a spring-mass model,
5 schematically shows a robot control system according to an embodiment of the present invention,
FIG. 6 is a graph showing a magnetostatic region according to an embodiment of the present invention,
FIGS. 7 and 8 are graphs showing a scale and a gain change of a flexible leg robot according to an embodiment of the present invention,
FIG. 9 is a graph showing a change in speed of a flexible leg robot according to an embodiment of the present invention, and FIG.
FIG. 10 is a graph showing the mechanical cost of transport (energy index) indicating the degree of energy efficiency of a flexible legged robot according to an embodiment of the present invention.

Hereinafter, embodiments disclosed in this specification will be described in detail with reference to the drawings. Like reference numerals in the drawings denote like elements, unless the context clearly indicates otherwise. The exemplary embodiments described above in the detailed description, the drawings, and the claims are not intended to be limiting, and other variations are possible without departing from the spirit or scope of the subject matter disclosed herein. The components of the present disclosure, that is, components that are generally described herein and described in the drawings may be arranged, configured, combined, and configured in a variety of different configurations. It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit and scope of the invention.

A simple model of the walking / running state of the robot having the flexible leg structure is as shown in Figs. Figs. 1 and 2 show a spring-mass model of a leg in a state where the robot is walking and running. Fig. 3 and Fig. 4 show a model of a running position of the leg in a state in which the robot is running, It is. Hereinafter, a robot dynamic model (spring-mass model) used in the present invention will be described with reference to FIGS.

In the embodiment shown in Figs. 1 to 4 m represents the mass of the robot with a flexible bridge structure to the material point mass (point mass), x, y are location coordinates of m, K is the spring modulus, l ₁ is a robot L ₂ denotes the second leg length of the robot, l ₀ denotes the original length of the first and second legs of the robot (i.e., the length when the spring is not compressed), F _1, F ₂ denotes a spring elastic force _{(F i = K (l 0} -l i) i = 1,2) of the robot's _feet, τ _1, τ ₂ represents the torque applied to the robot's feet, θ ₁ , θ ₂ is the angle of the legs of the robot from the vertical axis, and x _g1 , y _g1 , x _g2 and y _g2 are the positions where the legs touch the ground. Here, the subscript 1 indicates the bridge 1, the subscript 2 indicates the bridge 2, and the subscript 1 or 2 does not mean either 1 or 2.

In addition, α represents an angle formed by the leg of the robot with the ground, and α ₀ represents a touch down angle when the leg of the robot lands on the ground. _{y apex, p (p = i} , i + 1, i + j, i + j + 1) is the height p-th _{vertex, V x, apex, p (} p = i, i + 1, i + j, i + j + 1) represents the x-direction velocity at the p-th vertex, and g represents the gravitational acceleration (g = 9.81 m / s ² ). Here, the leg mass is assumed to be zero.

The equation of motion for a walking or running state of the robot in the embodiment represented by the spring-mass model as shown in FIGS. 1 to 4 is expressed by Equation 1 below.

Hereinafter, a robot control system according to the present invention will be described with reference to Figs. 5 and 6. Fig.

FIG. 5 is a schematic view of a robot control system to which a robot travel route planning apparatus according to the present invention is applied. FIG. 6 is a diagram illustrating a self-stable region of a landing angle according to a target speed according to an embodiment of the present invention. Fig.

5, the control system of the robot includes a robot leg driving unit 300 of a flexible leg robot as a plant to be controlled, a controller 200 for controlling the robot leg driving unit 300, And a traveling path generating unit 100 for providing a control target value for driving the vehicle.

Traveling route generation section 100 is the x direction in the initial peak from the x direction, the target speed (V _{x, apex, desired)} for the input receives the magnetic stability region (self-stable region) shown in Figure 6 at the desired vertex with respect to the running direction the speed of the landing leg with respect to the stable robot (V _{x, apex, 0)} of each (α ₀₎ is calculated. From the landing angle (α ₀ ), the leg angle at landing,

θ ₀ = - (90 ° -α ₀ ) (in the case where θ is counterclockwise from the vertical axis in FIGS. 1 and 2, is defined as a negative sign). In Equation (1), numerical integrations (where the numerical integration method applied is a fourth-order Runge-Kutta method) of the equations of motion in which τ ₁ and τ ₂ are 0 are generated to generate desired paths x _d and y _d of the robot, The desired path? _D (? _Id = tan ^-1 ((x _d - x _gi ) / (y _d - y _gi )), i = 1, 2).

Then, in the controller 200, the path of the robot leg is set to a desired path? _D (? _Id = tan ^-1 ((x _d - x _gi ) / (y _d - y _gi )), i = 1, 2). At this time, P, PD, PID, nonlinear control technique and the like can be applied to the control method.

FIG. 6 is a graph showing a self-stable region of a landing angle according to a target speed according to an embodiment of the present invention. The stability angle is obtained by the stability analysis of the landing angle, which allows the robot to travel stably even if the input (eg, torque) is not applied. The magnetostrictive region has a stable landing angular range at a constant velocity over V _{x, apex, 0} indicated by an arrow, as shown.

Therefore, if the command speed Vx _{, apex, desired} is directly applied at the minimum speed in the self-stabilizing region in the stopped state, the robot leg driving unit 300 having the limited capacity as shown by the arrow in FIG. 6 saturates, So that the traveling path generating unit 100 generates the control target value at time intervals of ts / scale in the present invention. ts is the control sampling time of the control signal applied by the controller 200 and the scale value is greater than or equal to 1 (scale≥1). The control target value generation from the stop to the walking speed in the self-stabilizing zone can be generated at a time interval smaller than the control sampling time interval and the speed can be arbitrarily lowered. On the other hand, at this time, the gain scheduling can be made to be able to stand in a stopped state and the speed of walking can be increased to reach the walking speed in the self-stabilizing region.

4, the controller 200 calculates a torque? To be applied to the robot leg on the basis of the robot leg target angle provided by the traveling path generating unit 100, and outputs the torque? And controls the leg driving unit 300.

On the other hand, the angle? Of the robot leg is input again to the input value of the controller 200 to perform a closed loop control. For reference, the controller 200 may use various methods such as P (Proportional) control, PD (Proportional Derivative) control, PID (Proportional Integral Derivative) control, and nonlinear control.

6 is a graph showing a self-stable region in a graph of a landing angle according to a target speed.

As shown in FIG. 6, the magnetostrictive region has a stable landing angular range at Vx _{, apx, 0} equal to or greater than a constant velocity value. As shown by the red arrow in FIG. 6, when the command speed is given at the minimum speed in the self-stabilizing area in the stop state, there is a danger that the robot leg driver 300 having a limited capacity may reach a saturation state. Accordingly, the robot control system according to the present invention uses a method of adjusting the sampling time.

That is, when ts is the feedback control loop sampling time, the traveling path generating unit 100 generates a desired path at a time interval of ts / scale, scale≥1.

Therefore, if the scale is larger than 1, ts / scale becomes smaller than ts, so that desired path generation is performed in a shorter time interval, and the desired path is arbitrarily reduced by giving it as a command value to the feedback control loop. For example, if ts = 0.001 sec and scale = 10, then the path generation time interval is 0.0001 sec, and if it is given as a command value to the feedback control loop with sampling time ts = 0.001 sec, . In other words, when scale = 1, the desired path is generated at a time interval of 0.001 sec. When scale = 10, the desired path is generated at a time interval of 0.0001 sec. The path is generated at a time interval 10 times shorter than that in the case where the sampling time ts is 0.001 sec. Therefore, the command speed is optionally lowered 10 times because it is used as the command value of the feedback control loop.

In the method of applying the above path planning method, when the robot receives the command speed at the minimum speed in the self-stabilizing zone in the stop state, the scale value is set to a sufficiently large value to reduce the command speed arbitrarily, And the scale value is set to 1 so that the desired path generation time interval is equal to the feedback control loop sampling time. This makes it possible to avoid the saturation state of the driving unit when the command speed is immediately given from the stop state to the minimum speed in the self-stabilizing region in the state of scale 1.

And performs gain scheduling of the controller 200 as in the above path planning. Here, the proportional control (P control) is treated as one embodiment. FIGS. 7 and 8 show change values of the scale with respect to time and changes in the gain Kp of the controller 200 at this time.

As shown in FIGS. 7 and 8, when the scale value is large, the gain value Kp is increased. The reason for doing this is to increase the holding force of the leg at the stop and arbitrarily lowered command speed. The reason why the gain value Kp is decreased when the scale value is small is that when the scale value becomes small, τ ₁ = τ ₂ = 0 The gain Kp can be made small and the gain Kp can be made small so that the driving unit 300 having a limited capacity can be driven at a high speed.

The path planning and gain scheduling method according to the present invention allows the robot having a flexible leg structure to stand in a stopped state, to avoid the saturation state of the actuator, to reach a speed in the self-stabilizing region at a low speed, This allows the driver to drive at a high speed.

Hereinafter, a path planning method according to the present invention and a control system operation of a flexible leg robot through gain scheduling will be described.

In one embodiment of the present invention, when generating a path from the stationary state to the speed in the self-stabilizing zone by starting to walk, the path (ts / scale, scale? 1) . That is, the actual command speed is lowered by setting the path generated in a time interval shorter than the control sampling time interval to the control loop as a set value. At this time, the gain K _p of the control signal is subjected to gain scheduling (see FIG. 7). The reason for doing this is to avoid saturation of the actuator when the setpoint is given directly to the speed within the self-stabilizing zone for walking from stop. Further, as shown in FIG. 7, the gain Kp is increased when the scale value is large, and the supporting force of the legs can be increased at a commanded speed that is stopped and arbitrarily lowered. This can be said to be the case when a person is walking with a slow motion arbitrarily to the leg. And the gain Kp can be made smaller when the scale value becomes smaller because Kp can be made smaller because the scale value becomes closer to the equation of motion of Equation 1 where τ ₁ and τ ₂ are 0, And to reduce the gain Kp in order to be able to travel at a high speed with a limited driver. The above scale and the change value of the gain Kp with respect to time can be set as shown in FIGS.

Experimental results according to the above embodiment are as shown in the graphs of FIGS. FIG. 9 is a graph showing the change in the robot moving speed, and FIG. 10 is a graph showing the mechanical cost (Cmt) according to the movement.

As shown, in FIG. 9, Vxd represents the actual command speed at which ts / scale, scale? 1 is applied to the above-described desired path generation. Vx represents the actual moving speed of the robot. The reason why the variation width of Vx is large is that the variation of Vx is large as shown in the graph due to the vibration caused by the impact when the robot leg contacts the ground when the robot walks and runs in the embodiment. The average value of Vxd and Vx is about the same for about 50 seconds, and the average value of Vx is slightly larger than Vxd between 50 and 55 seconds, and the average value of Vx is larger than Vxd for 55 seconds to about 66 seconds. And, between 66 and 71 seconds, acceleration is the interval, Vx _{, apex, desired} It is the interval that includes the set interval of the transition speed that transitions from walking to running on the basis of standard. At this time, Vx does not follow Vxd well, but follows Vxd in the vicinity of 71 seconds, and when Vxd exceeds 71 seconds, the average value of Vx is similar to Vxd.

FIG. 10 is a graph showing the Cmt (mechanical cost of transport), where Cmt is an index indicating the degree of energy efficiency generally used, Cmt = P / mgV, P is an externally applied power , mg is the weight of the robot, and V is the moving speed of the robot. Cmt is a dimensionless value as defined, and a smaller value means higher energy efficiency. 10, the desired path generation is performed at a time interval (ts / scale, scale> 1, see FIG. 8) shorter than the control sampling time interval (ts = 0.001 sec) as described above. In the case of more than 55 seconds, the desired path generation time interval is equal to the control sampling time interval (ts / scale, scale = 1, see FIG. 8).

As mentioned above, setting the desired path generation to a time interval shorter than the control sampling time interval can arbitrarily lower the command speed, which is equivalent to instructing an unnatural walk such as slow motion in the case of a person. Therefore, it can be seen that the energy efficient walking is not performed in this case, and it can be confirmed that the value of Cmt in FIG. 10 is 55 seconds or less, which is larger than 55 seconds or more. When the desired path generation and control sampling time intervals are equal to or longer than 55 seconds, Cmt is _calculated as Vx _{, apex, desired} Cmt ≒ 0.3852 for walking below the reference transition speed, and Cmt ≒ 0.3481 for running over the transition speed.

The traveling robot of this embodiment has a relatively high energy efficiency, and it is possible to perform the transition from the stationary state to the walking, walking to running, and running.

That is, the traveling robot according to the present embodiment is a robot capable of performing walking, walking, running, and running from a stationary state, and there is no restriction on running, and energy efficiency is also high in walking and running

 The path planning and the gain scheduling method proposed in the present invention can allow the robot having the flexible leg structure to stand in the stopped state and avoid the saturation state of the actuator and can reach the speed in the self stable region at low speed, It is possible to drive at a high speed with a limited driver.

Claims (2)

In a control system of a flexible leg robot,
A robot leg driving unit for driving the legs of the robot;
A controller for applying a control signal to the robot leg driving unit to control driving of the robot leg driving unit; And
And a travel path generating unit for providing a control target value for running the robot to the controller,
Wherein the travel path generating unit provides the control target value at a time interval of ts / scale until the traveling speed of the robot reaches the walking speed in the self-stabilizing region from the stop,
Here, ts is the control sampling time of the control signal of the controller, and scale > = 1. The method according to claim 1,
Wherein the controller adjusts the gain of the control signal to a large value when the scale value is large and adjusts the gain of the control signal to be small when the scale value is small.

Images