CN101898594B - Walking method for dynamic biped robot - Google Patents

Abstract

A walking method for a powered biped robot, which belongs to the technical field of robot walking control, is characterized in that, during the walking process of the robot, power is added to the supporting legs to make the supporting legs actively complete elongation and shortening in one walking cycle, through Control the extension and shortening position of the supporting legs to realize the potential energy supplement to the system. The gait of this walking method is determined by five keyframes and described by two angle parameters, and the keyframes are connected by smooth curves with continuous first-order derivatives. On the prototype, the robot can walk stably and at different speeds under the control of this dynamic walking method.

Description

A walking method for a powered biped robot

technical field

The invention is a biped powered walking method based on the principle of passive walking, and realizes the open-loop walking control of the powered biped robot by adding power to the supporting legs of the robot.

Background technique

How to achieve fast and stable walking is the focus and difficulty in the research of biped robots. At present, the walking methods of biped robots mainly include static walking, ZMP walking, and limit cycle walking. Among them, static walking is the earliest and most basic walking method, which requires that the center of mass of the robot is always kept within the polygon formed by the feet on the ground during the walking process. This method is easy to maintain the stability of the robot, but it is also extremely limits the walking speed of the robot. The ZMP theory requires that the zero-moment point of the robot is always kept within the polygon formed by the feet. This method reduces artificial constraints to a certain extent compared to static walking, and has achieved great success in prototype applications, including Honda's ASIMO, Japan HRP4 of AIST Research Institute, and Qrio of Sony Corporation. However, it is difficult for robots designed according to the ZMP theory to make breakthroughs in terms of natural gait and energy efficiency.

Limit cycle walking is a new walking concept that has emerged in recent years. Its proposal is inspired by human walking and requires that the periodic gait sequence be orbitally stable, that is, the gait sequence can form a stable state in the state space. Limit cycle, but does not have local stability at any instant in the gait cycle. This method has fewer artificial constraints on the robot, and fully utilizes the dynamic characteristics of the robot itself under the gravity field, so it has a large space to improve the energy efficiency, speed, and anti-interference ability of the robot walking. At present, the successful examples using the principle of limit cycle walking include MIT's Spring Flamingo and its virtual model control method, the French Academy of Sciences' Rabbit and its hybrid zero dynamics control method, Geng et al.'s RunBot and its central nervous system control method, and CMU's Biped robot and its reinforcement learning method, etc. These robots have achieved great breakthroughs in terms of walking speed, energy efficiency, and anti-interference ability, but the gait generation method is relatively cumbersome, and some require the use of machine learning, which has higher requirements for the experimental environment.

Passive walking is a typical example of limit cycle walking. The robot walks down a slightly inclined slope without any control. The gravitational potential energy provided by the slope is converted into the kinetic energy required for the robot to walk. The gait generated by passive walking is very natural, and the energy efficiency can reach the human level, which is about one tenth of that of ZMP walking robot ASIMO. In order to achieve passive walking on flat ground, Cornell University uses a method of increasing power at the ankle of the robot. After each swinging leg collides with the ground, the soles of the feet push the ground to inject energy into walking. Deflt University uses the method of clamping the hip joint before the swing leg hits the ground, which also achieves the purpose of replenishing energy. However, the energy replenishment timing of the above two methods is located before and after the collision time, and the energy is replenished instantaneously, which requires extremely high energy density, which limits the walking speed of the robot to a large extent. At the same time, this energy replenishment method It will cause greater disturbance to the gait and reduce the stability of walking.

In the previous patent ZL200810116148.6, our research group proposed a method of power walking on flat ground, which supplements the potential energy of the center of mass of the system by actively extending the supporting legs and shortening the swinging legs during walking, so as to realize energy supplementation and loss during walking balance. The walking method of the powered biped robot in the present invention is based on the principle of passive walking, and supplements the potential energy of the center of mass of the system by first actively extending and then actively shortening the supporting legs during one-step walking. Compared with the method in the previous patent, this method also eliminates the impact on the walking stability caused by the energy added at the moment the swing leg collides with the ground, enabling the robot to achieve higher walking stability, and it only needs to start Loop control is simple to implement and has a very small amount of calculation, so it is suitable for occasions that require high real-time performance. In addition, it concentrates the control on the supporting legs instead of controlling the swinging legs, which reduces the variables of control, and this control method can make the energy added to the system a fixed value, thus having a progressive energy balance Advantage.

Contents of the invention

The purpose of the present invention is to propose a method of adding power to the supporting legs based on the principle of passive walking to realize the dynamic walking method of the biped robot on flat ground.

The biped robot model of the present invention is as shown in Figure 1, and wherein 1 is robot body, and quality is M, and 2 is supporting leg thigh, and 3 is supporting leg calf, and 4 is equivalent supporting leg (by supporting leg thigh 2 5 is the swing leg thigh, 6 is the swing leg calf, and 7 is the equivalent swing leg (constituted by the connection line from the top of the swing leg thigh 5 to the swing leg calf 6 end) . The robot has three angles θ, α, β, where θ and α are the key angles for controlling the walking of the robot, θ is the angle between the equivalent support leg 4 and the equivalent swing leg 7; α is the support leg thigh 2 and the equal The angle between the effective support legs 4 determines the length of the equivalent support legs 4, when the support leg knees are bent, α>0, and when the support leg knees are straight, α=0; β is the swing leg 5 and the equivalent swing leg 7 The angle between them determines the length of the equivalent swing leg. When the swing leg knee is bent, β>0, and when the swing knee is straight, β=0. Since the swing leg does not need to be controlled during walking, β remains a constant. It is convenient to define the key angle in this way to determine the angle between the two equivalent legs and the length of the two equivalent legs. The one-step walk of the robot consists of a swing process and a collision. The swing process refers to the end of the supporting leg of the robot touching the ground, and swings forward with the end as the axis. At the same time, the swing leg swings in the air from the back of the support leg to the front of the support leg. The end of the swinging leg hits the ground momentarily while the supporting leg lifts off the ground. After a collision the swing leg converts to a support leg and the support leg converts to a swing leg. The one-step walk of the robot starts after the collision of the previous step, and ends after the collision through the swing process.

The energy conversion principle of the biped robot walking method of the present invention is shown in Figure 2. In order to highlight the role of the supporting legs, the two legs are only drawn at the beginning and end of a step in the figure, and the swinging is omitted at the middle moments. leg. The robot starts at time A. At this time, the equivalent supporting leg is the same length as the equivalent swinging leg. The swinging leg is about to leave the ground. After leaving the ground, the supporting leg swings freely until time B. In the process of one-step swing, the robot first straightens the knee joint of the supporting leg (from time B to time C), and then bends the knee joint of the supporting leg (from time C to time D), so as to achieve the elongation and shortening of the equivalent supporting leg 4 length to supplement the potential energy of the system. After the knee joint of the supporting leg is bent, it remains unchanged and continues to swing forward. The two legs collide with the ground at a fixed angle to complete a cycle of walking. The process of straightening the supporting legs is the process of adding energy to the system, and the energy of the system increases by E ₁ . In order to ensure that the two legs are equal in length during a collision, and that the supporting legs cannot be extended infinitely, during a one-step walking process, the supporting legs must be bent back to their original length before colliding with the ground. The bending process of the supporting legs is a systematic In the process of energy reduction, the system energy decreases by E ₂ . In order to increase the total energy of the robot during one step of walking, that is, the difference E _c between E ₁ and E ₂ is greater than zero, it is necessary to strictly control the straightening and shortening positions of the supporting legs, that is, to ensure that the straightening position of the supporting legs is compared to The shortened position is closer to the vertical position, so that the total energy of the system during the swing process can be increased. In this method of energy supplementation, only the effect of the power added on the supporting leg on the energy supplementation of the system is considered. Since we only consider the mass concentrated at the hip and ignore the model of the mass of the leg, the impact of the swinging leg on the system The energy has no effect, so the bending remains unchanged during walking, and it is assumed that the swinging leg can always swing to a hip-locked state where the angle between the swinging leg and the supporting leg is a fixed value.

The biped robot described in the present invention has the characteristic of progressive balance of energy, and the performance of this characteristic is as follows. During one-step walking, the increase in system energy is only related to the straightened and bent positions of the supporting leg, but not to the initial velocity of the supporting leg. When the position of straightening and bending of the robot's supporting legs is fixed, the energy added in each step of walking is a fixed constant. When the added energy is the same as the energy lost in the collision of the swinging legs, the robot's walking will enter the limit The initial velocity after the collision is the velocity at the fixed point. If the initial speed of the robot is greater than the speed at the fixed point, since the energy lost in the collision of the swinging leg is proportional to the initial speed, the energy lost in the collision is greater than the added energy, so that the initial speed of the next step is reduced until it is equal to The speed at the fixed point is equal; on the contrary, if the initial speed of the robot is lower than the speed at the fixed point, the energy lost in the collision is less than the added energy, so that the initial speed of the next step increases until it is at the fixed point. speeds are equal.

The biped robot walking method of the present invention is characterized in that it contains the following steps in sequence:

Step (1), construct a biped robot, as shown in Figure 3(b), the steps are as follows:

Step (1.1), establish the connection between the trunk 1 and the first thigh 4 and the second thigh 5: the trunk 1 is fixedly connected with the bodies of the first hip joint motor 2 and the second hip joint motor 3 placed coaxially on the left and right respectively, and The rotation output shaft of the first hip joint motor 2 is connected with the first thigh 4, the rotation output shaft of the second hip joint motor 3 is connected with the second thigh 5,

Step (1.2), establishing the connection between the first thigh 4 and the first calf 7, the second thigh 5 and the second calf 9: the end of the first thigh 4 is fixedly connected to the body of the first knee joint motor 6 , the rotation output shaft of the first knee motor 6 is connected with the first lower leg 7, the end of the second thigh 5 is fixedly connected with the body of the second knee motor 8, and the rotation output of the second knee motor 8 is The shaft is connected with the second lower leg 9,

In step (1.3), the four motors described in step (1.1) and step (1.2) are all servo motors, and _Ship1 and _Ship2 represent the first hip joint motor 2 and the second hip joint motor 3 respectively The rotation angles of the first knee joint motor 6 and the second knee joint motor 8 are represented by S _knee1 and S _knee2 respectively, and a host computer is used to control the four motors, wherein: the _Ship1 is the The angle between the first thigh 4 and the vertical direction of the trunk 1, when the end of the first thigh 4 is located in front of the trunk 1, S _hip1 >0, and when it is located behind the trunk 1, S _hip1 <0, said S _hip2 is the The angle between the second thigh 5 and the vertical direction of the trunk 1, when the end of the second thigh 5 is located in front of the trunk 1, S _hip2 >0, and when it is located behind the trunk 1, S _hip2 <0; the S _knee1 is the first The angle between the lower leg 7 and the first thigh 4, when the first lower leg 7 is bent backward relative to the first thigh 4, S _knee1 >0, when the two are parallel, S _knee1 =0, and the S _knee2 is the angle between the second calf 9 and the second thigh 5, when the second calf 9 bends backward relative to the second thigh 5, S _kneee2 >0, and when the two are parallel, S _knee2 =0,

Step (1.4), the control signal input end of each motor described in the step (1.1), step (1.2), step (1.3) is connected with the control signal output end of a host computer respectively;

Step (2), a gait cycle T is set in the host computer, and the gait cycle T refers to the time elapsed from the start of one step to the impact of the swing leg, wherein, the start time t=0 means that the swing leg At the moment of leaving the ground, the collision time t=T refers to the collision between the swing leg and the ground, that is, the moment when one gait cycle ends and the next gait cycle begins. At this moment, the previous supporting leg becomes a swing leg, and The former swinging leg becomes the supporting leg.

Step (3), in the host computer, in the one gait cycle, set five key frames, two key joint angles θ and α, as shown in Figure 4:

The first key frame (figure A), at t=0, determines the initial posture of the robot one step, where θ=-θ ₀ , θ ₀ is a non-negative constant, representing the two equivalent legs when t=0 The angle between them determines the size of the stride; α=α ₀ , α ₀ is a non-negative constant, which means that t=0 is the bending angle of the support leg thigh 2 relative to the equivalent support leg 4 .

The second key frame (Fig. B), when located at t=T ₁ , wherein: α=α ₀ , which is the same as α in the first key frame, represents that the length of the equivalent support leg 4 is between the first key frame and the second key frame remains constant between frames. This keyframe represents the starting point for the straightening of the supporting leg.

The third key frame (Fig. C) is located at t= _T2 , wherein: α=0, which means that the knee joint of the supporting leg is straightened, and the center of mass of the hip rises to the highest point. From the second keyframe to the third keyframe is the process of adding energy to the system, and since the knee joint of the swinging leg is still in a bent state at this time, it is possible to avoid collisions between the lower leg of the swinging leg and the ground during walking.

The fourth key frame (Figure D), at t=T ₃ , where: α=α ₀ , means that the knee joint of the supporting leg is bent back to the original state, and the supporting leg is shortened from the third key frame to the fourth key frame process, this is to ensure that the legs are of equal length during the collision and remain bent when the supporting leg becomes the swinging leg in the next step.

The fifth key frame (Fig. E), at t=T, determines the attitude of the robot at the time of collision, where: θ=θ ₀ represents the angle between two equivalent legs at the time of collision; α=α ₀ , from the fourth Keyframe to the fifth keyframe, the supporting leg remains unchanged. After a collision with the ground, the supporting leg becomes a swinging leg and the swinging leg becomes a supporting leg.

Step (4.1), the host computer controls the robot to walk according to the following steps in turn, setting the period T of each step of walking, three key times T ₁ , T ₂ , T ₃ , the calculation time is t, and t starts from 0 , the upper computer calculates the value of θ and α every time t according to the formula

(1)

$\mathrm{<span\; class="notranslate">\; \&alpha;</span>}<span\; class="notranslate">\; =</span>\left\{\begin{array}{cc}{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}& \mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; t</span>}\mathrm{<span\; class="notranslate">\; mod</span>}\mathrm{<span\; class="notranslate">\; T</span>}<span\; class="notranslate">\; <</span>{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="T"\; filenames="HSA00000209917400013.png,HSA00000209917400021.png"\; state="\{\{state\}\}">T</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}\\ \frac{{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}}\mathrm{<span\; class="notranslate">\; cos</span>}\frac{\mathrm{<span\; class="notranslate">\; \&pi;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; T</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; )</span>}{{\mathrm{<span\; class="notranslate">\; T</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="T"\; filenames="HSA00000209917400013.png,HSA00000209917400021.png"\; state="\{\{state\}\}">T</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}}<span\; class="notranslate">\; +</span>\frac{{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}}& {\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="T"\; filenames="HSA00000209917400013.png,HSA00000209917400021.png"\; state="\{\{state\}\}">T</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; t</span>}\mathrm{<span\; class="notranslate">\; mod</span>}\mathrm{<span\; class="notranslate">\; T</span>}<span\; class="notranslate">\; <</span>{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="T"\; filenames="HSA00000209917400011.png,HSA00000209917400012.png"\; state="\{\{state\}\}">T</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}\\ \frac{{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}}\mathrm{<span\; class="notranslate">\; cos</span>}\frac{\mathrm{<span\; class="notranslate">\; \&pi;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; T</span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}<span\; class="notranslate">\; )</span>}{{\mathrm{<span\; class="notranslate">\; T</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="3"\; label="T"\; filenames="HSA00000209917400011.png,HSA00000209917400012.png"\; state="\{\{state\}\}">T</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}}<span\; class="notranslate">\; +</span>\frac{{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}}& {\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="T"\; filenames="HSA00000209917400011.png,HSA00000209917400012.png"\; state="\{\{state\}\}">T</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; t</span>}\mathrm{<span\; class="notranslate">\; mod</span>}\mathrm{<span\; class="notranslate">\; T</span>}<span\; class="notranslate">\; <</span>{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="3"\; label="T"\; filenames="HSA00000209917400011.png,HSA00000209917400012.png"\; state="\{\{state\}\}">T</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}\\ {\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}& {\mathrm{<span\; class="notranslate">\; <figure-callout\; id="3"\; label="T"\; filenames="HSA00000209917400011.png,HSA00000209917400012.png"\; state="\{\{state\}\}">T</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; t</span>}\mathrm{<span\; class="notranslate">\; mod</span>}\mathrm{<span\; class="notranslate">\; T</span>}<span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; T</span>}\end{array}\right.$

(2)

According to the above formula, two curves of θ=f _θ (t) and α=f _α (t) can be obtained, which are respectively the curves of the hip angle and the bending angle of the knee joint of the supporting leg, as shown in Figure 5, the variable θ, The first derivative of α with respect to t is continuous.

Step (4.2), in the above-mentioned upper computer, when the number of steps is n (n is the integer part of the division result of t and T), the above-mentioned walking parameters θ, α and the bending angle of the swinging leg knee joint are calculated according to the following formula The rotation angle of each motor of the biped robot under the value of β (since the knee joint remains unchanged after the supporting leg becomes a swinging leg, that is, β=α ₀ ). When n is an odd number, _Ship1 and S _knee1 are the angles of the hip joint and the knee joint of the supporting leg respectively, and _Ship2 and S _knee2 are the angles of the hip joint and the knee joint of the swing leg respectively, and when n is an even number, _Ship2 and S _knee2 are the angles of the hip joint and knee joint of the supporting leg respectively, and _Ship1 and S _knee1 are the angles of the hip joint and knee joint of the swing leg respectively, that is to say, when n changes from an odd number to the odd number plus When there is an even number formed by 1, the supporting leg as the first thigh and the swinging leg as the second thigh are interchanged after one step of walking.

$\left\{\begin{array}{c}{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="S\; hip"\; filenames="HSA00000209917400013.png,HSA00000209917400021.png"\; state="\{\{state\}\}">S</figure-callout></span><figure-callout\; id="1"\; label="S\; hip"\; filenames="HSA00000209917400013.png,HSA00000209917400021.png"\; state="\{\{state\}\}">\; </figure-callout>}}_{\mathrm{<span\; class="notranslate">\; </span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; -</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; \&alpha;</span>}\\ {\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="S\; hip"\; filenames="HSA00000209917400011.png,HSA00000209917400012.png"\; state="\{\{state\}\}">S</figure-callout></span><figure-callout\; id="2"\; label="S\; hip"\; filenames="HSA00000209917400011.png,HSA00000209917400012.png"\; state="\{\{state\}\}">\; </figure-callout>}}_{\mathrm{<span\; class="notranslate">\; </span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; \&beta;</span>}\\ {\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="S\; knee"\; filenames="HSA00000209917400013.png,HSA00000209917400021.png"\; state="\{\{state\}\}">S</figure-callout></span><figure-callout\; id="1"\; label="S\; knee"\; filenames="HSA00000209917400013.png,HSA00000209917400021.png"\; state="\{\{state\}\}">\; </figure-callout>}}_{\mathrm{<span\; class="notranslate">\; </span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; \&alpha;</span>}\\ {\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="S\; knee"\; filenames="HSA00000209917400011.png,HSA00000209917400012.png"\; state="\{\{state\}\}">S</figure-callout></span><figure-callout\; id="2"\; label="S\; knee"\; filenames="HSA00000209917400011.png,HSA00000209917400012.png"\; state="\{\{state\}\}">\; </figure-callout>}}_{\mathrm{<span\; class="notranslate">\; </span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; \&beta;</span>}\end{array}\right.<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1,3,5,7,9</span>}<span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span>$

(3)

$\left\{\begin{array}{c}{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="S\; hip"\; filenames="HSA00000209917400013.png,HSA00000209917400021.png"\; state="\{\{state\}\}">S</figure-callout></span><figure-callout\; id="1"\; label="S\; hip"\; filenames="HSA00000209917400013.png,HSA00000209917400021.png"\; state="\{\{state\}\}">\; </figure-callout>}}_{\mathrm{<span\; class="notranslate">\; </span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; \&beta;</span>}\\ {\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="S\; hip"\; filenames="HSA00000209917400011.png,HSA00000209917400012.png"\; state="\{\{state\}\}">S</figure-callout></span><figure-callout\; id="2"\; label="S\; hip"\; filenames="HSA00000209917400011.png,HSA00000209917400012.png"\; state="\{\{state\}\}">\; </figure-callout>}}_{\mathrm{<span\; class="notranslate">\; </span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; -</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; \&alpha;</span>}\\ {\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="S\; knee"\; filenames="HSA00000209917400013.png,HSA00000209917400021.png"\; state="\{\{state\}\}">S</figure-callout></span><figure-callout\; id="1"\; label="S\; knee"\; filenames="HSA00000209917400013.png,HSA00000209917400021.png"\; state="\{\{state\}\}">\; </figure-callout>}}_{\mathrm{<span\; class="notranslate">\; </span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; \&beta;</span>}\\ {\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="S\; knee"\; filenames="HSA00000209917400011.png,HSA00000209917400012.png"\; state="\{\{state\}\}">S</figure-callout></span><figure-callout\; id="2"\; label="S\; knee"\; filenames="HSA00000209917400011.png,HSA00000209917400012.png"\; state="\{\{state\}\}">\; </figure-callout>}}_{\mathrm{<span\; class="notranslate">\; </span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; \&alpha;</span>}\end{array}\right.<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 6</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 8,10</span>}<span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span>$

(4)

The parameters involved in the gait design method of the present invention are mainly time variables T, T ₁ , T ₂ , T ₃ and two angles of θ and α. During the operation, it is necessary to adjust the parameters correspondingly according to the actual results of the robot. .

The walking method of the powered biped robot in the present invention is based on the principle of passive walking, and supplements the potential energy of the center of mass of the system by first actively extending and then actively shortening the supporting legs during one-step walking. This method eliminates the impact on walking stability caused by adding energy when the swinging leg collides with the ground, and enables the robot to achieve high walking stability, and it only requires open-loop control, which is simple to implement and has a very small amount of calculation. Therefore, it is suitable for occasions that require high real-time performance. In addition, it concentrates the control on the supporting legs instead of controlling the swinging legs, which reduces the variables of control, and this control method can make the energy added to the system a fixed value, thus having a progressive energy balance Advantage.

Description of drawings

Figure 1 is a schematic diagram of the robot model.

Fig. 2 is a schematic diagram of the energy conversion of walking, which shows the principle of one-step walking process and the energy added by the supporting legs and the energy lost by swinging and colliding.

Fig. 3 is a mechanical diagram of the robot and a schematic diagram of four key motor rotation angles, wherein Fig. 3 (a) shows a side view of the robot, and (b) shows a front view of the robot.

Fig. 4 is a schematic diagram of gait key frames, A is the first key frame, B is the second key frame, C is the third key frame, D is the fourth key frame, and E is the fifth key frame.

Fig. 5 is the change curve of two key angles, wherein the realization shows the trajectory of θ, and the dotted line shows the trajectory of α.

Fig. 6 is a flowchart for realizing the power walking method.

Detailed ways

Fig. 3 is the schematic diagram of the structure of the robot used in the present invention and four joint angles, the realization of biped robot walking method of the present invention needs five keyframes, as shown in Fig. 4, in these five keyframes, the first The first key frame determines the posture at the initial moment of a step, and the fifth key frame determines the posture at the moment of a step collision, and also determines the size of the stride. The second keyframe to the third keyframe is the process of extending the supporting leg, which determines the amount of energy added. The third keyframe to the fourth keyframe is the process of shortening the supporting leg to ensure that the two legs are equal in length when they collide. , and keep the support leg shortened when it becomes a swing leg in the next step. This process is a process of energy loss. In order to ensure that the total energy of the system increases before the collision, the energy added by the extension of the support leg must be greater than that of the support The energy lost by the shortening of the legs requires strict control of the relative positions of the second, third and fourth keyframes. As long as the extended position is closer to the vertical position than the shortened position, the overall system performance can be guaranteed. Energy is increased.

The above parameters T, T ₁ , T ₂ , T ₃ and θ ₀ , α ₀ take values in the following manner.

Given a set of initial values based on experience, first determine the three time values of T ₁ , T ₂ , and T ₃ . This is to ensure that the extended position of the supporting leg is closer to the vertical position than the shortened position. For convenience, generally Take the T ₂ key frame as the vertical position, that is, T ₂ = T/2, T ₁ and T ₃ only need to satisfy T ₂ -T ₁ <T ₃ -T ₂ , and then operate the actual robot, keeping T , T ₁ , T ₂ , and T ₃ remain unchanged, if the robot starts walking at t=0 and then falls forward, it means that α ₀ is too large, and the added energy is too large, so α ₀ is reduced by 1 °, repeat this until the robot can walk; if the robot starts walking at t=0 and then falls backwards, it means that α ₀ is too small, and the added energy is too small, so add 1° to α ₀ , and so on Repeat until the robot is able to walk.

If adjusting α ₀ cannot make the robot walk, you need to adjust other parameters. If the robot falls forward, you can also reduce T ₁ (to make the supporting leg straighter farther away from the vertical position) or reduce T3 (to make the supporting leg shortened). The position is closer to the vertical position) to reduce the energy added by the system; if the robot falls backwards, the energy added by the system can also be increased by increasing _T1 or _T3 . Finally, by reselecting θ ₀ and T, and then adjusting the parameters according to the above steps, the robot can realize stable walking with different amplitudes and different periods on flat ground.

During walking, the values of the key angles θ and α shown in Figure 1 are calculated according to the above-mentioned key frames, and its motion trajectory is the connection curve between the key frames. In order to ensure that the angular velocity of each joint of the robot is continuous, a smooth curve is used to connect the key frames, that is, the first derivative of the curve is continuous. Fig. 5 shows a connection method using straight lines and trigonometric function curves, but the scope of the present invention is not limited to this connection method.

The realization of the walking method of the biped robot in the present invention is an iterative process of the following two steps: (1) the upper computer control walking stage; (2) the manual parameter adjustment stage. In stage (1), firstly, a set of θ ₀ =50°, α ₀ =25°, T=0.4s, T ₁ =0.1s, T ₂ =0.2s, T ₃ =0.36s is manually set, and after determining After setting these parameter values, the computer as the upper computer uses a smooth curve to connect the key frames, calculates the θ and α values at each moment, and then converts them into the four motor angle values of the robot through formulas (3) and (4) , and send it to each motor on the robot to make it rotate according to the specified trajectory to realize the walking action. When n is an odd number, _Ship1 and S _knee1 in Figure 3 are the joint angles of the supporting legs, _Ship2 and S _knee2 are the joint angles of the swinging legs, and when n is an even number, _Ship2 and S _knee2 are the joint angles of the supporting legs, and _Ship1 , S _knee1 is the swing leg joint angle. After the robot is placed in the air and starts to move, put it on the ground to walk and observe the walking effect. If the robot falls down and cannot walk, then turn to stage (2). Under the parameters set above, the robot will fall backwards when walking on flat ground.

Stage (2) is a manual adjustment stage. In this stage, people manually adjust the parameter settings in the host computer according to the walking effect of the robot in stage (1). The adjustment method is as follows: firstly, two key angles are adjusted, Among them, α ₀ determines the amount of supplementary energy for walking. If the robot falls forward after walking in stage (1), α ₀ can be reduced by 1°; if the robot falls backward, α ₀ can be added by 1°. If the robot can walk, it means that the walking condition has been met, and α ₀ can be fine-tuned to optimize the collision time. If adjusting α ₀ can not make the robot walk, you need to adjust other parameters. If the robot falls forward, you can also reduce the energy supplied by the system by reducing T ₁ or T ₃ ; if the robot falls backward, you can also use Increasing T ₁ or T ₃ increases the energy supplied by the system. Finally, by reselecting θ ₀ and T, and then adjusting the parameters according to the above steps, the robot can walk stably on the flat ground. Under the parameters set in stage (1), the robot will start to fall forward. By reducing α ₀ , when α ₀ is reduced to 16°, the robot can walk stably on flat ground.

Claims (5)

1. A kind of power type biped robot walking method is characterized in that, contains following steps successively: Step (1), constructing a biped robot, the steps are as follows: Step (1.1), establish the connection between the torso and the first thigh and the second thigh: The torso is respectively fixedly connected with the body of the first hip joint motor and the second hip joint motor placed coaxially on the left and right sides, and the rotation output shaft of the first hip joint motor is connected with the first thigh, and the second hip joint motor The rotation output shaft of the joint motor is connected with the second thigh, Step (1.2), establishing the connection between the first thigh and the first calf, the second thigh and the second calf: The end of the first thigh is fixedly connected to the body of the first knee motor, and the rotation output shaft of the first knee motor is connected to the first lower leg, The end of the second thigh is fixedly connected to the body of the second knee motor, and the rotation output shaft of the second knee motor is connected to the second lower leg, In step (1.3), the four motors described in step (1.1) and step (1.2) all adopt servo motors, and the rotations of the first hip joint motor and the second hip joint motor are represented by _Ship1 and _Ship2 respectively Angle, respectively use S _knee1 , S _knee2 to represent the rotation angles of the first knee joint motor and the second knee joint motor, and use a host computer to control the four motors, wherein: The S _hip1 is the angle between the first thigh and the vertical direction of the trunk, when the end of the first thigh is located in front of the trunk, S _hip1 >0, and when the end of the first thigh is located behind the torso, S _hip1 <0, The S _hip2 is the angle between the second thigh and the vertical direction of the trunk, S _hip2 >0 when the end of the second thigh is located in front of the torso, and S _hip2 <0 when it is located behind the torso, The S _knee1 is the angle between the first calf and the first thigh, when the first calf is bent backward relative to the first thigh, S _knee1 >0, and when the two are on the same straight line, S _knee1 =0, The S _knee2 is the angle between the second calf and the second thigh, when the second calf is bent backward relative to the second thigh, S _knee2 >0, and when the two are on the same straight line, S _knee2 =0, Step (1.4), the control signal input end of each motor described in the step (1.1), step (1.2), step (1.3) is connected with the control signal output end of a host computer respectively; Step (2), set a gait cycle T in the host computer, the gait cycle T refers to the time elapsed from the moment of one step start to the impact of the swinging leg, wherein, the start moment t=0 means regarded as The moment when the swing leg of the second thigh leaves the ground, the collision time t=T refers to the moment when the swing leg collides with the ground, that is, the moment when one gait cycle ends and the next gait cycle begins. At this moment, it is regarded as The supporting leg of the second thigh becomes a swinging leg, and the previous swinging leg becomes a supporting leg, and the walking parameters include: θ, α, β, the unit is angle, wherein: θ is the angle between the equivalent support leg (4) and the equivalent swing leg (7), the equivalent support leg (4) is connected from the top of the support leg thigh (2) to the support leg calf (3 ) at the end represents that the equivalent swing leg (7) is represented by the connection line from the top of the swing leg thigh (5) to the end of the swing leg calf (6), α is the angle between the supporting leg thigh (2) and the equivalent supporting leg (4), which determines the length of the equivalent supporting leg (4), β is the angle between the swing leg thigh (5) and the equivalent swing leg (7), which determines the length of the equivalent swing leg (7). During walking, the swing leg remains unchanged, so β is a fixed constant, When the equivalent swing leg (7) is located before the equivalent support leg (4), θ>0, and after that, θ<0, When the knee joint of the supporting leg is bent, α>0, and when the knee joint of the supporting leg is straightened, α=0, When the knee joint of the swing leg is bent, β>0, and when the knee joint of the swing leg is straight, β=0; Step (3), in the host computer, in the one gait cycle, set five key frames, two key joint angles θ and α; The first key frame, at t=0, determines the initial posture of the robot for one step, where θ=-θ ₀ , θ ₀ is a non-negative constant, representing the clamp between the two equivalent legs at t=0 Angle determines the size of the stride; α=α ₀ , α ₀ is a non-negative constant, which means that t=0 is the bending angle of the supporting leg thigh relative to the equivalent supporting leg, The second key frame is located at t=T ₁ , where: α=α ₀ , which is the same as α in the first key frame, means that the length of the equivalent support leg remains constant between the first key frame and the second key frame change, this keyframe represents the starting point of the support leg straightening, The third key frame is located at t=T ₂ , where: α=0, which means that the knee joint of the supporting leg is straightened, and the center of mass of the hip rises to the highest point. From the second key frame to the third key frame is the extension of the supporting leg The long process is the process of adding energy to the system, and since the knee joint of the swinging leg is still in a bent state at this time, it can avoid the collision between the lower leg of the swinging leg and the ground during walking. The fourth key frame, at t=T ₃ , wherein: α=α ₀ , indicates that the knee joint of the supporting leg is bent back to the original state, and the process from the third key frame to the fourth key frame is the shortening of the supporting leg, which is In order to ensure that the legs are of equal length during the collision and remain bent when the supporting leg becomes a swinging leg in the next step, The fifth key frame, at t=T, determines the pose of the robot at the time of collision, where: θ=θ ₀ represents the angle between two equivalent legs at the time of collision; α=α ₀ , from the fourth key frame to the first Five keyframes, the supporting leg remains unchanged, after colliding with the ground, the supporting leg becomes a swinging leg, and the swinging leg becomes a supporting leg, The parameters T, T ₁ , T ₂ , T ₃ and θ ₀ , α ₀ take values in the following manner: First give a T, and take T ₂ =T/2, then determine T ₁ and T ₃ , in order to ensure that the position of the support leg is extended closer to the vertical position than the shortened position, T ₁ and T ₃ must satisfy T ₂ -T ₁ <T ₃ -T ₂ , θ ₀ , α ₀ are given randomly under the conditions of 30°<θ ₀ <65°, 10<α ₀ <50; Step (4), the host computer controls the walking of the robot according to the following steps in turn: Step (4.1), set the cycle T of each step, three key times T ₁ , T ₂ , T ₃ , the calculation time is t, and t starts from 0, and the host computer calculates the time of each t time according to the following formula θ, the value of α (1)

(2) According to the above formula, two curves of θ=f _θ (t), α=f _α (t) can be obtained, which are respectively the curves of the included angle of the hip and the bending angle of the knee joint of the supporting leg, and the first derivatives of variables θ and α with respect to t continuous, Step (4.2), in the above-mentioned upper computer, when the number of steps is n, n is the integer part of the division result of t and T, in the above-mentioned upper computer, the above-mentioned walking parameters θ, α and the bending angle β of the swinging leg knee joint Under the value of , since the knee joint remains unchanged after the supporting leg becomes a swing leg, β=α ₀ , the rotation angle of each motor of the biped robot, (3)

(4) When n is an odd number, _Ship1 and S _knee1 are the angles of the hip joint and the knee joint of the supporting leg respectively, and _Ship2 and S _knee2 are the angles of the hip joint and the knee joint of the swing leg respectively, and when n is an even number, _Ship2 and S _knee2 are the angles of the hip joint and knee joint of the supporting leg respectively, and _Ship1 and S _knee1 are the angles of the hip joint and knee joint of the swing leg respectively, that is to say, when n changes from an odd number to the odd number plus When there is an even number formed by 1, the supporting leg as the first thigh and the swinging leg as the second thigh are interchanged after one step of walking. 2. A kind of power type biped robot walking method according to claim 1, is characterized in that, if the operator finds the following situations, be dealt with respectively: The robot falls forward after starting to walk at t=0, indicating that α ₀ is too large, and the added energy is too large, so α ₀ is subtracted by 1°, and so on, until the robot can walk, The robot starts to walk at t=0 and then falls backwards, indicating that α ₀ is too small and the added energy is too small, so add 1° to α ₀ and repeat this until the robot can walk. 3. A walking method for a powered biped robot according to claim 1, wherein if the operator finds that no matter how he adjusts α ₀ , the robot cannot walk stably, then the following situations shall be dealt with separately: The robot starts to walk at t=0 and then falls forward, indicating that the added energy is too large. By reducing T ₁ , the supporting leg is farther away from the vertical position when it is straightened, or by reducing T ₃ , so that When the supporting legs are shortened, the position is closer to the vertical position, so as to reduce the energy supplied by the system, and so on, until the robot can walk, The robot starts to walk at t=0 and then falls backwards, indicating that the added energy is too small, then increase T ₁ to make the supporting legs closer to the vertical position when straightened, or increase T ₃ to make The position when the supporting legs are shortened is farther away from the vertical position, to reduce the energy added by the system, and so on, until the robot can walk. 4. a kind of power type bipedal robot walking method according to claim 1 is characterized in that, at t=T, if the operator finds the following situations, be dealt with respectively: The body of the robot leans forward, indicating that the time of collision is too early, so _α0 is subtracted by 0.2°, and so on, until the body of the robot is vertical when it collides. The body of the robot leans backward, indicating that the collision time is too late, so α ₀ is added by 0.2°, and so on, until the body of the robot is vertical when the collision occurs. 5. A kind of power type biped robot walking method according to claim 1, it is characterized in that, if no matter how operator adjusts, can not make described robot walk stably, or can walk, but needs to change pace, then Increase or decrease θ ₀ by 10°, or increase or decrease T by 0.1s, repeat steps (2) and (3), so as to obtain different strides and gait cycles, and realize walking at different speeds.

Images