Based on the pneumatic muscle Biped robot joint

Describes a pneumatic muscle building Biped robot joint, the joint use of the flexible pneumatic muscle characteristics, you can effectively control the Biped robot quickly walking or running landing foot impact issues.

Detail gives the pneumatic muscle works as well as by its composition of the joint system hardware architecture.

At the same time describes the hardware and joint structures of control software system.

Biped robot compared to typical mobile robot in unstructured environments with better mobility, thus causing researchers.

Control robot for fast travel speed and running gait is still a Biped robot in challenging areas. Robot walking or running fast, swing feet at the moment of landing will have a greater impact, this forces the landing foot rally or zmp (zeromoment point) produce larger jump, resulting in a robot stability margin decrease and fell. This phenomenon is called impact effect, it is restricting the Biped robot improves walking speed and running.

Pneumatic muscle is in recent years developed a new type of drive, McKibben-type pneumatic muscle is one of the most widely used one.

It has a soft, power/weight ratio, in force, the length property with similar human muscle, and other advantages. Because it has the advantage of compliance, application controlled pneumatic muscle as the drive can effectively address the Biped robot floor foot impact issues. Therefore, the pneumatic muscle as a Biped robot drive has good prospects. However, artificial muscles highly nonlinear characteristics. And accompanied by hysteresis phenomena, makes it difficult to build the home front and control. At present, based on the pneumatic muscle of study on Biped robot has just started, only a few Biped robot projects in this study. This article uses MeKibben pneumatic muscle to build similar biological antagonistic joints of single-degree-of artificial joints. This system hardware section includes pneumatic driven subsystems, sensor subsystem and control subsystems. This hardware system built the software system that enables this artificial joint trajectory tracking control. Based on this work can be further studied and resolved pneumatic artificial muscles and joints modeling and control issues, to design and build a drive based on pneumatic muscle Biped robot.

1 pneumatic muscle and joint systems hardware and software design

1.1 pneumatic muscle

McKibben pneumatic muscle is the invention by the United States physician joseph.l.mekibben and use the name of a flexible pneumatic drives.

McKibben pneumatic artificial muscles of the body consists of outer knitted NET and inner elastic rubber tube. Its structure as shown in Figure 1.

　　

Figure l for muscle chart, where Pi = input pressure, its size by the controller under actual working conditions.

Only when the input-side pressure increases, the inner rubber tube expansion, as the outer braid rigidity is large, almost no elongation, limit the muscle can only radial deformation (diameter larger, length shortening), produce axial contractility; and when input pressure decreases, the Pi leads to artificial muscle elongation (relaxation), muscle stiffness and driving force is reduced accordingly. Muscle stiffness can control rubber tube pressure to achieve, this muscle stiffness characteristics from a variable may be equivalent to a variable stiffness of the spring.

1.2 sdof joint system

Because the pneumatic muscle can only provide one-way driving force, so the two muscle to similar biological antagonistic muscle ways constitute a confrontational rotary joint for operations arm of force closed.

This article uses McKibben pneumatic muscle as drive a sdof system of antagonistic joints. This system is driven by the pressure of hardware subsystems, sensor subsystem and control subsystems. System chart as shown in Figure 2.

　　

1.2.1 pressure-driven subsystems

Pneumatic driven subsystems by air, pressure servo McKibben proportional valve, pneumatic muscle and body parts.

Provided by the air pressure 0.6 ~ 0.9MPa compressed gases, compressed gases catheterization by-passes away servo proportional valve pneumatic muscle. Each muscle is a servo proportional valve is connected with a vented valves and an inlet valve. By controlling the servo valve on the voltage to control muscle in gas pressure. Pressure of pneumatic muscle tension and exported part of driving mechanism, through the joint rotation on muscle pressure control to achieve the desired trajectory tracking joint torque. The system adopts pneumatic muscle of McKibben for FESTO's type, its work MAS-20-300N pressure range 0 ~ O.6MPa, maximum operating frequency for the biggest contraction 3Hz, for muscle length of 25%, when the theory forcing to O.6MPa 300N, repeat accuracy of less than 1%. Pressure servo control-proportional valve accept incoming voltage input and by adjusting the air intake valve and the valve control air pressure inside the muscle. This system incorporates the SMC company "∏" ITVOO5C-2ML pressure proportional valve. This valve input range is 0 to 5VDC, output 0.001 ~ 0.9MPa pressure between.

1.2.2 sensor subsystem

Sensor subsystem consists of force sensor and linear displacement sensor.

By linear displacement sensor can measure out the amount of muscle contraction, according to the contraction of the muscles and joints can be used for tracking control model. Force sensor for measuring muscle tension, according to this tension and joint torque of the linear relationship between joint torque can be calculated in order to complete the joint closed-loop control of a servo. This system uses the force transducer is the CASC 7Ol BK-2F type high precision s-force/Weighing sensor. The maximum measured force range 80kg, precision 0.05%. Output after TS-2-amplifier, the output voltage range is-5V to + 5V. Linear displacement sensor using WDL type straight sliding-conductive plastic potentiometer.

1.2.3 control subsystem

Control subsystem consists of industrial control computer (IPC), A/D sampling card, D/A conversion card.

Software control system operation in industrial control computer, and d/a converter will convert the digital control of analog. This analog to control the pressure in the output of the Servo-pneumatic proportional valve. A/D converter collection tension sensor and linear displacement sensor data and made available to industrial control computer can be used by software processing of digital signals. This system uses the d/A converter for PCL-726-6 channel analog output card. It offers six 12-bit double-buffered analog output channels to meet the needs of the muscle servo control. MD acquisition card use a 12-bit 32 PCL-813B-channel a/D card that provides 32-channel isolated a DC voltage measurement accuracy to meet the system requirements.

1.3 software system

Pneumatic muscle driven antagonistic joint has 2 muscle pressure is a system of control variables.

Due to the system via 2 free variables control the movement of a degree of freedom and therefore constitutes a redundant-drive system. You can prove that this system is available on the joint torque and joint stiffness for independent control. Among them, former and 2 muscle pressure difference, the latter and muscle pressure and relevant. Through the joint torque control can implement precise joint tracking and through joint stiffness control can lower standing foot shock and system power consumption.

This article in industrial control computer produced pneumatic muscle joint system of control software and operation interface.

This software system can achieve muscle model parameter set

And stable closed-loop control and sensor returns the value of real-time display and recording capabilities.

The main part of the software includes trajectory planning module and pressure control module. Trajectory planning module implements the top joint trajectory planning and realization of joint model calculation of expected joint track joint torque is required. Pressure control module implementing the underlying calculation, the amount of its input to the expectations of the upper plan is joint torque. Pressure calculation module under pneumatic muscle model calculation of actual control muscle movement in the desired pressure, it holds out the hope that muscle pressure value. Smart PID control algorithm module according to the expected pressure value and A/D sampling actual muscle pressure data intelligently PID control, thereby achieving closed-loop control. Its output through numerical and D/A link to output to the actual hardware system control voltage. Software system block diagram shown in Figure 3.

　　

2 experimental and system application

2.1 pneumatic joints system model

In order to realize on this build of pneumatic muscle driven antagonistic joint precise servo control. first of all to be modeled pneumatic muscle.

MeKibben pneumatic muscle with time-varying characteristics of nonlinear, and work with a hysteresis phenomena, which are difficult to be modeling and control. Most existing research for McKibben pneumatic muscle modeling using Chou and Hannaford is based on the principle of virtual work gives a theoretical model. This model gives the ideal muscle effort estimate, however this theoretical model directly to the actual control and will not be able to get good results. This study used the Reynolds and dynamic model of ternary muscles will approximate pneumatic muscles as nonlinear damping factor, nonlinear spring factor and non-linear shrinkage force factor of dynamic systems in parallel, the model equation is:

　　

Where x is the length of the muscle contraction, when muscles completely done at x = 0.

K0, K1 factor for spring, B0, B1 coefficient as damping factor coefficient, F0, F1 to shrink force factor. For this article adopts pneumatic muscle, through this experiment on the system, you can type (1) in a ternary muscle to accurately estimate the model parameters. When the muscle pressure p value in between 200kPa ~ 650kPa, experimentation is the resulting model coefficients can be satisfied with the approximate effect. Application of this three-way muscles, described software system on joint was closed-loop tracking control, the control accuracy better than traditional models.

2.2 system application

In this article use flexible drive built McKibben pneumatic muscle as a driver of robot single degree of freedom on the antagonistic joint system, McKibben-type pneumatic muscle modeling and control for further research.

The system of the ternary muscle model parameters for an accurate estimate, based on the use of this implementation of the control software system can achieve a closed-loop tracking control of joints.

In further work will be based on the existing platform and closed-loop control method to focus on two issues.

First research joint stiffness and controllable tracking control. This redundancy system for joint realization of rigid metric optimization purposes, the robot by making better use of their joint of passive dynamic features to reduce energy loss. The second study in shock over control joint stiffness and reduce shocks, which in turn is driven by pneumatic muscle structures Biped robot provides theoretical preparation.