機(jī)械外文文獻(xiàn)翻譯--電動(dòng)下肢輔助外骨骼的開(kāi)發(fā)與分析【中文4480字】【PDF+中文WORD】,中文4480字,PDF+中文WORD,機(jī)械,外文,文獻(xiàn),翻譯,電動(dòng),下肢,輔助,骨骼,開(kāi)發(fā),分析,中文,4480,PDF,WORD Corresponding author:Wei Dong E-mail: Journal of Bionic Engineering 14(2017)272283 Development and Analysis of an Electrically Actuated Lower Extremity Assistive Exoskeleton Yi Long1,Zhijiang Du1,Chaofeng Chen1,Weidong Wang1,Long He2,Xiwang Mao2,Guoqiang Xu2,Guangyu Zhao2,Xiaoqi Li2,Wei Dong1 1.State Key Laboratory of Robotics and System,Harbin Institute of Technology(HIT),Harbin 150001,China 2.Weapon Equipment Research Institute,China South Industries Group Corporation,Beijing 102202,China Abstract An electrically actuated lower extremity exoskeleton is developed,in which only the knee joint is actuated actively while other joints linked by elastic elements are actuated passively.This paper describes the critical design criteria and presents the process of design and calculation of the actuation system.A flexible physical Human-Robot-Interaction(pHRI)measurement device is designed and applied to detect the human movement,which comprises two force sensors and two gasbags attached to the inner surface of the connection cuff.An online adaptive pHRI minimization control strategy is proposed and implemented to drive the robotic exoskeleton system to follow the motion trajectory of human limb.The measured pHRI information is fused by the Variance Weighted Average(VWA)method.The Mean Square Values(MSV)of pHRI and control torque are utilized to evaluate the performance of the exoskeleton.To improve the comfort level and reduce energy consumption,the gravity com-pensation is taken into consideration when the control law is designed.Finally,practical experiments are performed on healthy users.Experimental results show that the proposed system can assist people to walk and the outlined control strategy is valid and effective.Keywords:exoskeleton,pHRI measurement,data fusion,pHRI minimization,adaptive control Copyright 2017,Jilin University.Published by Elsevier Limited and Science Press.All rights reserved.doi:10.1016/S1672-6529(16)60397-9 1 Introduction A lower extremity exoskeleton is an automatic de-vice worn by human users to improve strength and en-durance.In recent decades,many advances and progress have been made in the development of wearable exo-skeletons.The Berkeley Lower Extremity Exoskeleton(BLEEX)was designed to assist people to carry heavy loads,which could walk at the speed of 0.9 ms1 while carrying 34 kg payload1.A single robot leg has four actuators which leads to the complexity and heaviness of the whole system2.The developed exoskeletons latter,i.e.,ExoHiker,ExoClimber and HULC,simplify the mechanical structure and reduce the number of active Degrees of Freedoms(DoFs).Those DoFs of joints with the highest power consumption during gait cycles should be actuated while the rest DoFs are passively connected with elastic elements to simplify the system3.Generally,the lower extremity exoskeleton leg is composed of serial or serial-parallel leg segment.A cam-mechanism is implemented in the design of ankle joint of the exo-skeleton4.A novel serial-parallel mechanism is de-signed in a lower extremity exoskeleton to augment load-carrying for walking5,6.However,parallel mechanism will increase the complexity of mechanism and decrease its portability.Compact serial-link mechanisms are utilized in many advanced exoskeletons,e.g.,Ekso7 and ReWalk8.The actuation system design and development is a significant aspect for the exoskeleton.There are several popular actuation system modes that have been applied in lower extremity exoskeleton,including hydraulic actuators,electrical motors,serial elastic actuators and pneumatic muscle actuators.Torque-mass ratio,velocity,range of motion and controllability are significant specifications when choosing actuators for exoskele-tons9.Hydraulic actuators and electrical motors are utilized frequently due to their high torque-mass rates10.Long et al.:Development and Analysis of an Electrically Actuated Lower Extremity Assistive Exoskeleton 273Comprehensively,electrical motors are suitable for exoskeletons due to its wide range of motion,ease of control and maintenance,and simplicity of the system.Normally,the electrical motors are placed parallel with the joint of mechanical legs,which causes increment of structure complexity.Different from the traditional electrical actuators,a novel electrical actuator consisting of DC motor,gear pair and ballscrew is developed,which has a more compact mechanical structure com-pared with the traditional actuator.pHRI-based human motion intent estimation is a critical step for the exoskeleton control.It is essential to get an accurate measurement of the pHRI for the robotic exoskeleton control and the assessment of the assistance grade11.The connection cuff is a widely-used device to fasten the human limb to the exoskeleton,which affects the wearing comfort and the walking performance.A flexible connection cuff is a suitable device to interact with the human user.A flexible sensor technique is de-veloped to measure the pHRI pressure,where the sen-sory system is composed of several optical-electronic sensors12.The system has at least six sensors on the surface of the cuff.The pHRI force can be measured by a strain gauge,where a circle sensor is utilized13.How-ever,the structure of the interaction cuff is complex and has multiple sensors.In this work,a flexible pHRI measurement device is designed and applied in the ro-botic exoskeleton control,which is composed of two gasbags and one force sensor connected to each gasbag.The gasbag can enlarge the interaction area and guar-antee that the pHRI can be measured easily.In addition,the usage of gasbag can increase the wearing comfort and adapt to different human users.The control strategy design is the core issue for the exoskeleton.Control strategies can be divided into three categories according to methods of estimating human motion intent,i.e.,approaches based on signals meas-ured from the human body,approaches based on inter-action force measurement and approaches based on signals measured from exoskeletons14.In order to guarantee the natural gait of human users,sensitivity amplification control,model-based control and hybrid assistive strategy are suitable for load-carrying task15.However,sensitivity amplification control is sensitive to external disturbances and model-based control is de-pendent on the accurate dynamic model.Compared with other assistive strategies,hybrid control strategy is ap-plied successfully in exoskeletons for load-carrying and shows an increased performance.The pHRI-based con-trol strategy is an effective approach for exoskeleton.To minimize the pHRI force,a RBF neural network was proposed to compensate the dynamic uncertainties,where there is no force sensor in the interaction cuff16.However,there are some sensors on the human limbs.The goal of human exoskeleton collaboration is to re-duce or eliminate the misalignment between the human user and the exoskeleton.In this work,we propose an online adaptive strategy to drive the robotic exoskeleton.With the discussion above,we highlight contributions of this paper:1)A lightweight and compact electrically actuated robotic exoskeleton system is designed and implemented for walking assistance.2)A pHRI meas-urement device is designed,which consists of a force sensor and a gasbag embedded in the connection cuff.3)A control strategy called model-free adaptive pHRI minimization is proposed to drive the exoskeleton to follow the human limbs.The remainder of this paper is organized as follows.The specific mechanical design of exoskeleton and the details of actuator are given in the second section.The sensory system is illustrated in the third section.In the fourth section,the adaptive pHRI minimization control strategy is proposed.Experiments are performed and experimental results are presented in the fifth section.Discussion and conclusion are presented in the final section.2 System architecture description 2.1 Mechanical structure Exoskeletons are anthropomorphic devices that work in parallel with the human body.In the design of lower extremity exoskeletons,the number of DoF is required to be close to the number of human lower limbs DoFs in order to achieve a comfortable walk-ing17.The lower extremity assistive exoskeleton has two mechanical segments attached to human limbs,in which there are seven DoFs on the thigh,the knee and the ankle joints in total.Fig.1 shows the architecture of the lower extremity exoskeleton,which includes three main components as below18:(1)The leg segment including a shank and a thigh is attached to the human user leg by the interaction cuffs.The knee joint is activated by an actuator while the hip joint is passive.The actuator of the knee joint consists of Journal of Bionic Engineering(2017)Vol.14 No.2 274 Fig.1 Mechanical structure of the exoskeleton.1:power supply,2:control system enclosure,3:carried load,4:springs connecting the backpack with the waist part,5:waist part,6:thigh segment,7:shank segment,8:wearable shoes which have pressure sensors,9:hip joint,10:actuation system consists of DC motor,gear pair and screw,11:connection cuff,12:optical encoder on knee joint,13:ankle joint,14:signal transfer for foot pressure.a DC motor and a ballscrew,which helps legs walk and supports the weight of system when the leg is in the stance phase.The leg segment is made of aluminum and its weight is about 4 kg.The leg segment is attached to the waist part of the exoskeleton through a connection mechanism.All wires of the control module are inte-grated into the mechanism through the wiring slot con-necting to the central controller.The lower extremity exoskeleton is suitable for the user whose height is in the range of 168 cm 188 cm,where the length of the ro-botic thigh can be adjusted in the range of 430 mm 470 mm while the robotic shank can be ad-justed in the range of 470 mm 510 mm.(2)The trunk,which is used as a platform to inte-grate the control system,the power supplies and the payload to be carried.The trunk of the exoskeleton consists of a connector to the leg,an adjustment mechanism for the waist width,a backpack and an elas-tic element connecting the backpack with the robotic waist.The elastic element with enough stiffness is util-ized to support the weight of the mechanism.The length of the robotic waist can be adjusted in the range of 340 mm 360 mm.The backpack will be connected to an ergonomic mechanism which is tied to human torso to carry the control enclosure,the power unit and other equipments.(3)The wearable shoes connected with the shanks,each of which contains three pressure sensors to collect ground reaction force.The wearable shoe is made of rubber to enhance the flexibility.The thin upper sole is fixed by bolts.Three pressure sensors or foot switches are placed into the lower sole and all wires are integrated together.The wearable shoe is connected with the leg segment through the ankle joint,which is designed as a curved ball bearing with three DoFs.To make the weight with a flexible support,some elastic elements,e.g.,flexible rubber plates,are placed into the connecting part between the ankle joint and the shoe.2.2 Design and performance analysis of actuator Since the joint space is limited,it is impossible to use larger and heavier actuators to actuate each DoF of the lower extremity exoskeleton.In our work,the knee joint is actuated while other joints are supported by unpowered elastic elements,e.g.,the springs to support the load from the humans leg joints.The selection of the actuator should be determined according to appropriate specifications because the in-crease in required torque will cause the increase in weight19.Electrical actuators should be able to actuate the whole device including the weight of exoskeleton system,the weight and inertial torques of legs.Drillis stated that the maximum angular acceleration at the instant of kicking a ball can be 453 rads2 and the mass moment of inertial of lower legs can be 0.35 Nms220.In our work,the maximum angular acceleration is,the mass moment of inertial is I,the carrying load is m,and the inclination angle of weight is,as shown in Fig.2.Then the joint moment of knee can be expressed as:coscos.mg lI=+(1)For the developed exoskeleton,the carrying load is 40 kg,and the maximum =54.4 rads2,then the cal-culated moment of knee joint is about 30 Nm.We define the angular velocity of knee joint is =57 rads1.The ballscrew should be chosen according to architecture size of the joint and motor parameters.There are some equations for the relationship between driving torque and thrust force F,which are shown as follows:()()22221212,sin(),arctan,arctan,aFeacbdefcadb=+=(2)Long et al.:Development and Analysis of an Electrically Actuated Lower Extremity Assistive Exoskeleton 275 Weight of loadMoment armlGGear pairMotorEncoderSensorDesigned electrical actuatorBallscrewConnecting pinInclination angle of loadThrust forceFaJoint momentGround support forcebdefaFaFg Fig.2 Force analysis of exoskeleton carrying load in the stance phase.where a,b,c and d are mechanical parameters of the leg,and f is the length of ballscrew.Based on Eq.(2),the thrust force F can be calculated.Then the moment of ballscrew can be determined by the following formula:()/(2),aMFp=(3)where M is the input moment to the ballscrew,p is the thread pitch,and is the mechanical efficiency.Then input rotation speed to the ballscrew can be expressed as:/.snfp=(4)Defining the reduction ratio of gear pair transmis-sion as,then the torque and the rotation speed of DC motor can be determined by:/,mmsMMnn=(5)where Mm and nm are the driving torque and the rotation speed of DC motor,respectively.The actual structure of the electrical actuator is shown in Fig.2.This prototype transfers the rotation movement of the motor to the lin-ear movement by the transmission mechanism,and at the same time generates the supporting force to drive the rotation movement of the mechanical joint.2.3 Control architecture of the exoskeleton The developed robot control system architecture mainly consists of the embedded controller,the CAN bus(Controller Area Network)and the sensing system Force sensor with gasbagPowerCANARMCANCANWearable shoesOptical encoderBallscrewDriveMotorActuation systemExoskeletonPressure sensor and foot switch Fig.3 The control architecture of the robotic exoskeleton.The control architecture consists of central control parts,e.g.,the PC and the PMAC,sensory system,e.g.,the pressure sensor,the force sensor and the optical encoder and the actuation system,e.g.,DC motor,drive(Copley)and ballscrew.as shown in Fig.3.The embedded core controller is based on ARM-Cortex-A9 and dual core processor of Freescale.The core controller supports the Linux 3.0 plus QT 4.8 and Android 4.2.All data processing and control commands computation are conducted on this embedded board.The sensing part consists of the force switches,the foot pressure sensors,the force sensors and the optical encoders.The data collected by the force sensor and the foot pressure sensor are processed by the low-pass filter.The communication between the high-level controller and the low-level actuation system or the sensing system is connected by the CAN bus.The CAN bus has high data transfer rate of 1 Mbits1 to achieve real-time performance.The information of Journal of Bionic Engineering(2017)Vol.14 No.2 276 sensors is mounted on one channel CAN bus and the control command is transferred to the actuation system by another channel CAN bus.The control software consists of three parts,i.e.,the data acquisition module,the CAN bus module and the control module.Module programming can reduce complexity of the central con-trol program and achieve a high speed control.All of the data collection is completed in the data acquisition module and the control module produces the control commands to the actuation system to drive the exo-skeleton.The CAN bus module consists of the relevant files about CAN bus communication protocol.There are some safety precautions implemented on the software program,e.g.,the commanded motor currents will be limited using a saturation function to protect the system.3 Sensor system The sensor system is the outermost layer of the control prototype and consists of three kinds of sensors,i.e.,the digital and analogue pressure sensor embedded in the insole,the force sensor with gasbag for pHRI measurement and the encoder to obtain the joint angular position,which is shown in Fig.4.3.1 Pressure sensor in the wearable shoe Two digital pressure sensors(foot switches)are placed in the wearable shoe to measure the interaction force between the exoskeleton system and the ground.As Fig.4 shows,there is one sensor placed on the con-nection area to measure the weight of the exoskeleton system.The selected digital sensor outputs binary values,i.e.,0 and 1,which outputs 1 when the human user presses upon the sensor.The pressure sensor used in the connection area,which can output the voltage signal proportional to the applied force.This kind of sensor has enough resolution to record the GRF changes during the stance period.In addition,this kind of sensor is small and thin enough to be embedded into the wearable shoes and to ensure the wearing comfort.The used pressure sensor has a sensing area of 4.52 cm2 and has a meas-uring range from 0 kg to 100 kg.3.2 Force sensor with gasbag Physical HRI measurement is a critical issue for exoskeleton control.Based on pHRI,the human motion intent can be obtained.The measurement range of the force sensor is from 0 kPa to 1 kPa and the hose has a CoinForce sensorWearable shoeGasbag based sensorOptical encoderAir pipelineForce sensorSignal output Fig.4 The sensor system of the exoskeleton,which is composed of the wearable shoe,the force sensor with gasbag and the optical sensor.diameter of 7.5 mm.The air pipeline is connected to the gasbag,the force sensor changes as the air pressure changes and the signal output wire is connected to the collection part.The force sensor signals can be collected and the range of output voltage is 0 V 10 V and the sensitivity is 2.0 mvV1.The force sensor is convenient to measure the interaction signals between the human limb and the mechanical limb.The two gasbags are attached to the interaction cuff,each of which has an area of 100 mm by 70 mm.3.3 Encoder on the mechanical joint The angular position of the exoskeleton joint is collected by the optical encoder.This sensor is a kind of incremental encoder(HEDSS,German),whose resolu-tion is 1024 line for a rotation.The angular position is collected to illustrate the current mechanical position.The value of the optical encoder is recorded and trans-ferred to the central controller.The specific parameters of all applied sensors are shown in Table 1.4 Control strategy for exoskeleton When the user wears the robotic exoskeleton,the estimation of human motion intention is the first step for the control of the exoskeleton.In this project,we control this exoskeleton through minimizing the pHRI.There are two force sensors introduced previously to detect the pHRI,as shown in Fig.5.The interaction cuff on the thigh segment has two embedded gasbags,one of which is placed on the back of the interaction cuff and the other is placed on the front.Each gasbag is connected to a force sensor through a flexible hose,when the gasbag is oppressed then the force sensor will output the corre-sponding voltage signals.This measurement of pHRI has the following advantages,i.e.,1)it improves the Long et al.:Development and Analysis of an Electrically Actuated Lower Extremity Assistive Exoskeleton 277 Table 1 Specific parameters of the sensory system Parameters Range Resolution Sensitivity Accuracy Supply Pressure sensor 0 kg 100 kg 0.4%2.0 mvV1 5%5V 10%HRI sensor 0.1 kgcm2 10 kgcm2 0.5%2.0 mvV1 10%5V 10%Optical encoder 12 bits 0.5 5V 10%Notation:the symbol“”means that the current item does not exist.Interaction cuffGasbagActuationForce sensorSwitch Fig.5 T