CN108888478A - A kind of upper limb rehabilitation robot based on binocular vision - Google Patents

Abstract

The invention discloses a kind of upper limb rehabilitation robot based on binocular vision, the device includes mechanical framework, main control module, remote control module, motor module, limbs Information Collecting & Processing module, detection module, communication module, power module.It is characterized in that：The mechanical framework includes pedestal assembly, waist revolute joint, principal arm rotary joint, principal arm, the external fixing sleeve of principal arm, auxiliary rotary joint, auxiliary, the external fixing sleeve of auxiliary；The main control module is STM32 single-chip microcontroller；The motor module includes DC speed-reducing, motor driver；The limbs Information Collecting & Processing module includes binocular camera, computer；The detection module is surface myoelectric sensor；The communication module is SX1278 chip；Surface myoelectric sensor is communicated by SPI with main control module, and binocular camera is communicated by USB interface with computer；The invention discloses a kind of upper limb rehabilitation robots based on binocular vision.

Description

A kind of upper limb rehabilitation robot based on binocular vision

Technical field

The present invention relates to field of medical rehabilitation, in particular to a kind of upper limb rehabilitation robot based on binocular vision.

Background technique

In the 21st century, intelligent robot technology has played huge effect in field of industrial production, greatly improves Productivity；In field of medical rehabilitation, robot technology is also evolving, and has obtained the extensive concern of researcher；With me State steps into rapidly aging society, and the disease incidence of cerebral apoplexy, illness rate rise year by year, and 70% has different degrees of movement in patient Dysfunction, wherein hemiplegia is most common, most serious sequelae, and 55% ~ 75% has a upper limb obstacle in hemiplegic patient, and upper limb health Multiple robot can promote stroke hemiplegia upper extremity function to restore by rehabilitation training.

In traditional rehabilitation training, need one-to-one professional physical therapy personnel that patient is assisted to carry out periodic rehabilitation instruction Practice, be likely to reach certain rehabilitation efficacy, traditional rehabilitation training consumes great human resources and at high cost, and rehabilitation machines The appearance very good solution of device people these problems；Current healing robot in the market, information acquisition system are more multiple Miscellaneous, at high cost, acquired data error is big；The problem of biggish acquisition data error is caused to robot control accuracy difference, It is easy to cause patient's secondary injury；Therefore, it is necessary to design, a kind of structure is simple, and acquisition information is accurate, and control precision is high, rehabilitation The apparent upper limb rehabilitation robot of effect.

A kind of upper limb rehabilitation robot based on binocular vision of the invention can provide health for upper limb disorder patient Refreshment is practiced, and the rehabilitation of ipsilateral upper limb function is promoted；When rehabilitation training starts, by Physical Therapist according to patient's actual conditions, manually It assists patient to complete one group of calibration training, letter is moved using the arm space that binocular vision method obtains patient during the calibration process Breath improves the precision of acquisition information；The rehabilitation experience of Physical Therapist is applied in calibration process, the control of rehabilitation training is used in Information processed is more accurate, can obtain better rehabilitation efficacy；In the training process, upper limb rehabilitation robot is external solid by principal arm Fixed set and the external fixing sleeve of auxiliary realize the fixation of arm, and structure is simple；Using segmented intelligent control method, make training process Closer to the information of calibration, and small-scale self-adjusting can be carried out to training.

Therefore, the present invention has invented a kind of based on binocular vision aiming at the problem that existing upper limb rehabilitation robot technical deficiency The upper limb rehabilitation robot of feel.

Summary of the invention

The present invention is long for cerebral apoplexy upper limb hemiplegia patient's spontaneous recovery period after therapeutic treatment, and low effort is asked Topic and the not accurate enough problem of existing upper limb rehabilitation robot acquisition information, it is simple to propose a kind of structure, acquires information Accurately, control is accurate high, and rehabilitation efficacy significantly based on the upper limb rehabilitation robot of binocular vision, can assist patient to complete shoulder Portion's outreach/containing, the movement of shoulder vertical direction and ancon extension/flexion；It is completed using main control chip to entire robot Control, control information assist the calibration information of patient's training from Physical Therapist；Suffered from using binocular vision method acquisition process The affected limb motion information of person；In training process, pass through the external fixing sleeve of principal arm, the external fixing sleeve of auxiliary and end fixing piece The fixation for realizing arm, can train left and right arm；Using segmented intelligent control method, small model can be carried out to training The self-adjusting enclosed, makes training process closer to the information of calibration, and effect is more excellent.

In order to achieve the above objectives, technical solution provided by the invention is.

A kind of upper limb rehabilitation robot based on binocular vision, the device includes mechanical framework, main control module, long-range Control module, motor module, limbs Information Collecting & Processing module, detection module, communication module, power module；The machinery Framework includes pedestal assembly, waist revolute joint, principal arm rotary joint, principal arm, the external fixing sleeve of principal arm, auxiliary rotation pass Section, auxiliary forearm, auxiliary pull rod, auxiliary back seat, auxiliary, the external fixing sleeve of auxiliary, end fixing piece；The main control module is STM32F4；The motor module includes DC speed-reducing, motor driver；The limbs Information Collecting & Processing module Including binocular camera, computer；The detection module is surface myoelectric sensor；The wireless communication module is The LoRa of SX1278 chip is spread.

A kind of rehabilitation training of upper limbs method is devised according to the upper limb rehabilitation robot based on binocular vision is somebody's turn to do, with right arm shoulder For the outreach in portion is containing, recovery training method is：1. press the demarcating key of remote control module, start training action Calibration, Physical Therapist draw the palm of patient, and patient is helped to complete one group of slow shoulder abduction/containing training；2. calibrated Cheng Zhong by the acquisition of limbs Information Collecting & Processing module, processing calibration information, and this group information is saved at times；3. passing through Fixing sleeve（Part）On length-adjustable flexible-belt it is respectively that the postbrachium of the external fixing sleeve of principal arm and arm is fixed together, outside auxiliary It connects fixing sleeve to be fixed together with forearm, palm and end fixing piece are fixed on one piece, realize the servo-actuated of arm；4. pressing long-range Calibration information grouping is sent main control module by the training key of control module, Information Collecting & Processing module；5. main control module receives To after segment information, control waist turning motor rotation, motor moves the movement that mechanical arm executes rotation by gear band, until The whole grouping informations for having executed calibration information complete outreach/containing training of a shoulder；Wherein, main control chip STM32 Acceleration, deceleration, positive and negative rotation and the braking of motor are controlled by sending pwm signal；Information Collecting & Processing module real-time monitoring hand The motion information of arm, and compared with calibration group information, when there is large deviation, system alarm, mechanical arm restores to initial position； When surface myoelectric sensor detects that fatigue or spasm occurs in patient's finger, system alarm, mechanical arm restores to initial position.

Present invention employs binocular cameras as information acquisition device, and the angle different from two captures target information, lead to It crosses USB interface and acquired image information is transferred to computer；Different training actions, the position that binocular camera is put is not Together, convenient for the accurate image information of acquisition；For shoulder outreach with it is containing, binocular camera is disposed vertically in human body front hand On the right side of arm；Lift on the vertical direction of shoulder is fallen, binocular camera is disposed vertically in human body side；For bending for ancon It stretches, binocular camera level is put in the underface of limbs.

The present invention uses computer as message handler, and computer is shot by the image processing system of customization The shape information of target；During the calibration process, record is demarcated group information by computer, and according to sampling principle from the group information Extract one group of efficiently sampling information, fragmented storage, the reference information as rehabilitation；After training starts, computer passes through double The pictorial information that mesh camera captures obtains the shape information of arm, and compares with reference information.

Present invention employs main control modules integrally to control the upper limb rehabilitation robot based on binocular vision, control Data information derives from calibration group information acquired in Information Collecting & Processing module；Using remote control module to main control chip STM32 issues control instruction, and control instruction is transferred to main control chip STM32, main control chip by WIFI by remote control module According to data information control motor rotation, motor moves mechanical framework movement by gear band, realizes the servo-actuated of arm.

The present invention passes through fixing sleeve（Part）On length-adjustable flexible-belt respectively by the external fixation of the principal arm of mechanical framework It covers and is fixed together with the postbrachium of arm, the external fixing sleeve of auxiliary is fixed together with forearm, and palm is fixed on end fixing piece One piece, and left and right arm can be conveniently realized by the limited screw of adjusting fixing sleeve support rod and master/slave arm contact position Switching.

The present invention is attached to arm using surface myoelectric sensor acquisition muscular fatigue or spasm signal, surface myoelectric sensor The muscle surface of forearm and postbrachium, acquires and transmits the spasm or fatigue signal of arm muscles, and is transferred to master control mould by SPI Block, main control module are controlled the deceleration or braking of motor by pwm signal according to the fatigue or spasm signal of transmission, stop arm The safety of patient is protected in rehabilitation training.

The present invention uses segmented intelligent control method；Calibration group information is acquired by binocular vision method, will be collected Calibration group information be scaled 80%, which is stored at times, as control information；Main control module receives the One-stage control information, and motor rotation is controlled, motor moves mechanical framework movement by gear band, realizes the servo-actuated training of arm；? During this, position and angle information in arm servomechanism process are acquired with binocular vision method, and analyze with control information Compare, if error exceeds allowed band, system closure operation；If error is within the allowable range, just weaken beyond error next The level of enforcement of group control information, the level of enforcement of next group of control information is just increased lower than error, until set training It completes.

The main control module is STM32.

The communication module is based on SX1278 chip, using the LoRa spread spectrum of 2.4G mould group.

The motor module uses DC speed-reducing, model GW4058-3162.

The detection module is surface myoelectric sensor：KIT0012 muscle electric transducer.

Using Analyzing on Building A Planning Scheme of the invention, compared with prior art, have the advantages that.

(1) the invention proposes a kind of structure is simple, acquisition information is accurate, on control precision is high, and rehabilitation efficacy is apparent Limb healing robot can be used for the rehabilitation training of left and right arm functional disorder, and the Jian Buwaizhan that can complete arm is containing, shoulder Portion's vertical oscillation and ancon bend and stretch training.

(2) a kind of upper limb rehabilitation robot based on binocular vision of the invention, the mechanical framework can by length The flexible-belt of tune realizes the connection of principal arm external fixing sleeve, auxiliary external fixing sleeve and end fixed plate and arm, avoids hand Arm is contacted with the direct of structure is executed, and be can be convenient and be switched to another arm.

(3) a kind of upper limb rehabilitation robot based on binocular vision of the invention, the limbs Information Collecting & Processing mould The motion information of the method acquisition process patient arm to be trained of block binocular vision；It is different from two using binocular camera Angle captures the image information of arm, is handled by computer custom-built system the image information, obtains the three-dimensional space of arm Between the important parameters such as position and angle, improve the data precision of acquisition.

(4) a kind of upper limb rehabilitation robot based on binocular vision of the invention, the control information of the main control module The calibration information of patient's training is assisted from Physical Therapist；During the calibration process, Physical Therapist determines health according to patient's actual conditions The amplitude of movement and movement that refreshment is practiced obtains one group of calibration information, group calibration letter by limbs data collecting and processing system Breath derives from the working experience of physical therapy personnel, is more in line with the arm situation of patient, training effect is more preferably.

Detailed description of the invention

Fig. 1 is the design main view of the mechanical framework of the upper limb rehabilitation robot based on binocular vision.Fig. 2 is based on for this The plan view of the upper limb rehabilitation robot remote control module of binocular vision.

Fig. 3 is the Control system architecture block diagram of the upper limb rehabilitation robot based on binocular vision.

Fig. 4 is the binocular vision method of the upper limb rehabilitation robot based on binocular vision.

Fig. 5 is the system control method of the upper limb rehabilitation robot based on binocular vision.

Specific embodiment

The invention will be further described with the following Examples.

Referring to Fig. 1, the principal arm circumscribed fixing sleeve a is fixed by connecting rod and 5 phase of principal arm, auxiliary circumscribed fixing sleeve B is fixed by connecting rod and 11 phase of auxiliary, on principal arm circumscribed fixing sleeve, auxiliary circumscribed fixing sleeve and end fixing piece 12 Length-adjustable flexible-belt is installed, the fixation for arm and mechanical framework；The connecting rod of the external fixing sleeve of principal arm passes through principal arm, There are two screws with opposite threads on connecting rod, this two screws are located at the two sides of principal arm, form interlocking, tightly by connecting rod On the fixed principal arm in ground, when needing to switch trained arm, can quickly and easily it switch, auxiliary is same.

Referring to Fig. 1, the waist revolute joint 2 is mounted on pedestal assembly 1 by bearing, waist turning motor 3 It drives the joint to rotate in the horizontal direction by gear, realizes the containing training of Wai Zhan of shoulder；

The principal arm 5 is mounted on principal arm assembly, and end and major-minor arm seat 6 are hinged, principal arm motor by gear drive around Axis rotation, realizes that the lift of shoulder falls training；The auxiliary 11 and auxiliary back seat 10 is hinged, and auxiliary motor drives crank gear, Gear pulls auxiliary back seat by connecting rod 9, drives auxiliary to be pivoted, that realizes ancon bends and stretches training.

Referring to Fig. 2, before training starts, the calibration of training action is first carried out, presses demarcating key, starts limbs information collection The data input function of processing module, Physical Therapist drive patient's arm slowly to complete one group of instruction by drawing patient's palm mode Practice, limbs Information Collecting & Processing module is in running order in the process, obtains one group of effective calibration information.

Referring to Fig. 2, press trained key, training starts, main control module start to receive location information in calibration group information and Angle information, control motor rotation, single machine drive mechanical framework to act by gear drive, arm

It is servo-actuated.

Referring to Fig. 2, in the training process, if surface myoelectric sensor detects arm Fatigue or spasm signal, directly Main control module is fed back, main control module control motor stops immediately, and issues system alarm；If patient issue danger early warning information or Physical Therapist has found that training is abnormal, directly presses the stop key of remote control module, and motor stops operating immediately, and mechanical framework restores To initial position；If having ability to exchange patient to indicate that tired out or training activity reaches Physical Therapist and formulates the time, press long-range The rest key of control module, motor stalls, mechanical framework are restored to initial position.

Referring to Fig. 3, binocular camera captures the image of arm from two different angles, and passes through USB interface for image Information is transferred to computer；The reconstruct of two-dimensional image to three-dimensional space may be implemented in the image information of two different angles.

Referring to Fig. 3, after computer receives image information, by handling image information, the three-dimensional of arm is obtained The important parameters such as spatial position and angle；During the calibration process, computer can extract a group information, carry out fragmented storage, make For benchmark information；During arm training, computer is sent to master control mould for group information is demarcated by the period by communication module Block, main control module control motor rotation, and motor moves mechanical framework movement by gear band, and arm is servo-actuated, meanwhile, limbs information is adopted Collection module is in running order, obtains the position of arm and angle information in training process, and the information with calibration group in real time It makes comparisons, if the two error is larger, motor stalls have the function that protect patient's arm, can also be with by comparing result Judge the effect of rehabilitation.

Referring to Fig. 3, main control chip STM32 controls the acceleration of motor by sending pwm signal, slows down, positive and negative rotation and system It is dynamic；The included magneto-electric encoder of motor can provide the absolute position of rotor, which is fed back to master control by motor module Module is used for training of judgement accuracy of action.

Referring to Fig. 3, remote control module is communicated by WIFI with main control chip STM32, sends control instruction to master control Chip；Surface myoelectric sensor is communicated by SPI with main control chip STM32, and the forearm and postbrachium muscle that will test are tired Labor and spasm signal are transferred to main control chip；When surface myoelectric sensor detects the fatigue or spasm signal of arm muscles, Surface myoelectric sensor transmits the signal to main control chip STM32 by SPI, and main control module control motor stops turning at once It is dynamic, that is, stop the rehabilitation training of arm, protects the safety of patient.

Referring to Fig. 3, communication module be based on SX1278 chip, using the LoRa spread spectrum of 2.4G mould group, have it is accurate, Efficiently, jamproof characteristic communicates limbs Information Collecting & Processing module quick and precisely with main control module, reduce collection terminal with The signal delay of actuating station.

Referring to Fig. 4, this system acquires arm information using binocular vision method, and its step are as follows.

1. carrying out the stereo calibration of binocular camera, left and right camera is demarcated respectively, eliminate camera distortion, obtains camera Intrinsic Matrix and outer parameter matrix.

2. carrying out the three-dimensional correction of binocular camera, camera is remedied to ideal state, two images is made strictly to go It is corresponding, to obtain the accurate depth of field.

3. carrying out binocular ranging, the reconstruct of X-Y scheme to three-dimensional space is realized, search the strong texture between two width pictures, The depth of base point is only calculated under weak texture environment.

4. obtaining three-dimensional data.

5. obtaining gray scale diagram data, true parallax value is obtained.

6. identification arm simultaneously obtains binaryzation data, when arm appears in the camera lens visual field, arm can be identified, and Arm segment is rendered into white, other parts are rendered into black.

7. profile is extracted, by judging monochrome pixels point boundary come the irregular profile of preextraction.

8. finding the minimum circumscribed rectangle and the extraneous rectangle of maximum of arm, and according to minimum, maximum boundary rectangle, calculate hand The move angle of arm.

9. obtaining range of the arm apart from camera, i.e. measurement arm distance, extracted in depth bounds with for circulation Information.

Referring to Fig. 5, this system uses segmented intelligent control method, can carry out small-scale self-adjusting to training, make Closer to the information demarcated, effect is more excellent for training；Steps are as follows for the control method.

1. assisting the calibration group information of completion by Physical Therapist with the acquisition of binocular vision method, calibration group is believed by computer Breath is all scaled 80%, and is stored at times to the scalability information, as control data, i.e., by one group of training action point Solution is multiple processes.

2. main control module receives the first segment information of control data, and with segment information control motor rotation, motor passes through Gear band moves mechanical framework movement, realizes the servo-actuated training of arm.

3. acquiring location information and the angle in arm servomechanism process with binocular vision method while executing control information Spend information.

4. collected arm is servo-actuated information and first segment control data perform an analysis comparison, if error is beyond permission model It encloses, then system closure is run, and shows that this section of rehabilitation movement be not up to standard, arm may be damaged by continuing to execute.

5. judging the angle whether actual angle of arm is greater than in control data if the two error is within the allowable range Degree, if more than the level of enforcement of lower one-stage control data is then reduced by 5%；If being less than, by the execution of lower one-stage control data Degree mentions 5%.

6. judging whether the segment information of all control data is finished, if being finished, show one group of training action Terminate；If not finishing, main control module receives lower one-stage control data information, and controls motor rotation, executes step 3., directly It is finished to all control data segmentation informations.

By above-mentioned segmented intelligent control method, small-scale self-adjusting can be carried out to training, make training closer to mark Fixed information, rehabilitation efficacy are more excellent.

Claims (5)

1. the device includes mechanical framework, master the invention discloses a kind of upper limb rehabilitation robot based on binocular vision Control module, remote control module, motor module, limbs Information Collecting & Processing module, detection module, communication module, power module； It is characterized in that：The mechanical framework include pedestal assembly, waist revolute joint, principal arm rotary joint, principal arm, outside principal arm Connect fixing sleeve, auxiliary rotary joint, auxiliary, the external fixing sleeve of auxiliary；The main control module is STM32 single-chip microcontroller；Described Motor module includes DC speed-reducing, motor driver；The limbs Information Collecting & Processing module include binocular camera, Computer；The detection module is surface myoelectric sensor；The communication module is SX1278 chip；Surface myoelectric sensing Device is communicated by SPI with main control module, and binocular camera is communicated by USB interface with computer；The present invention discloses A kind of upper limb rehabilitation robot based on binocular vision.

2. a kind of upper limb rehabilitation robot based on binocular vision according to claim 1, it is characterised in that:The machine Tool framework realizes the fixation of arm by the external fixing sleeve of principal arm, the external fixing sleeve of auxiliary and end fixing piece.

3. a kind of upper limb rehabilitation robot based on binocular vision according to claim 1, it is characterised in that:The master The control information for controlling module assists the calibration information of patient's training from Physical Therapist.

4. a kind of upper limb rehabilitation robot based on binocular vision according to claim 1, it is characterised in that:The limb Body Information Collecting & Processing module uses the motion information of binocular vision method acquisition process arm.

5. a kind of upper limb rehabilitation robot based on binocular vision according to claim 1, it is characterised in that：The control System processed uses segmented intelligent control method, and the step of control method is as follows：1. passing through the acquisition calibration of binocular vision method Group information, and zoom in and out and store at times, as control information；2. main control module receives first segment and controls information, and holds Row；3. being analyzed with arm in binocular vision method acquisition implementation procedure servo-actuated position and angle information, and with control information Compare；4. if error exceeds allowed band, system closure operation；5. just weakening if error is within the allowable range beyond error The level of enforcement of next group of control information just increases the level of enforcement of next group of control information lower than error；6. main control module connects One-stage control information is accepted, executes step 3.；7. set training is completed until all control data are finished.