Hand-eye-force Coordination And Human-robot Interaction For Acupuncture Robot

The development of acupuncture robots is in the embryonic stage.Most of the existing works focus on the design of the mechanism.There is no complete theoretical system guidance for an acupuncture robot,which can help people to complete tasks.This thesis digs out that hand,eye and force are the essential parts of the acupuncture treatment,and when it comes to the acupuncture robot,the hand-eye-force coordination framework is proposed.With the guidance of the framework,the acupuncture tasks are divided into three parts: hand-eye,eye-force and hand-force coordination.For each of the three parts,solutions are proposed according to the task-specificity and they are validated by experiments.The main works include:1)Hand-eye-fore coordination framework for acupuncture robotIn the framework,the hand represents the end-effector of the robot,which can move in the guidance of the task.The eye,by definition is the data gained by vision sensors.The force is corresponding to the degree of physical contact.To complete the task,the three parts should coordinate with each other.With the guidance of the framework to decompose the tasks,the hand-eye coordination task need the robot to locate where the acupuncture point is and insert the needle into it by rule and line.Eye-force coordination task requires the robot to estimate the force with the help of visual information.The hand-force coordination task means that the robot can control the force while controlling the position of the end-effector.2)Solutions to the three parts of the hand-eye-force frameworkFor the hand-eye task,to locate where the acupuncture point is,the visual system is composed of the label detection based on HU contour invariant moment and the target discrimination based on KCF tracker.To insert the needle,the monocular eye-to-hand stratified control system is designed to complete the task.For the eye-force task,the problem is transformed to observe the deformation of the end-effector.The locations of three labels are chosen to estimate the force,and to locate where the pixel coordinate of the label is,HSV color space and morphology operation are chosen.In addition,to eliminate the error of camera distortion,the camera need to be calibrated.For the hand-force task,open-loop control strategy is chosen to deal with the press the constant force problem,least-square method and newton method are used to calculate the control quantity.To limit the force of the needle-holding hand,close-loop control strategy is proposed with the help of the designed convolution index.3)Control strategy based on imitation for human-robot interaction and the human-robot coordination system based on hand gestureTo control the multi-freedom of the robots fluently,a strategy based on posture imitation is proposed.To interface with people naturally,hand gesture recognition based on fHog feature is used to construct the human-robot coordination system.