Research On Flexible Upper Limb Posture Detection And Recognition Based Fiber Bragg Grating

Research on the flexible posture detection and recognition methods has a great significance for exoskeleton robots to perceive human posture and understand human intentions.Especially there are few effective ways to achieve the interaction between exoskeleton robots and people at the current stage.With the continuous innovation of science and technology,people have higher requirements for posture detection and recognition.Flexibility,convenience,and wearability are the design directions of related sensors.In this thesis,we design a upper limb posture detection and recognition experimental prototype based on fiber grating sensing network,it can be used for human-computer interaction.The prototype can not only provide ideas for human-computer information interaction of industrial exoskeleton,but also can be used for joint point capture and posture classification in the fieid of medicai rehabilitation or animation.The specific research contents are as follows:(1)The defects are found in existing posture recognition sensing devices by searching and refining research results in related fields.And study the principle of fiber grating sensing and the principle of demodulation.We propose the idea that apply fiber grating to posture recognition by analyzing the defects about previous results.(2)Based on the principle of reducing the redundancy and complexity of fiber grating,the upper limb freedom is simplified according to the range of daily motion of upper limb joints.Upper limb posture reproduction and palm coordinate algorithm are determined by analyzing forward kinematics of simplified upper limbs' model.Then the simulation proves the correctness of the algorithm.(3)The principle and method of integrated measurement of upper limb multi-joint angle based on fiber grating are proposed.Measure the joint angle by arranging the fiber grating at a suitable position on the upper limb of the human body.We optimized the position of the fiber grating in the upper limb by the experiment that directly arranging fiber gratings at joint points.Thereby completing the construction of a flexible sensor network for measuring the joint angle.Then the feasibility of the scheme is demonstrated by performing performance tests on the fiber gratings used for the measurement.(4)Based on previous research results,manufacture a flexible wearable joint angle measuring fabrics.The calibration method is used to establish the corresponding relationship between the measuring point on the fabric and the angle of the joint.The problem of inaccurate output angle signals has been corrected by a large number of experiments.And establish a LabView-based attitude detection platform.Applying the prototype to a 5-degree-of-freedom angle measurement comprehensive experiment for different individuals.Achieved the goal of upper limb posture detection.(5)Constructing the upper limb posture data set that regard joint angle as a feature by the prototype output.The machine learning algorithm are applied to the dataset.Analyze the experimental results corresponding to different parameters,and the feasibility of applying the experimental prototype combined with machine learning algorithm to pose recognition is verified.The classification recognition of the posture is realized.Finally,the writer summarizes the situation of the whole subject,finds out and analyzes the shortcomings of the research.Proposes some improvement plans,and points out the research direction of the future topics.