| In the "Internet of Things" era,the indoor activity time for pedestrian is increasing,and the demand for indoor location services is rapidly expanding.With the popularity of miniaturization and low-power inertial measurement units(IMU),micro-inertial sensor-based pedestrian dead reckoning(PDR)autonomous positioning technology has become one of the most effective methods for providing indoor location services.However,the high-precision autonomous positioning technology of pedestrian under the current complex motion behavior mainly faces the technical problem that the motion behavior is limited and the error accumulates with time.Aiming at the above problems,this thesis studies the autonomous positioning of human motion behavior constraints based on micro-inertial sensors,and studies the accurate identification of multiple sports behaviors and the high-precision PDR autonomous positioning technology based on behavior constraints.The main contents are as follows:(1)In the accurate identification of human sports behavior: the basic theory of human motion behavior recognition is studied,and the multi-directional walking recognition technology based on time domain-slope feature is proposed,and an average recognition of 97% for motion behavior is obtained in the two-dimensional plane.Identification: A multi-feature threshold fusion algorithm based on attitude angle and acceleration is proposed to realize the recognition of abnormal motion behavior with recognition rate up to 97.9%.A technique based on isochronous height change of barometer is proposed to realize three-dimensional motion behavior recognition.The recognition rate is approximately 92%.Finally,the hierarchical classification strategy is used to fuse multiple motion behavior recognition algorithms to realize the construction of multi-motion behavior recognition system.(2)For the high-precision PDR autonomous positioning technology based on behavior constraint: the basic theory of PDR autonomous positioning is studied.For the conventional gait detection algorithm has poor adaptability to complex motion behavior,a gait detection based on multi-motion behavior constraint is proposed.The gait detection accuracy is 20% higher than that of the conventional algorithm.For low accuracy of the common nonlinear step estimation model,an adaptive behavior step estimation technique is proposed for multi-motion behavior.Compared with the empirical model algorithm,the overall improvement is about 10.8%.For the drift of the height for barometer,a height drift compensation algorithm based on the plane motion behavior constraint is proposed,which can accurately distinguish the floor where the person is located,and the error of height is less than 1 m for the same layer.The optimized parameters are applied to the human space trajectory estimation algorithm adapting to various motion behaviors,and the autonomous positioning system algorithm for the whole motion behavior constraint is experimentally verified.Compared with the non-behavior-constrained autonomous algorithm,the closed-loop positioning accuracy is improved by 4 times approximately.The real-time,continuous and accurate position information of the pedestrian can be obtained under the complex movement behavior. |
