Research On Imaging Positioning Technology Based On Visible Light Fusing Attitude And Track Information

In recent years,the location-based services have grown explosively.Relying on satellite navigation and mobile network,outdoor positioning technology starts research early and is widely applied,but indoor positioning technology is not yet mature.Visible light technology,with its unique advantages such as green environmental protection,broadband high-speed and indoor ubiquitous,provides a new way for application of indoor positioning,and now has become a hotspot in the research field of indoor positioning technology at home and abroad.Based on technical links of visible indoor imaging positioning technology,this paper undertakes research for high-dimensional parameter estimation,constraint object positioning optimization and object dead reckoning,mainly including:1.As for the solution for high-dimensional parameter nonlinear equations of visible light imaging positioning,this paper introduces UKF(Unscented Kalman Filter)design idea,and puts forward a high-dimensional parameter estimation method based on UKF.By building initial sample space,this method achieves constraint iteration in limited scope,and completes the solution of nonlinear high-dimensional equation quickly and efficiently.Simulation results show that in the case of building reasonable initial sample space,mm-level positioning can be achieved.When the initial sample randomly selected within a wide range,positioning error increases,part of the solver can't converge.2.As for slow convergence in high-dimensional space of object positioning estimation,this paper introduces dimension reduction design idea,combined with instantaneous attitude estimation of inertial units,which reduces the search space to sample space with constrained features.Accordingly,the paper puts forward a high-dimensional space search method of joint target instantaneous attitude estimation.The method takes instantaneous attitude information characteristics to build object optimization function as constraint condition,and reduces the evolution scope of individual good genes in genetic algorithm,which improved the evolution speed significantly,avoided the dependence of initial sample space at the same time,and increased the robustness of algorithm.Simulation results show that with the same parameters,the mean square error of positioning estimation is no more than 40 mm.3.As for the problem of inconsistent positioning caused by missing observable information in object moving state,this paper introduces inertia navigation information,such as acceleration and direction,and presents an object dynamic positioning method of tracking information fusion.Combining historical track and inertial navigation information,this method measures stride frequency,stride length and moving direction,the three main parameters,and achieves continuous positioning of moving object through prediction and filtering.