A Seamless Indoor-Outdoor Positioning Transition Method Based on HMM/Bayesian Inference Using GNSS and Bluetooth

Aiming at the problem that traditional indoor-outdoor positioning transition method relies on sensor threshold discrimination for transition resulting in insufficient accuracy when the number of collected features is small, this paper takes the GNSS/BLE indoor-outdoor seamless positioning system as the carrier, and proposes a method based on HMM (Hidden Markov Model)/Bayesian Inference machine learning method for seamless indoor-outdoor positioning transition, which utilizes the continuous time feature of HMM to describe the seamless continuous positioning process, and solves the optimal hidden state of HMM by Bayesian Inference algorithm to achieve stable, smooth and high-precision positioning during indoor-outdoor scene transition. The simulation and practical experimental validation show that the proposed method has a minimum transition error of 0.1870 meters in the simulation experiment and 0.1098 meters in the practical experiment. Compared with other methods such as HMM/Viterbi, CRF, SVM, etc., the proposed method has a higher accuracy when the number of captured features is small, and it also reflects a better Stability. The method based on HMM/Bayesian Inference is excellent for indoor-outdoor scene transition, and its fastness and accuracy provide a new idea for seamless indoor-outdoor transition localization, which has a wide range of application prospects and practical value.