Research On UWB-based Relative Localization Algorithms In Blind Envrionment

For soldier and first responder operations,like military action,anti-terrorism act and fire fighting rescue,it is crucial to get the location information of drones,ships,and first responders.Most rescue events happen in blind environments with no localization infrastructure,no WiFi fingerprinting database and no map.This is to say,most rescue and emergency scenarios are unknown environment for localization and it is challenging to get location information with high reliability and accuracy.In order solve the problem,this paper focus on the localization with no infrastructure in blind environment.The contributions of this thesis are summarized as follows:(1)By the analysis of two-piont positional relationship,a system-level solution is proposed for point-to-point localization system with no infrastructure in blind environment in this paper.In this system,relative position can be obtained with relative altitude,relative distance and relative direction.Atmospheric pressure is used for altitude and TOA for distance."Rotation TOA measurement"and"template matching"are utilized for direction estimation.The anchor holds the measurement device in his hand and turns around to get a measurement sequence and the template sequence can be calculated with with direction angle,rotation radius and the smallest ranging value.Field test experiments verify the feasibility of the system.(2)By the analysis of rotating anchor-based TOA measurements,characteristic modeling of TOA ranging error based on rotation measurement in presented in this paper.The a-stable distribution was adopted,and it can better fit the empirical data than Gaussian distribution,generalized extreme value distribution and log-normal distribution.(3)A direction estimation algorithm based on Maximum Correntropy Criterion and Dynamic Time Warping is proposed in this paper,to reduce the influence of impulsive ranging error and non-uniform sampling.By using exponential decline weights,maximum correntropy criterion is robust to impulsive data and performs well with better steady state performance.In addition,Dynamic programming is adopted by stretching and compressing the sequence to establish the optimal wraping path.Empirical measurenments were conducted in different scenarios.Empirical tests in different senarios show that the reliability of the algorithm is verified.(4)An exact initial coordinate free relative localization algorithm based on exterior penalty function is proposed in this paper,to reduce the sensitivity on initial coordinate accuracy.Penalty stratergy is involved to integrate the constraints as a penalized part of the objective function and approach the boundary of feaisible region from whether a feasible solution or not.In addition,geometric constraint is adopted in this algorithm for the targets having inherent shape features.Simulation and empirical tests show that the algorithm has improved profermance than LM and Powell algorithms even with random initial coordinates.And better performance can be achieved with geometric constraints.