Research On The Trajectory Patterns Mining And Path Planning For Moving Objects In Uncertain Environments

In recent years, as an important branch of mobile computing technology and as one of the support technologies of location based services, moving object database has received widely attention and research. For intelligent transportation systems, digital military battlefield and driver assistance systems, it is of great practical value to predict the trajectories of moving objects with uncertainty in a real-time, accurate and reliable fashion. Intelligent trajectory prediction can provide accurate location based services, as well as monitor and estimate the traffics in order to provide the best path, which has grown to be an active research direction. It is essential for us to develop accurate and effective location prediction approaches. The prediction methods have been proposed and improved continuously.This thesis aims to do research on trajectory prediction algorithms for moving objects, which can greatly improve the performance of prediction error and time. A new trajectory prediction model based on Gaussian mixture models called GMTP (Gaussian Mixture Model Trajectory Prediction) is proposed, In view of the complex and simple trajectory using the Gaussian mixture model and Gaussian Process, then mining trajectory patterns. Compared with the hidden Markov model, Kalman Filter, continuous time Bayesian network prediction model not only improves the prediction error and time, also has certain anti-interference for noise data. Finally, combined with the rapidly exploring random tree algorithm for trajectory prediction algorithm and improved the development of the simulation system of three scenarios.The major contributions of this thesis are given as follows:(1) A new Trajectory Prediction algorithm called GMTP is proposed in this study, This thesis introduces modeling the complex motion patterns based on Gaussian mixture models and how to use EM to train model, calculating the probability distribution of different types of motion patterns by using Gaussian mixture model in order to partition trajectory data into distinct components, and inferring the most possible trajectories of moving objects via Gaussian process regression.(2) Comparing three kinds of trajectory prediction algorithm based on real trajectory data sets. The GMTP algorithm is naturally a Gaussian nonlinear statistical probability model and the advantage of the proposed model is that the result is not only a predicted value, but also a whole distribution beyond the future trajectories, in order to infer the location in regard to some motion pattern, e.g, uniformly accelerated motion, by using statistical probability distribution. Extensive experiments are conducted on real trajectory data sets and the results show that:the prediction accuracy of the GMTP algorithm is improved by 22.2% and 23.8%, and the runtime can be reduced by 92.7% and 95.9% on average, respectively, when compared to the Gaussian process regression model and Kalman filter prediction algorithm with similar parameter setting.(3) Proposes a recommendation algorithm incorporating user similarity based on the location of social networking, user similarity by combining new social patterns of friend’s familiar user interest similarity degree and the interest of the calculation, and may be recommended location of interest. Finally, comparing three algorithms based on the Foursquare’s real checkin data.(4) Based on the proposed trajectory prediction algorithm in this thesis, a system called PPNS(Path Planning and Navigation Simulation) is developed. Results show the performance of the navigation algorithm for a car-like robot moving among dynamic obstacles with probabilistic trajectory prediction.