Context-adaptive Sensor Optimization And Selection In All Source Positioning And Navigation

With the development of navigation and positioning technology,the application scenario of location-based services has gradually become more and more complex,and the reliability and accuracy requirements for location-based services keep increasing.However,existing sensors or positioning technologies have drawbacks,because of these drawbacks,using sensors in the unsuitable scene may cause large errors or even positioning failures.Therefore,all source fusion positioning and navigation technology that can obtain reliable navigation and positioning information at any time,in any environment,and on any platform,has emerged.Firstly,the paper introduces the overall architecture of all source positioning and navigation system,including the application layer,algorithm layer,signal layer,and hardware layer,supporting for contextawareness and context-adaptive sensor selection and optimization,designing a ROS-based plug-and-play fusion navigation software framework.Secondly,for the all source fusion positioning and navigation technology,this paper studies the influence of environments and platform motion state on sensor positioning information for several commonly used sensors.Combined with the measurement model and error model of sensors,this paper proposes a sensor optimal selection method based on contextawareness of environment and platform motion context.Different from the widely used method of positioning and navigation using fixed sensors,this method has the context adaptability.Then,this paper designs an all source positioning and navigation algorithm based on sensor error model,environment model and platform motion model,and implements with three algorithms: extended Kalman filter,particle filter and graph optimization.Among them,the paper focuses on the factor graph optimization fusion positioning and navigation algorithm,uses the sensor quality evaluation through sensor cross-checking method to optimize the reliability of factor graph optimization fusion positioning and navigation algorithm,so that it can according to the context and automatically adjust the weight of sensors according to their real-time performance in the fusion process.Finally,this paper implements a complete all source positioning and navigation system,equipped with Global Navigation Satellite System,2D laser radar,inertial measurement unit,stereo camera and optical flow meter five kinds of sensors.Experiments are done in both indoor simple scenario and indoor-outdoor-switching complex scenario.Experiments show that the context-adaptive sensor selection and optimization proposed in this paper can automatically select the applicable sensor according to the environment,can achieve robust positioning.In complex scenario,the 67% positioning error is less than 3 meter while in simple scenario,it is less than 0.5 meter.