Method and system for precise positioning of intelligent vehicles using hybrid sampling filtering

Abstract

The invention discloses a method and system for performing accurate positioning by adopting hybrid sampling filter by an intelligent vehicle. The accurate positioning of a vehicle in an environment without / with a weak satellite signal can be realized by utilizing the hybrid sampling filter and combining pre-constructed high-precision map data or real-time point cloud data collected by a laser radar or a camera, the method and system can be used for aided driving and unmanned driving of the vehicle; a new probability positioning framework is adopted, and a random sampling consistency algorithmand a particle filter / histogram filter algorithm are efficiently integrated by using a nuclear density estimation method, the disadvantages of the opposite parties are remedied by sufficiently utilizing the respective advantages, and then more robust, accurate and quick positioning is performed by using a probability grid map based on a Gaussian hybrid model.

Description

technical field [0001] The invention belongs to the technical field of intelligent driving, and in particular relates to a method and system for precise positioning of an intelligent vehicle by using hybrid sampling filtering. Background technique [0002] Intelligent driving technology involves cognitive engineering, vehicle engineering, electrical and electronic engineering, control science and engineering, artificial intelligence and many other disciplines. Intelligent vehicles are an important symbol to measure a country's scientific research strength and industrial level. The emergence of unmanned vehicles has fundamentally changed the vehicle driving mode in the traditional "human-vehicle-road" closed-loop system, liberating the driver from fatigue driving, using advanced sensors and information technology to control vehicle driving, Automating routine, long-lasting, low-level, and repetitive operations in driving activities can greatly improve the efficiency and safet...

AI Technical Summary

Problems solved by technology

For example, in a tunnel, the occlusion of the tunnel will lead to the failure of satellite positioning, and the single feature in the longitudinal direction of the tunnel will lead to the failure of LiDAR positioning. Only dead reckoning can be used for relative positioning. Vehicles drive in this environment for a certain distance and time Finally, a large cumulative error will be generated. At this time, if the vehicle travels to a fork in a tunnel, it needs to use LiDAR and the map to locate under a large initial deviation.

In addition, unmanned vehicles will also encounter problems similar to "kidnapping". For example, when starting up in an underground garage, the initial position error at this time is usually very large, sometimes hundreds of meters, and the vehicle's heading is also completely unknown.

Images