Urban monocular mapping method and system based on GPS constraint and dynamic correction

Abstract

The invention relates to an urban monocular mapping method and system based on GPS constraint and dynamic correction, and the method comprises the steps: screening a curved road section according to the trajectory curvature of a vehicle during driving, and detecting the internal and external parameters of a monocular camera according to the sequence image of the screened curved road section in the monocular camera; determining pose absolute coordinate transformation of the monocular camera according to whether the change of the internal reference or the external reference of the monocular camera is greater than a threshold value or not; and outputting a map according to the pose absolute coordinate transformation and the multi-view intersection measurement of the monocular camera. According to the invention, the monocular camera and the consumer-level GPS equipment are fused for single-vehicle mapping, so that the dependence on a laboratory calibration environment is reduced, and the mapping reliability is improved.

Description

technical field [0001] The invention belongs to the field of automatic driving navigation and high-precision maps, and relates to a monocular mapping method and system, in particular to a monocular mapping method and system for urban areas based on GPS constraints and dynamic correction. Background technique [0002] In related fields such as automatic driving navigation and high-precision maps, in order to implement safe route planning and navigation decisions, three-dimensional geometric information of the road environment around the vehicle is required. Currently, map dealers need professional acquisition vehicles to collect and produce corresponding road environment maps, including lasers. Radar, binocular cameras, high-precision inertial navigation, wheel speed information, millimeter-wave radar and other professional equipment generally have relatively high acquisition costs. Due to the frequent construction and maintenance of the urban road environment, the road eleme...

AI Technical Summary

Problems solved by technology

However, the fusion of GPS and monocular camera for measurement Figure 1 Generally, it is necessary to accurately calibrate the sensor in advance, including the internal parameters of the camera and the external parameters from the camera to GPS, time delay and other parameters. In addition, due to the long-term driving bumps of the vehicle, these key parameters will also change dynamically over time. These require the laboratory The key parameters calibrated in the environment are difficult to obtain stably in natural scenes. In addition, the scale drift of the monocular camera mapping itself also increases the practical difficulties of cameras and consumer-grade GPS equipment used in urban road environment mapping.

[0004] The current method of solving monocular camera mapping is difficult to take into account the requirements of automation, accuracy and reliability at the same time, and the advantages of low cost, simple deployment and wide application of monocular camera make it an ideal choice for crowdsourcing mapping. Bicycle mapping with cameras and consumer-grade GPS devices is the basis for crowdsourced mapping with vision as the core. However, the difficulty of calibrating cameras and GPS devices in urban environments and the scale drift of monocular maps have negative impacts on crowdsourced mapping. Accuracy and reliability pose major challenges

Images