Planning method of automatic driving system

Abstract

The invention discloses a planning method of an automatic driving system. The invention discloses a planning method of an automatic driving system. The method comprises the steps that S1, acquiring acurrent road environment image g by the system, using a filter in the convolutional neural network to extract the structure and road condition information Phi of the road image; and obtaining reward information R of the current road environment image through the mapping function fR, obtaining initialization of a kernel function Kwp for irregular graph convolution operation through the mapping function fP, and using an adjacent matrix of the image as an activation parameter of the Kwp to obtain an initial convolution operation sub-P. The deep reinforcement learning method based on the generalized asynchronous value iterative network model has the beneficial effects that the problem that an existing automatic driving system cannot carry out high-success-rate and high-accuracy road planning in a road environment with a complex structure and unknown road condition information can be well solved.

Description

technical field [0001] The invention belongs to the technical field of road planning in an automatic driving system, and specifically relates to a variety of irregular road environments. Using a deep reinforcement learning method embedded in a generalized asynchronous value iterative network model and a simulated automatic driving system in a complex structure and unknown road condition information Strategic planning in the road environment. Background technique [0002] At present, in the field of autonomous driving vehicles, it is more and more common to adjust the road planning strategy according to the road structure and the complexity of road information. For example, in a road condition that contains multiple curves and forks, the vehicle must not only adapt to the difficulty of each curve and pass through each curve with different curvatures smoothly, but also accurately select the fork that can lead to the target. mouth and finally reach the target point. However, ...

AI Technical Summary

Problems solved by technology

[0005] The technical problem to be solved by the present invention is to provide a planning method for an automatic driving system, based on a generalized asynchronous value iterative network model, a deep reinforcement learning method that can perform road planning in an environment with complex road structures and unknown road condition information. Solve the problem that the existing automatic driving system cannot perform accurate strategy planning due to the complex road structure and insufficient environmental information

Images