Unmanned vehicle intelligent identification method

Abstract

The invention provides an unmanned vehicle intelligent identification method. A vehicle controller calculates the collision probability between any two vehicles in a jurisdiction communication area, and the vehicle collision risk is the number of times that the distance between the vehicles on a two-dimensional plane falls on an absolute safe distance and a relative safe distance; the vehicle MCUsets attributes which are representative and suitable for pedestrian identification in road pedestrian attributes, including biological characteristics of limbs, trunk, head and face of a human body and related attributes of wearing colors, upper garments, lower garments and shoes; a deep learning model is constructed, wherein the deep learning model adopts a convolutional neural network model; and the vehicle controller obtains parameters of vehicles in a region managed by the roadside unit through communication with the roadside unit, calculates the vehicle collision risk of the whole road,publishes information, and pushes the information to the accident early warning information of the vehicles managed by the roadside unit. Compared with a traditional identification method, the convolutional neural network method is adopted, and compared with a method without deep learning, the better identification rate is reflected.

Description

technical field [0001] The invention relates to the technical field of pattern recognition of unmanned vehicles, in particular to an intelligent recognition method for unmanned vehicles. Background technique [0002] At present, the calculation amount of the pattern recognition algorithm of unmanned vehicles is very large, and the performance requirements of the calculation unit are very high, which makes the productization and miniaturization difficult. Therefore, it is relatively difficult to solve the dual-purpose computing problem on chips or FPGAs. [0003] Many research institutions or manufacturers in the world, most of them use servers for image processing and calculation, and some of them simplify the algorithm and use FPGA for processing, which greatly affects the effect, and there are many noises and holes. , which is unfavorable to subsequent calculations and poses a security risk. In reality, it is often very simple to process pedestrian features, vehicle obst...

AI Technical Summary

Problems solved by technology

[0003] Many research institutions or manufacturers in the world, most of them use servers for image processing and calculation, and some of them simplify the algorithm and use FPGA for processing, which greatly affects the effect, and there are many noises and holes. , which is unfavorable to subsequent calculations and poses a security risk

Images