A remote takeover method for unmanned vehicles based on scene familiarity

Abstract

The invention discloses a remote takeover method for an unmanned vehicle based on scene familiarity. When an unmanned vehicle faces a new scene, it processes sensor data to generate corresponding fused scene feature information; and then compares it with the information in the local memory. , to calculate the familiarity, if the familiarity exceeds the threshold, the scenario strategy with the highest similarity will be issued directly; otherwise, communicate with the cloud to obtain the historical scenario strategy, if there is a historical scenario with the highest similarity Issue historical scenario policies; if they do not exist, query the consensus policy library in the cloud, and issue consensus policies that exceed the threshold; otherwise, submit a remote takeover application to the remote control center. The method disclosed in the present invention can make full use of the historical experience strategy of the unmanned driving terminal and the consensus strategy verified by a large number of unmanned driving terminals, thereby effectively reducing the frequency of manual request to take over, improving the level of autonomy of unmanned driving and ensuring safety .

Description

technical field [0001] The invention belongs to the technical field of information processing, and in particular relates to a remote takeover method for unmanned vehicles based on scene familiarity. Background technique [0002] At present, with the increase of computing power and local memory capacity of automobile terminal processors and the development and progress of technologies such as sensors, data mining technology, deep learning and intelligent decision-making planning, unmanned driving has become the focus of research and development in the current academic and industrial circles field, and is expected to enter people's lives in the near future, bringing subversive changes to the transportation system. [0003] However, there are still many problems in the current driverless cars, including poor generalization ability to the scene and difficult decision-making in complex scenes. The current unmanned driving is still in the stage of low intelligence between L3-L4. ...

AI Technical Summary

Problems solved by technology

However, how to effectively judge when manual intervention should be performed is a difficult problem. Therefore, in practical applications, special personnel are often required to monitor at all times and manually judge the timing of takeover.

This method is not only inefficient and unsafe, but also increases the cost and reduces the practicality of driverless cars

Images