A vehicle-mounted voice control method and device, electronic equipment and storage medium

By deploying large language models and environmental representation models in the in-vehicle voice assistant, voice input is converted into text in real time and high-level driving task instructions are generated, solving the problems of network latency and personalization, and realizing real-time safe voice control and personalized driving with low latency.

CN122369445APending Publication Date: 2026-07-10AUTOLINK INFORMATION TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
AUTOLINK INFORMATION TECHNOLOGY CO LTD
Filing Date
2026-04-21
Publication Date
2026-07-10

AI Technical Summary

Technical Problem

Existing in-vehicle voice assistants suffer from problems such as high network latency, difficulty in understanding open-domain natural language, and difficulty in adapting to different users' driving style preferences, resulting in low real-time performance and personalization.

Method used

By deploying large language models and environmental representation models on the vehicle, user voice input is converted into text in real time and classified into tasks to generate high-level driving task instructions. These instructions are then processed in conjunction with environmental representation vectors to achieve localized and low-latency voice control of the vehicle.

Benefits of technology

It achieves real-time safe voice control of the vehicle with low latency, can accurately understand open-domain natural language, adapt to different users' driving style preferences, improve personalization, and work in conjunction with the autonomous driving decision system to execute dynamic driving tasks.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122369445A_ABST
    Figure CN122369445A_ABST
Patent Text Reader

Abstract

The application provides a vehicle-mounted voice control method and device, electronic equipment and a storage medium. The method comprises the following steps: receiving voice input information from a user, identifying the voice input information in real time through a voice recognition module to transcribe the voice input information into text information, sending the text information into a task classification model to obtain a task classification result, constructing a task prompt based on the task classification result, the text information and an environment representation vector, inputting the task prompt into a large language model for processing, sending a high-level driving task instruction back to a smart driving server when the processing result is the high-level driving task instruction, receiving the high-level driving task instruction through a discrimination model of the smart driving server, taking the high-level driving task instruction as a decision target, generating a corresponding high-level driving task trajectory, and outputting a high-level driving task control signal corresponding to the high-level driving task trajectory to a vehicle actuator, so that the vehicle actuator performs a corresponding action based on the high-level driving task control signal.
Need to check novelty before this filing date? Find Prior Art