CN-121980681-A - Unmanned vehicle virtual remote cockpit end intelligent body construction method and system

CN121980681ACN 121980681 ACN121980681 ACN 121980681ACN-121980681-A

Abstract

The application relates to the technical field of automatic driving and remote control, in particular to a method and a system for constructing an intelligent body at a virtual remote cockpit end of an unmanned vehicle. The method comprises the steps of receiving a remote driving request sent by the unmanned vehicle after triggering an abnormal event, responding to the request, dynamically distributing or creating a virtual driving agent from a virtual driving capability resource pool uniformly managed through a containerization technology, retrieving a matched target scene record from a remote driving scene library classified according to road conditions by the agent based on event information, and controlling the unmanned vehicle according to a disposal mode indicated by the record. The corresponding system comprises a cloud management platform and a virtual driving capability resource pool. The application realizes the flexible supply and intelligent decision of remote driving capability, obviously reduces the operation cost and improves the consistency and reliability of takeover response.

Inventors

CHEN XINHAI
Zu hui
LIU CHEN
CHEN JUNPENG
Wang bosi
Lou Fangdi
YANG JINGYAN

Assignees

招商智行(重庆)科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260107

Claims (10)

1. The unmanned vehicle virtual remote cockpit end intelligent body construction method is characterized by being applied to an unmanned vehicle virtual remote cockpit end intelligent body construction system, and comprises the following steps: Receiving a remote driving request sent by a target unmanned vehicle, wherein the remote driving request is generated by triggering an abnormal event during operation of the target unmanned vehicle; Dynamically allocating or creating a virtual driving agent from a virtual driving capability resource pool in response to the remote driving request to take over control of the target unmanned vehicle by the virtual driving agent; The virtual driving intelligent agent is specifically used for acquiring event information sent by a target unmanned vehicle, searching target scene records matched with the abnormal event from a preset remote driving scene library based on the event information, wherein the remote driving scene library comprises a plurality of scene records classified according to road conditions; the virtual driving capability resource pool is formed by uniformly managing and generating a plurality of virtual driving intelligent agents through a containerization technology by the unmanned vehicle virtual remote cockpit end intelligent agent construction system, and an intelligent driving function is arranged in the virtual driving capability resource pool.
2. The method of claim 1, wherein retrieving a target scene record matching the abnormal event from a preset remote driving scene library comprises: Generating at least one safety dimension characteristic according to the event information, wherein the event information comprises an abnormal event and real-time state data of the target unmanned vehicle; Sequentially searching in a micro scene library, a mesoscopic scene library and a macroscopic scene library by taking the at least one safety dimension feature as an index, wherein the micro scene library corresponds to a single-vehicle-level road condition, the mesoscopic scene library corresponds to a multi-vehicle-level road condition, and the macroscopic scene library corresponds to a road network-level road condition; comparing the security dimension feature with a plurality of dimension thresholds in the retrieved scene record; and if all the comparison dimensions accord with the corresponding threshold range, judging that the matched target scene record is retrieved.
3. The method of claim 2, wherein the treatment regimen comprises at least one of: Executing, by the virtual driving agent, a full authority takeover; generating auxiliary decision information by the virtual driving agent, and sending the auxiliary decision information to an entity cockpit to be manually taken over; and generating control parameters by the virtual driving intelligent agent, and carrying out parameter correction on preset parameters of a vehicle end controller of the target unmanned vehicle.
4.A method according to claim 3, wherein the preset parameters include steering coefficient, accelerator pedal coefficient or brake pedal coefficient.
5. The method of claim 1, wherein the remote drive request carries brand information and model information for the target drone.
6. The method of claim 1, wherein control of the target drone is taken over by a pre-set human cockpit if there is no virtual driving agent that cannot match the remote driving request.
7. The method according to claim 1, wherein the method further comprises: monitoring a control link state between the virtual driving agent and the target unmanned vehicle; When the control link is monitored to be abnormal, starting a plan according to the severity level of the target scene record, wherein the plan comprises switching to a preset manual cockpit takeover or issuing a safe parking instruction to the target unmanned vehicle.
8. The unmanned vehicle virtual remote cockpit end intelligent body construction system is characterized by comprising a cloud management platform; The cloud management platform is used for receiving a remote driving request sent by a target unmanned vehicle, wherein the remote driving request is generated by triggering an abnormal event during operation of the target unmanned vehicle; dynamically allocating or creating a virtual driving agent from a virtual driving capability resource pool in response to the remote driving request to take over control of the target unmanned vehicle by the virtual driving agent; The virtual driving intelligent agent is specifically used for acquiring event information sent by a target unmanned vehicle, searching target scene records matched with the abnormal event from a preset remote driving scene library based on the event information, wherein the remote driving scene library comprises a plurality of scene records classified according to road conditions; the virtual driving capability resource pool is formed by uniformly managing and generating a plurality of virtual driving intelligent agents through a containerization technology by the unmanned vehicle virtual remote cockpit end intelligent agent construction system, and an intelligent driving function is arranged in the virtual driving capability resource pool.
9. The unmanned vehicle virtual remote cockpit end agent construction system of claim 8 further comprising an artificial cockpit; And if the virtual driving intelligent agent is not available and cannot be matched with the remote driving request, taking over the control of the target unmanned vehicle by a preset manual cockpit.
10. The unmanned vehicle virtual remote cockpit end agent construction system of claim 8, wherein the target unmanned vehicles are respectively in communication connection with the cloud management platform by means of wireless communication.

Description

Unmanned vehicle virtual remote cockpit end intelligent body construction method and system Technical Field The application relates to the technical field of automatic driving and remote control, in particular to a method and a system for constructing an intelligent body at a virtual remote cockpit end of an unmanned vehicle. Background Along with the large-scale landing of low-speed functional unmanned vehicles such as unmanned delivery vehicles, unmanned sanitation vehicles and the like on open roads, a technical scheme based on 'single-vehicle intelligent + remote manual connection' has become the mainstream of the industry. Unmanned vehicles may encounter various types of abnormal events in operation, such as sensor faults, complex traffic conflicts, path blocking, etc., and when their own automated driving system (bicycle intelligence) is unable to handle, a take over request may be sent to a remote control center. Currently, remote takeover relies primarily on physical remote cabins deployed at the control center and human operators. However, this model has the significant disadvantage that, first, the cost of manual take-over is high, and the cost of operator incubation, management, and physical cockpit deployment grows linearly with fleet size, severely limiting the scalability of the business model. Secondly, there is a delay in manual response and differences in experience and judgment of different operators, resulting in non-uniformity of the efficiency of take over and the treatment criteria. Furthermore, existing systems lack unified and flexible management of remote driving capability and cannot intelligently allocate processing resources according to event type and severity. Therefore, a new scheme is needed in the industry to reduce the operation cost, improve the response efficiency and the decision consistency, and realize the large-scale and intelligent supply of the remote driving capability. Disclosure of Invention In view of the above, the embodiments of the present application are directed to providing a method and a system for constructing an intelligent agent at a virtual remote cockpit of an unmanned vehicle, which aims to overcome the defects of the prior art that the artificial remote connection is relied on, and the problems of low response efficiency and unstable decision are solved. The first aspect of the application provides a method for constructing an intelligent agent at a virtual remote cockpit of an unmanned vehicle, which is applied to a system for constructing the intelligent agent at the virtual remote cockpit of the unmanned vehicle, and comprises the following steps: Receiving a remote driving request sent by a target unmanned vehicle, wherein the remote driving request is generated by triggering an abnormal event during operation of the target unmanned vehicle; Dynamically allocating or creating a virtual driving agent from a virtual driving capability resource pool in response to the remote driving request to take over control of the target unmanned vehicle by the virtual driving agent; The virtual driving intelligent agent is specifically used for acquiring event information sent by a target unmanned vehicle, searching target scene records matched with the abnormal event from a preset remote driving scene library based on the event information, wherein the remote driving scene library comprises a plurality of scene records classified according to road conditions; the virtual driving capability resource pool is formed by uniformly managing and generating a plurality of virtual driving intelligent agents through a containerization technology by the unmanned vehicle virtual remote cockpit end intelligent agent construction system, and an intelligent driving function is arranged in the virtual driving capability resource pool. Optionally, retrieving a target scene record matched with the abnormal event from a preset remote driving scene library, including: Generating at least one safety dimension characteristic according to the event information, wherein the event information comprises an abnormal event and real-time state data of the target unmanned vehicle; Sequentially searching in a micro scene library, a mesoscopic scene library and a macroscopic scene library by taking the at least one safety dimension feature as an index, wherein the micro scene library corresponds to a single-vehicle-level road condition, the mesoscopic scene library corresponds to a multi-vehicle-level road condition, and the macroscopic scene library corresponds to a road network-level road condition; comparing the security dimension feature with a plurality of dimension thresholds in the retrieved scene record; and if all the comparison dimensions accord with the corresponding threshold range, judging that the matched target scene record is retrieved. Optionally, the treatment mode includes at least one of the following: Executing, by the virtual driving agent, a full authority takeover;