CN-121979498-A - Intelligent driving scene management method and device based on configuration table

CN121979498ACN 121979498 ACN121979498 ACN 121979498ACN-121979498-A

Abstract

The application discloses an intelligent driving scene management method and equipment based on a configuration table, wherein the method comprises the steps of maintaining the current output state of each scene element as a part of a decision context, respectively accessing various condition inputs associated with the element into an independent condition state machine according to the condition type, outputting the current condition state according to a preset transition rule based on the previous time state, combining the current output state with a condition state set to generate condition combination information, executing matching inquiry in a state machine conversion configuration table to obtain an expected output state, and driving the output state to update and respond to action switching. Therefore, the traditional decision logic of 'branch condition stacking' is abstracted into a configurable decision framework of 'state machine + configuration table' for complex management requirements of multi-scene elements, multi-level judgment and multi-strategy switching, so that structuring, configurability and expandability of the decision logic are realized, the combination complexity is reduced, and the consistency and maintainability are improved.

Inventors

ZHU XIAOBING
YUAN LONGJIAN
LIN GUANGYI
ZHANG LINHUI

Assignees

深圳市卓驭科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260205

Claims (14)

1. An intelligent driving scene management method based on a configuration table comprises the following steps: acquiring a scene element set to be managed, and maintaining corresponding current output states for all scene elements in the scene element set, wherein the current output states are used for representing output response actions corresponding to the scene elements at the current moment; for any target scene element, acquiring at least one condition input associated with the target scene element, and calling a condition state machine corresponding to the condition type of the condition input for each condition input to perform state transition processing on the condition input so as to determine at least one condition state corresponding to the target scene element, wherein the condition state machine determines the condition state at the current moment according to a preset state transition rule based on the condition input and the state of the condition state machine at the last moment; Combining the current output state of the target scene element with the at least one condition state to generate condition combination information for characterizing a current decision context of the target scene element; executing matching inquiry in a preset state machine conversion configuration table based on the condition combination information to obtain an expected output state matched with the condition combination information, wherein the state machine conversion configuration table comprises a plurality of configuration records, and each configuration record at least comprises the condition combination information and the corresponding expected output state; And updating the current output state of the target scene element into the expected output state, and switching and executing the output response action corresponding to the expected output state.
2. The method of claim 1, wherein for the generation of the state machine transition configuration table, comprising: defining an output state machine for representing the output response action category to form an output state set, and respectively binding corresponding output response actions for each output state in the output state set in advance; Defining corresponding condition state machines for each condition type respectively to form a condition state set of the condition type, wherein the condition state machines corresponding to different condition types are mutually independent; Arranging and combining the current output state in the output state set and the condition state set of each condition type to generate a plurality of state machine decision entries covering a combination space, wherein each state machine decision entry is used for representing a group of joint values of the current output state and each condition state; and determining an expected output state corresponding to each state machine decision entry, and writing the current output state, each condition state and the expected output state into the state machine conversion configuration table.
3. The method of claim 2, wherein the writing the current output state, the condition states, and the desired output state to the state machine transition configuration table comprises: Executing field integrity check and field value validity check aiming at the current output state, each condition state and the expected output state; Mapping the checked field values of the current output state, each condition state and the expected output state into state codes which can be analyzed by a program respectively, and generating corresponding structured configuration data based on a mapping result; writing the structured configuration data into the state machine conversion configuration table to form a configuration record in the state machine conversion configuration table.
4. The method of claim 1, wherein the output response action corresponding to the desired output state is a standardized behavioral action pre-bound to the desired output state; Wherein the standardized behavioral actions include one or more of the following action types: triggering the on-site data grabbing action associated with the target scene element, and executing storage and upper transmission of the grabbed data; Outputting a human-computer interaction prompt action corresponding to the expected output state; Triggering a function limiting action or a degradation action corresponding to the desired output state.
5. The method of claim 1 or 4, wherein each configuration record is further provided with a differential supplemental field; wherein the switching executing the output response action corresponding to the expected output state includes: Determining corresponding pre-bound standardized behavior actions according to expected output states in the configuration records obtained by the matching inquiry; And if the differentiated supplementary field is detected to be a non-null field, superposing and executing the expected supplementary behavior action indicated by the differentiated supplementary field on the basis of executing the standardized behavior action.
6. The method of claim 1, wherein the at least one condition state comprises at least a debounce condition state and the condition state machine comprises a debounce condition state machine; The method for determining the condition state of the target scene element comprises the steps of obtaining at least one condition input associated with the target scene element aiming at any target scene element, calling a condition state machine corresponding to the condition type of the condition input aiming at each condition input, and performing state transition processing on the condition input to determine at least one condition state corresponding to the target scene element, wherein the steps comprise: Acquiring event duration information associated with the target scene element, and comparing the event duration information with a preset multi-level debounce threshold; Switching the debounce condition state to a slightly-jittered confirmation state when the event duration information meets a first debounce threshold and does not meet a second debounce threshold, and switching the debounce condition state to a slightly-state cleared state when the event duration information changes from meeting to not meeting the first debounce threshold; And switching the debounce condition state to a serious debounce confirmation state under the condition that the event duration information meets the second debounce threshold, and switching the debounce condition state to a serious state clearing state under the condition that the event duration information changes from meeting to not meeting the second debounce threshold.
7. The method of claim 1, wherein the at least one condition state comprises at least a mask request condition state and the condition state machine comprises a mask request condition state machine; Wherein the obtaining at least one condition input associated with the target scene element for any target scene element, and invoking, for each condition input, a condition state machine corresponding to a condition type of the condition input to perform a state transition process on the condition input to determine at least one condition state corresponding to the target scene element, includes: Receiving a shielding request instruction or an opening request instruction, and analyzing to obtain a shielding object and a shielding grade, wherein the shielding object is used for indicating at least one target scene element or used for indicating a target sensor, and the target sensor is a sensor associated with at least one scene element in the scene element set; Determining at least one target scene element as a state update object of a mask request condition state machine if the mask object is used to indicate at least one target scene element, or determining all scene elements associated with a target sensor as state update objects of a mask request condition state machine if the mask object is used to indicate the target sensor; Mapping the shielding level to a target state of the shielding request condition state machine, and updating the current state of the shielding request condition state machine corresponding to the state updating object according to the target state to obtain an updated shielding request condition state, wherein the target state comprises at least one of no adjustment, serious level shielding, slight level shielding, uniform shielding and uniform opening; and writing the updated shielding request condition state into at least one condition state corresponding to the state updating object, and combining the updated shielding request condition state with the current output state to generate the condition combination information, so that the matching query result of the state machine conversion configuration table can reflect the influence of the shielding request on the expected output state.
8. The method of claim 1, wherein after obtaining a set of scene elements to be managed and maintaining a corresponding current output state for each scene element in the set of scene elements, the method further comprises: Establishing and maintaining element state context records corresponding to each scene element in the scene element set one by one, wherein the element state context records at least comprise the current output state of the scene element and the internal state of a condition state machine instance corresponding to each condition type; The method for determining the condition state of the target scene element comprises the steps of obtaining at least one condition input associated with the target scene element aiming at any target scene element, calling a condition state machine corresponding to the condition type of the condition input aiming at each condition input, and performing state transition processing on the condition input to determine at least one condition state corresponding to the target scene element, wherein the steps comprise: Positioning the element state context records corresponding to the target scene elements one by one based on the element identifiers of the target scene elements, and reading the internal state of the condition state machine instance corresponding to each condition type from the element state context records at the last moment; Invoking a condition state machine instance corresponding to a condition type of the condition input for the at least one condition input, and inputting the condition input and the internal state of the last time to the condition state machine instance to execute state transition processing and output a condition state of the current time; and writing the updated internal state obtained by the state transition processing of the condition state at the current moment and the condition state machine instance back to the element state context record, and keeping the element state context records corresponding to other scene elements not updated.
9. A method according to claim 1 or 3, wherein, for any target scene element, the obtaining at least one condition input associated with the target scene element, and for each condition input invoking a condition state machine corresponding to a condition type of the condition input to perform a state transition process on the condition input to determine at least one condition state corresponding to the target scene element, comprises: performing periodic traversal on the scene element set according to a preset scheduling period to sequentially determine at least one target scene element in each scheduling period; Aiming at a target scene element determined in a current scheduling period, acquiring at least one condition input associated with the target scene element, and identifying a condition type corresponding to each condition input; for any condition input, acquiring the state of a condition state machine corresponding to the condition type of the condition input at the last moment, and inputting the condition input and the state of the last moment to the condition state machine so that the condition state machine completes state transition processing according to a preset state transition rule and outputs the condition state of the current moment; and collecting the condition states of the current moment output by each condition input to determine at least one condition state corresponding to the target scene element.
10. The method of claim 9, wherein the state machine conversion configuration table is stored in a key-value index structure that establishes a mapping relationship of a query key to a desired output state based on a hash index; The query key corresponding to the condition combination information is generated by splicing the state code of the current output state and the state code of the at least one condition state according to a preset coding sequence, and is used for executing index query on the key value index structure to obtain the expected output state.
11. The method of claim 1, wherein the method further comprises: Maintaining a historical condition state set corresponding to each scene element in the scene element set for the scene element; the combining the current output state of the target scene element with the at least one condition state generates condition combination information for characterizing a current decision context of the target scene element, including: Comparing at least one condition state obtained by determining the target scene element through a corresponding condition state machine with a historical condition state set corresponding to the target scene element so as to detect whether at least one condition state changes; in case that no condition state change is detected, keeping the history condition state set corresponding to the target scene element unchanged and keeping the current output state of the target scene element unchanged, and In the event that a change in at least one condition state is detected, combining the current output state of the target scene element with the current condition state set to generate the condition combination information, and updating the historical condition state set with the current condition state set.
12. A computer device comprising a memory, a processor and a computer program stored on the memory, wherein the processor executes the computer program to implement the steps of the method of any one of claims 1-11.
13. A computer program product comprising a computer program stored on a storage medium, the computer program comprising program instructions, wherein the program instructions, when executed by a computer, cause the computer to perform the steps of the method of any of claims 1-11.
14. A movable platform, comprising: A platform body, and The system comprises a platform body, a control system and a control system, wherein the control system is configured to execute the steps of the method of any one of claims 1-11 so as to manage intelligent driving scenes in the running process of the movable platform.

Description

Intelligent driving scene management method and device based on configuration table Technical Field The application relates to the technical field of intelligent driving systems, in particular to an intelligent driving scene management method and equipment based on a configuration table. Background In intelligent driving systems, a "scene understanding/scene management" module often requires continuous evaluation and response control of a large number of scene elements. Because the number of the scene elements can reach hundreds or more, and different operation requirements such as mass production experience and shadow mode are required to be considered, a single scene element is generally divided into multiple processing levels (such as slight, medium, serious, etc.), and different triggering time and response targets are respectively corresponding to the different triggering time and response targets. Meanwhile, the service side usually also provides shielding or enabling requirements of 'per element/per grade', namely, the same set of grade system may need to shield, release or forcedly switch the judging result of part of grades under different elements, different vehicle model versions or different strategy stages, so as to form a large number of engineering rule sets of 'condition combination+behavior switching'. At present, it is more common practice to solidify these combinational logics in code in the form of a large number of branch sentences or superposition conditions, but hidden corner case (corner case) risks are often introduced, and problems of excessively high circle complexity, difficult completion of test, difficult evaluation of modification influence range, high iterative maintenance cost and the like are caused, so that systematic quality and delivery risks are formed. Disclosure of Invention The embodiment of the application provides an intelligent driving scene management method, equipment, a storage medium and a program product based on a configuration table, which are used for at least solving one of the technical problems. In a first aspect, an embodiment of the application provides an intelligent driving scene management method based on a configuration table, which comprises the steps of obtaining a scene element set to be managed, maintaining corresponding current output states for all scene elements in the scene element set, combining the current output states of the target scene elements with at least one condition state according to preset state transition rules, generating condition combination information for representing the current context of the target scene elements, aiming at any target scene element, obtaining at least one condition input associated with the target scene element, calling a condition state machine corresponding to the condition type of the condition input for each condition input, performing state transition processing on the condition input to determine at least one condition state corresponding to the target scene element, wherein the condition state machine is used for determining the condition state of the current moment according to preset state transition rules based on the condition input and the state of the condition state machine at the last moment, performing state transition processing on the condition input in a preset state machine conversion configuration table based on the condition combination information, and performing state transition processing on the condition state conversion table corresponding to the condition input, and recording the expected state of the expected state, and outputting the expected state. In a second aspect, embodiments of the present application provide a storage medium having stored therein one or more programs including execution instructions that are readable and executable by an electronic device (including, but not limited to, a computer, a server, or a network device, etc.) for performing the steps of the method of any of the above applications. In a third aspect, a computer device is provided comprising at least one processor and a memory communicatively coupled to the at least one processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the steps of the method of any one of the preceding claims. In a fourth aspect, embodiments of the present application also provide a computer program product comprising a computer program stored on a storage medium, the computer program comprising program instructions which, when executed by a computer, cause the computer to perform the steps of the method of any of the preceding claims. In a fifth aspect, an embodiment of the present application further provides a mobile platform, including a platform body, and a control system disposed on the platform body, where the control system is configured to perform the steps of any one of the methods described above, so as to manage an intellig