An automatic driving planning method for target intention ambiguity

Abstract

The application provides an automatic driving planning method for target intention ambiguity, and belongs to the technical field of automatic driving. The automatic driving planning method for target intention ambiguity is applied to an automatic driving vehicle and comprises the following steps: obtaining an intention prediction result of a neighboring target vehicle, wherein the intention prediction result comprises multiple predicted driving intentions of the target vehicle and probability values corresponding to the predicted driving intentions; calculating an intention uncertainty index of the target vehicle according to the intention prediction result; in response to the intention uncertainty index satisfying a preset triggering condition and a collision time with the target vehicle being less than a preset safety threshold, activating an observation mode; in response to the observation mode being activated, constraining a longitudinal speed according to a longitudinal speed of the target vehicle; and in response to determining that the intention prediction result of the target vehicle is stable based on a sliding window mechanism, exiting the observation mode. When it is identified that the intention of a neighboring vehicle is ambiguous, the application takes over the speed planning authority and executes a companion observation strategy, thereby improving the safety of automatic driving.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, and in particular to an autonomous driving planning method for situations where the target intent is ambiguous. Background Technology

[0002] In the operation of autonomous vehicles (EVs), lane change decisions and longitudinal speed planning heavily rely on the prediction of the intentions of surrounding vehicles (TVs). Current prediction modules typically output multiple trajectories of surrounding vehicles over the next few seconds and their corresponding probabilities (e.g., 30% for left lane change, 30% for right lane change, 40% for lane keeping). When the probabilities of various intentions are very close (i.e., the prediction entropy is high), if the decision-making module simply selects the intention with the highest probability for planning, misjudgments are highly likely. If the autonomous vehicle rashly chooses to accelerate and overtake in this ambiguous state, a sudden lane change by a surrounding vehicle could lead to a serious collision risk or sudden braking, affecting comfort. Existing algorithms often lack an observation mechanism, i.e., they do not induce or wait for surrounding vehicles to reveal their true intentions through their own behavior (such as yielding speed without yielding lanes, or following the vehicle's path). Summary of the Invention

[0003] This invention aims to solve at least one of the technical problems existing in the prior art, and proposes an autonomous driving planning method for scenarios with fuzzy target intent that is optimized for scenarios with high difficulty in intent prediction.

[0004] In a first aspect, embodiments of the present invention provide an autonomous driving planning method for vehicles with fuzzy target intentions, applied to autonomous vehicles, comprising: acquiring the intention prediction results of adjacent target vehicles, the intention prediction results including multiple predicted driving intentions of the target vehicles and probability values ​​corresponding to the predicted driving intentions; calculating the intention uncertainty index of the target vehicles based on the intention prediction results; activating an observation mode in response to the intention uncertainty index satisfying a preset triggering condition and the collision time with the target vehicle being less than a preset safety threshold; constraining the longitudinal speed based on the longitudinal speed of the target vehicles in response to activating the observation mode; and exiting the observation mode in response to determining that the intention prediction results of the target vehicles are stable based on a sliding window mechanism.

[0005] According to an embodiment of the present invention, the intent uncertainty index includes information entropy and probability difference; the intent uncertainty index of the target vehicle is calculated based on the intent prediction result, including: calculating the information entropy of the target vehicle based on multiple predicted driving intentions and the probability values ​​corresponding to the predicted driving intentions; and calculating the difference between the largest probability value and the second largest probability value to obtain the probability difference.

[0006] According to an embodiment of the present invention, activating the observation mode further includes: determining whether the intention uncertainty index meets a preset triggering condition, specifically including: determining whether the information entropy is greater than a preset entropy threshold; if so, determining that the first preset triggering condition is met; if not, determining that the first preset triggering condition is not met; determining whether the probability difference is less than a preset probability difference threshold; if so, determining that the second preset triggering condition is met; if not, determining that the second preset triggering condition is not met.

[0007] According to an embodiment of the present invention, determining the intention prediction result of a target vehicle based on a sliding window mechanism includes: establishing a sliding window of a preset frame length; within the sliding window, determining the type of the predicted driving intention with the highest probability at each moment frame by frame, and accumulating the number of frames in which each type appears within the sliding window, wherein the type includes at least one of cutting in and lane keeping; in response to the proportion of the accumulated number of frames of cutting in to the total number of frames in the sliding window exceeding a first proportion threshold, determining that the predicted driving intention of the target vehicle is cutting in, and confirming that the intention prediction result is stable; in response to the proportion of the accumulated number of frames of lane keeping to the total number of frames in the sliding window exceeding a second proportion threshold, determining that the predicted driving intention of the target vehicle is lane keeping, and confirming that the intention prediction result is stable.

[0008] According to an embodiment of the present invention, the method further includes managing decision states through a finite state machine. The decision states include a default cruise state, a warning monitoring state, an observation state, a clear yielding state, and a resumed overtaking state. The method includes: in the default cruise state, controlling the vehicle to drive according to a conventional preset cruise logic; in response to detecting that a target vehicle enters a preset region of interest or that an intent uncertainty indicator meets an initial triggering condition, switching to the warning monitoring state to continuously monitor the intent uncertainty indicator, wherein the threshold of the initial triggering condition is less than the threshold of a preset triggering condition; in response to the intent uncertainty indicator meeting a preset triggering condition and the collision time with the target vehicle being less than a preset safety threshold in the warning monitoring state, switching to the observation state to activate the observation mode to constrain longitudinal speed; in response to determining, based on a sliding window mechanism, that the predicted driving intent of the target vehicle is to enter in the observation state, switching to the clear yielding state to execute deceleration and yielding logic; and in response to determining, based on a sliding window mechanism, that the predicted driving intent of the target vehicle is to maintain lane in the observation state, switching to the resumed overtaking state to release longitudinal speed constraints and execute overtaking.

[0009] According to an embodiment of the present invention, the method further includes: in the observation mode, counting the duration of the observation mode; in response to the duration being greater than the preset maximum observation duration, forcibly exiting the observation mode and switching to the overtaking state to increase the following distance with the target vehicle.

[0010] According to an embodiment of the present invention, constraining the longitudinal speed based on the longitudinal speed of the target vehicle includes: obtaining the relative distance between the target vehicle and the target vehicle and the dynamic safety observation distance, wherein the dynamic safety observation distance is determined according to the intention uncertainty index; calculating the upper limit of the longitudinal speed based on the difference between the longitudinal speed of the target vehicle, the relative distance and the dynamic safety observation distance and the speed adjustment gain coefficient; comparing the upper limit of the longitudinal speed with the road speed limit and taking the smaller value as the longitudinal speed command.

[0011] A second aspect of the present invention provides an autonomous driving planning system with fuzzy target intent, which can be used to implement the above-mentioned autonomous driving planning method with fuzzy target intent, comprising: an intent prediction module, configured to acquire intent prediction results of adjacent target vehicles, the intent prediction results including multiple predicted driving intentions of the target vehicles and probability values ​​corresponding to the predicted driving intentions; a fuzzy determination module, configured to calculate an intent uncertainty index of the target vehicles based on the intent prediction results; an activation observation module, configured to activate an observation mode in response to the intent uncertainty index satisfying a preset trigger condition and the collision time with the target vehicle being less than a preset safety threshold; a speed constraint module, configured to constrain the longitudinal speed of the target vehicles based on the longitudinal speed in response to the activation observation mode; and an exit observation module, configured to exit the observation mode in response to the determination that the intent prediction results of the target vehicles are stable based on a sliding window mechanism.

[0012] A third aspect of the present invention provides an electronic device comprising: one or more processors; and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to perform the aforementioned autonomous driving planning method with ambiguous target intent.

[0013] A fourth aspect of the present invention also provides a computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, cause the processor to perform the aforementioned autonomous driving planning method with ambiguous target intent.

[0014] The autonomous driving planning method provided by this invention overcomes the problem of misjudgment caused by relying on a single maximum probability in existing technologies by introducing an intent uncertainty index as an activation criterion; it achieves accompanying navigation when intent is ambiguous by constraining longitudinal speed in observation mode, avoiding the collision risk caused by rash overtaking; and it ensures the stability of exit decisions through a sliding window mechanism. Therefore, it at least partially solves the misjudgment problem caused by ambiguous target vehicle intent, improving the stability of autonomous driving. Attached Figure Description

[0015] Figure 1 A flowchart illustrating an autonomous driving planning method with ambiguous target intent, provided as an embodiment of the present invention;

[0016] Figure 2 This is a flowchart illustrating an optional specific implementation method of step S5 in an embodiment of the present invention;

[0017] Figure 3 This is a schematic diagram of the management decision of the finite state machine in an embodiment of the present invention;

[0018] Figure 4 A structural block diagram of an autonomous driving planning system with ambiguous target intent provided in an embodiment of the present invention;

[0019] Figure 5 This is a structural block diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0020] To enable those skilled in the art to better understand the technical solutions of the present invention, exemplary embodiments of the present invention are described below in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to aid understanding. These should be considered merely exemplary. Therefore, those skilled in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0021] Where there is no conflict, the various embodiments of the present invention and the features thereof may be combined with each other.

[0022] As used herein, the term “and / or” includes any and all combinations of one or more related enumerated entries.

[0023] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used herein, the singular forms “a” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will also be understood that when the terms “comprising” and / or “made of” are used in this specification, the presence of the stated feature, integral, step, operation, element, and / or component is specified, but the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or groups thereof is not excluded. Terms such as “connected” or “linked” are not limited to physical or mechanical connections but can include electrical connections, whether direct or indirect.

[0024] Unless otherwise specified, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art. It will also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having the meaning consistent with their meaning in the context of the relevant art and the invention, and will not be interpreted as having an idealized or overly formal meaning unless expressly so defined herein.

[0025] In the technical solution of this invention, the collection, storage, use, processing, transmission, provision, and disclosure of user personal information all comply with relevant laws and regulations and do not violate public order and good morals. The use of user data in this technical solution follows relevant national laws and regulations (e.g., the "Information Security Technology - Personal Information Security Specification"). For example: appropriate measures are taken for personal information access control; restrictions are imposed on the display of personal information; the purpose of using personal information does not exceed the scope of direct or reasonable association; and explicit identity targeting is eliminated when using personal information to avoid precisely locating a specific individual.

[0026] Figure 1 This is a flowchart illustrating an autonomous driving planning method with fuzzy target intent, as provided in an embodiment of the present invention. Figure 1 As shown, this embodiment of the invention provides an autonomous driving planning method for vehicles with fuzzy target intentions, applied to autonomous vehicles, including: S1, obtaining the intention prediction results of adjacent target vehicles, the intention prediction results including multiple predicted driving intentions of the target vehicles and the probability values ​​corresponding to the predicted driving intentions; S2, calculating the intention uncertainty index of the target vehicles based on the intention prediction results; S3, activating the observation mode in response to the intention uncertainty index meeting a preset trigger condition and the collision time with the target vehicle being less than a preset safety threshold; S4, constraining the longitudinal speed based on the longitudinal speed of the target vehicles in response to the activation of the observation mode; S5, executing the corresponding control decision based on the stable intention prediction results after exiting the observation mode in response to the determination that the intention prediction results of the target vehicles are stable based on a sliding window mechanism.

[0027] It should be noted that, optionally, the collision time is usually calculated based on a longitudinal kinematics model, obtaining the relative longitudinal distance and relative longitudinal approach speed between the vehicle and the target vehicle, i.e., the difference between the vehicle's speed and the target vehicle's speed, to calculate the collision time. If the vehicle is not approaching the target vehicle at that moment, it is considered that there is no collision risk, and the TTC is taken as infinity.

[0028] In this embodiment, the Defensive Observation Arbiter proposed in this invention is embedded in the planning and control layer of the autonomous driving system. When the intentions of adjacent vehicles are unclear, it takes over the speed planning authority and executes an accompanying observation strategy. This technical solution is specifically optimized for three types of scenarios with high difficulty in intention prediction. The first type is the ramp merging scenario, where the adjacent vehicle is at the end of the acceleration lane and hesitates between accelerating to cut in or decelerating to merge. The second type is the congestion game scenario, where the adjacent vehicle slightly deviates from its lane to probe, its intention oscillating between cutting in and giving up. The third type is the lane reduction scenario, where road construction or lane merging forces the adjacent vehicle to change lanes, but the specific timing is uncertain. A common feature of these three scenarios is that the probability distribution of the target vehicle's intention tends to be uniform, making it difficult for traditional decision-making methods based on maximum probability to provide reliable judgments.

[0029] The embodiments of this invention address the collision risk arising from uncertain decision-making when predicted probabilities are close. By actively maintaining distance observation while limiting speed, sufficient braking buffer space is reserved for potential target vehicles to cut in, mimicking the handling of an experienced driver when hesitating before encountering another vehicle (relaxing the accelerator, preparing to brake, and observing), rather than mechanically adhering to lane speed limits. The vehicle's decision not to overtake itself is a signal, helping other drivers understand their right-of-way and make a final decision more quickly, thus improving traffic efficiency and significantly enhancing the safety of autonomous driving.

[0030] Based on the above embodiments, the intent uncertainty index includes information entropy and probability difference; according to the intent prediction result, the intent uncertainty index of the target vehicle is calculated, including: calculating the information entropy of the target vehicle based on multiple predicted driving intentions and the probability values ​​corresponding to the predicted driving intentions; and calculating the difference between the largest probability value and the second largest probability value to obtain the probability difference.

[0031] In this embodiment, the key target screening and uncertainty quantification include: the system first screens out the nearest target vehicle (OT) in the left and right adjacent lanes, and then mathematically quantifies the uncertainty of its intent.

[0032] (1) Calculation of the difference in intent probability:

[0033] Δp = p_max - p_2nd, where: Δp represents the discrimination index; p_max is the probability with the largest value among all intent probabilities output by the prediction module; and p_2nd is the probability with the second largest value.

[0034] When Δp is less than a preset threshold (e.g., 0.2), the determination of intent is low and ambiguous.

[0035] (2) Calculation of information entropy (Shannon Entropy):

[0036] H(P) = -∑[p_k × log2(p_k)], where: H(P) represents the intention information entropy; K is the total number of possible intention categories (e.g., left lane change, right lane change, lane keeping, then K = 3); p_k is the probability value corresponding to the k-th intention.

[0037] The value of information entropy increases as the distribution of the probability values ​​tends to be uniform. The larger H(P) is, the more confused the system's judgment of the target's intention is, and the higher the uncertainty is.

[0038] The embodiments of this invention provide two specific methods for quantifying the uncertainty index of intent: information entropy reflects the overall disorder of the probability distribution, and probability difference reflects the distinguishability between the highest probability intents. This quantifies ambiguity from different dimensions.

[0039] Based on the above embodiments, before activating the observation mode, the method further includes: determining whether the intent uncertainty index meets the preset triggering conditions, specifically including: determining whether the information entropy is greater than the preset entropy threshold; if so, determining that the first preset triggering condition is met; if not, determining that the first preset triggering condition is not met; determining whether the probability difference is less than the preset probability difference threshold; if so, determining that the second preset triggering condition is met; if not, determining that the second preset triggering condition is not met.

[0040] It should be noted that, optionally, the preset triggering condition can be a first preset triggering condition, a second preset triggering condition, or a combination of the first preset triggering condition and the second preset triggering condition. That is, the observation mode can be activated as long as the information entropy is greater than the preset entropy threshold or the probability difference is less than the preset probability difference threshold.

[0041] Through the embodiments of the present invention, the greater the information entropy and the smaller the probability difference, the more ambiguous the intention. Quantifying this indicator facilitates subsequent decision-making.

[0042] Figure 2 This is a flowchart illustrating an optional specific implementation method of step S5 in an embodiment of the present invention, such as... Figure 2 As shown, the intention prediction result of the target vehicle is determined based on the sliding window mechanism, including: S51, establishing a sliding window with a preset frame length; S52, within the sliding window, determining the type of the predicted driving intention with the highest probability at each moment frame by frame, and accumulating the number of frames in which each type appears within the sliding window, the type including at least one of cutting in and lane keeping; S53, if the proportion of the cumulative number of frames of cutting in to the total number of frames in the sliding window exceeds a first proportion threshold, then the predicted driving intention of the target vehicle is determined to be cutting in, confirming that the intention prediction result is stable; S54, if the proportion of the cumulative number of frames of lane keeping to the total number of frames in the sliding window exceeds a second proportion threshold, then the predicted driving intention of the target vehicle is determined to be lane keeping, confirming that the intention prediction result is stable.

[0043] In this embodiment, the intent confirmation and exit mechanism (anti-jitter) is as follows:

[0044] To avoid vehicle jerking caused by abrupt changes in the predicted signal, a sliding window confirmation method is adopted. A time window of length N frames is established, and the cumulative frame count for each intention is calculated: Count_type=∑[I(Pred_t==Type)].

[0045] Where: N is the sliding window length; Pred_t is the intent type with the highest probability at time t; I(...) is the indicator function, which is 1 when the condition is met and 0 otherwise.

[0046] The exit criteria are as follows:

[0047] If Count_cut_in > 0.8 × N, the intention is clearly defined as cutting in, and the process transitions to a clear yielding state.

[0048] If Count_keep > 0.8 × N, the intention is clearly to maintain, and the system transitions to the overtaking state.

[0049] Through the embodiments of the present invention, by accumulating the number of occurrence frames of each intent type within a sliding window and using a proportional threshold as a stability criterion, the jitter of the prediction signal is filtered out, preventing the observation mode from erroneously exiting due to prediction jumps in a single frame or a few frames, and accurately determining the stability of the target vehicle's intent.

[0050] Figure 3 This is a schematic diagram of the management decision of the finite state machine in an embodiment of the present invention, such as... Figure 3 As shown, the method further includes managing decision states through a finite state machine. These decision states include a default cruise state, a warning monitoring state, an observation state, a clear yielding state, and a resumed overtaking state. Specifically: S61, in the default cruise state, control follows the conventional preset cruise logic; S62, in response to detecting that a target vehicle enters a preset region of interest or that an intent uncertainty indicator meets the initial trigger condition, jump to the warning monitoring state to continuously monitor the intent uncertainty indicator, where the threshold of the initial trigger condition is less than the threshold of the preset trigger condition; S63, in response to the intent uncertainty indicator meeting the preset trigger condition and the collision time with the target vehicle being less than a preset safety threshold in the warning monitoring state, jump to the observation state to activate the observation mode to constrain longitudinal speed; S64, in response to determining, based on a sliding window mechanism, that the target vehicle's predicted driving intent is to enter in the observation state, jump to the clear yielding state to execute deceleration and yielding logic; S65, in response to determining, based on a sliding window mechanism, that the target vehicle's predicted driving intent is to keep lane in the observation state, jump to the resumed overtaking state to release the longitudinal speed constraint and execute overtaking.

[0051] In this embodiment, to ensure the consistency of control decisions, a Finite State Machine (FSM) is adopted to manage the lifecycle of observation behaviors:

[0052] (1) IDLE (default cruise state):

[0053] The ego vehicle travels according to the conventional ACC / LCC logic, with no high-risk targets. The transition condition is that when a target enters the ROI region or H(P)>0.5, the state transitions to MONITOR

[0054] (2) MONITOR (early warning monitoring state):

[0055] The entropy value H(P) is continuously monitored in the background, and no control intervention is performed at this time. The transition condition is that when H(P)>H_high_thres (high uncertainty) and TTC<TTC_safe (collision risk exists) are satisfied, the state transitions to OBSERVE.

[0056] (3) OBSERVE (observation state):

[0057] Defensive control is activated to limit the speed of the ego vehicle and prohibit overtaking. When the intention is clearly determined to be cut-in, the state transitions to YIELD; when the intention is clearly determined to be lane keeping or the observation times out, the state transitions to RESUME.

[0058] (4) YIELD (clear yielding state):

[0059] It is confirmed that the target cuts in, and the conventional lane change deceleration logic is executed. The transition condition is that after the target completes the lane change and becomes a preceding vehicle, the state transitions to IDLE.

[0060] (5) RESUME (resume / overtake state):

[0061] It is confirmed that the target does not cut in, the speed limit is lifted, and overtaking is executed. The transition condition is that after the overtaking action is completed, the state transitions to IDLE.

[0062] Through the embodiment of the present invention, a finite state machine is introduced to manage the complete lifecycle of decision states, and the orderly transition of the five states ensures the consistency of system decisions and avoids control mutations caused by state jumps.

[0063] On the basis of the above embodiment, the method further includes: in the observation mode, counting the duration of the observation mode; in response to the duration being greater than a preset maximum observation duration, forcibly exiting the observation mode and transitioning to the resume overtake state to increase the following distance from the target vehicle.

[0064] In the embodiments of this invention, if the stability condition is not met even after the state duration T_duration > T_max (e.g., 5 seconds), the target is determined to be abnormal, and the distance between the vehicle and the target is increased. A timeout fallback mechanism for the observation mode is provided to prevent the vehicle from being trapped in the observation mode due to the target vehicle not revealing a clear intention for an extended period.

[0065] Based on the above embodiments, the longitudinal speed is constrained according to the longitudinal speed of the target vehicle, including: obtaining the relative distance between the target vehicle and the target vehicle and the dynamic safety observation distance, wherein the dynamic safety observation distance is determined according to the intention uncertainty index; calculating the upper limit of the longitudinal speed based on the difference between the longitudinal speed of the target vehicle, the relative distance and the dynamic safety observation distance and the speed adjustment gain coefficient; comparing the upper limit of the longitudinal speed with the road speed limit and taking the smaller value as the longitudinal speed command.

[0066] In this embodiment, when in the OBSERVE state, the planner executes the following formula to calculate the vehicle's target speed limit, ensuring that it accompanies but does not exceed it:

[0067] v_limit=v_OT+sign(d_rel-d_safe)×k_v×|d_rel-d_safe|;

[0068] v_cmd=min(v_cruise,v_limit);

[0069] Where: v_cmd is the final speed command output to the controller; v_cruise is the speed set by the driver or the road speed limit; v_OT is the current longitudinal speed of the observed object; d_rel is the actual longitudinal distance between the vehicle and the observed object; d_safe is the dynamic safe observation distance (this value is dynamically adjusted in a positive correlation with the entropy value H(P); k_v is the preset speed adjustment gain coefficient. Sign() is the sign function used to return the positive or negative sign of the input value.

[0070] Specifically, when the relative distance is greater than the dynamic safe observation distance, the upper limit of the longitudinal speed is greater than the longitudinal speed of the target vehicle; when the relative distance is less than the dynamic safe observation distance, the upper limit of the longitudinal speed is less than the longitudinal speed of the target vehicle, and a safe observation distance is maintained without overtaking.

[0071] The embodiments of the present invention provide a specific control strategy for longitudinal speed constraints in the observation mode. The strategy uses the target vehicle speed as a reference, adjusts the distance based on the deviation between the actual distance and the safe observation distance, and takes the smaller value between the upper speed limit and the road speed limit as the final instruction.

[0072] It should be noted that the input and output data streams of this solution encompass three categories: external input signals, internal computation signals, and output signals. Regarding external input signals, the system receives four inputs: the Ego_State signal provides the vehicle's current motion state, including longitudinal speed, acceleration, and positioning information; the Map_Info signal provides static environmental information from a high-precision map, including the distance to the merging point of the upcoming ramp and the lane line type; the Pred_Result signal is the set of target vehicle intent probabilities and the corresponding predicted trajectory output by the upstream prediction module; and the Obj_List signal is the list of targets in the region of interest after perception fusion, containing target-level information such as relative distance and relative speed. Regarding internal computation signals, the system calculates the Uncertainty_Idx signal based on the above inputs, which is the intent uncertainty index calculated by this module. This index is quantified based on information entropy or probability difference and is used for triggering judgments in subsequent observation modes. Regarding output signals, the system outputs four signals: the Obs_Mode_Active signal is the observation mode activation flag, used to notify the decision layer whether it is currently in observation mode; the Speed_Limit signal is the dynamically recommended speed limit generated in observation mode, used to constrain the vehicle's longitudinal speed planning; the Virtual_Obs signal is virtual obstacle information mapped to the center line of the current lane, used for following path planning; and the FSM_State signal is the current finite state machine state enumeration value, used for debugging and human-machine interface display. This interface design allows the present invention to be embedded in the planning and control layer of existing autonomous driving systems in a modular manner, achieving decoupling and integration with the upstream prediction module, perception fusion module, and downstream control execution module.

[0073] Figure 4 A structural block diagram of an autonomous driving planning system with fuzzy target intent provided in an embodiment of the present invention is shown below. Figure 4 As shown, a second aspect of the present invention provides an autonomous driving planning system with fuzzy target intent, which can be used to implement the above-mentioned autonomous driving planning method with fuzzy target intent, comprising: an intent prediction module, used to acquire intent prediction results of adjacent target vehicles, the intent prediction results including multiple predicted driving intentions of the target vehicles and probability values ​​corresponding to the predicted driving intentions; a fuzzy determination module, used to calculate the intent uncertainty index of the target vehicle based on the intent prediction results; an activation observation module, used to activate an observation mode in response to the intent uncertainty index satisfying a preset trigger condition and the collision time with the target vehicle being less than a preset safety threshold; a speed constraint module, used to constrain the longitudinal speed based on the longitudinal speed of the target vehicle in response to the activation observation mode; and an exit observation module, used to exit the observation mode in response to the determination that the intent prediction result of the target vehicle is stable based on a sliding window mechanism.

[0074] Based on the same inventive concept, embodiments of the present invention also provide an electronic device. Figure 5 This is a structural block diagram of an electronic device provided in an embodiment of the present invention. Figure 5 As shown, an embodiment of the present invention provides an electronic device including: one or more processors 101, a memory 102, and one or more I / O interfaces 103. The memory 102 stores one or more programs, which, when executed by the one or more processors, enable the one or more processors to implement an autonomous driving planning method with ambiguous target intent as described in any of the above embodiments; the one or more I / O interfaces 103 are connected between the processor and the memory, configured to enable information interaction between the processor and the memory.

[0075] The processor 101 is a device with data processing capabilities, including but not limited to a central processing unit (CPU); the memory 102 is a device with data storage capabilities, including but not limited to random access memory (RAM, more specifically SDRAM, DDR, etc.), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), and flash memory (FLASH); the I / O interface (read / write interface) 103 is connected between the processor 101 and the memory 102, and can realize information interaction between the processor 101 and the memory 102, including but not limited to a data bus (Bus).

[0076] In some embodiments, the processor 101, memory 102, and I / O interface 103 are interconnected via bus 104, and thus connected to other components of the computing device.

[0077] In some embodiments, the one or more processors 101 include a field-programmable gate array.

[0078] This invention also provides a computer-readable medium. The computer-readable medium stores a computer program, which, when executed by a processor, implements the steps in any of the above-described autonomous driving planning methods for vehicles with ambiguous target intentions. The computer-readable storage medium may be volatile or non-volatile.

[0079] This invention also provides a computer program product, including computer-readable code, or a non-volatile computer-readable storage medium carrying computer-readable code. When the computer-readable code is run in the processor of an electronic device, the processor in the electronic device executes the aforementioned autonomous driving planning method with ambiguous target intent.

[0080] Those skilled in the art will understand that all or some of the steps, systems, and apparatuses disclosed above, and their functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware implementations, the division between functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software can be distributed on a computer-readable storage medium, which may include computer storage media (or non-transitory media) and communication media (or transient media).

[0081] As is known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information, such as computer-readable program instructions, data structures, program modules, or other data. Computer storage media includes, but is not limited to, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), static random access memory (SRAM), flash memory or other memory technologies, portable compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, it is known to those skilled in the art that communication media typically contain computer-readable program instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0082] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0083] The computer program instructions used to perform the operations of this invention may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing state information from the computer-readable program instructions. This electronic circuitry can execute the computer-readable program instructions to implement various aspects of the invention.

[0084] The computer program product described herein can be implemented specifically through hardware, software, or a combination thereof. In one alternative embodiment, the computer program product is specifically embodied in a computer storage medium; in another alternative embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0085] Various aspects of the present invention are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer-readable program instructions.

[0086] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that, when executed by the processor of the computer or other programmable data processing apparatus, they create means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium that causes a computer, programmable data processing apparatus, and / or other device to operate in a particular manner; thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing aspects of the functions / actions specified in one or more blocks of the flowchart and / or block diagram.

[0087] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to perform the functions / actions specified in one or more boxes of a flowchart and / or block diagram.

[0088] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of an instruction, which contains one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than those shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0089] Example embodiments have been disclosed herein, and while specific terminology has been used, it is for illustrative purposes only and should be construed as such, and is not intended to be limiting. In some instances, it will be apparent to those skilled in the art that features, characteristics, and / or elements described in conjunction with particular embodiments may be used alone, or in combination with features, characteristics, and / or elements described in conjunction with other embodiments, unless otherwise expressly indicated. Therefore, those skilled in the art will understand that various changes in form and detail may be made without departing from the scope of the invention as set forth in the appended claims.

Claims

1. An autonomous driving planning method with fuzzy target intent, characterized in that, Applied to autonomous vehicles, including: Obtain the intention prediction results of adjacent target vehicles, the intention prediction results including multiple predicted driving intentions of the target vehicles and the probability values ​​corresponding to the predicted driving intentions; Based on the intent prediction results, calculate the intent uncertainty index of the target vehicle; If the intent uncertainty index meets the preset triggering condition and the collision time with the target vehicle is less than the preset safety threshold, then the observation mode is activated. In response to activating the observation mode, the longitudinal velocity is constrained based on the longitudinal velocity of the target vehicle; If the intention prediction result of the target vehicle is determined to be stable based on the sliding window mechanism, the observation mode is exited.

2. The method according to claim 1, wherein, The intent uncertainty index includes information entropy and probability difference; the calculation of the intent uncertainty index of the target vehicle based on the intent prediction result includes: Based on the various predicted driving intentions and the probability values ​​corresponding to the predicted driving intentions, the information entropy of the target vehicle is calculated; The probability difference is obtained by calculating the difference between the largest and second largest probability values.

3. The method according to claim 2, wherein, The activation of the observation mode previously included: Determining whether the intent uncertainty index meets the preset triggering conditions specifically includes: Determine whether the information entropy is greater than a preset entropy threshold. If yes, it is determined that the first preset triggering condition is met; otherwise, it is determined that the first preset triggering condition is not met. Determine whether the probability difference is less than a preset probability difference threshold. If yes, then the second preset triggering condition is satisfied; otherwise, the second preset triggering condition is not satisfied.

4. The method according to claim 1, wherein, The determination of the target vehicle's intent prediction result based on the sliding window mechanism includes: Create a sliding window with a preset frame length; Within the sliding window, the type of the predicted driving intention with the highest probability at each moment is determined frame by frame, and the number of frames in which each type appears within the sliding window is accumulated. The type includes at least one of cutting in and lane keeping. If the cumulative number of frames indicating a cut-in intention exceeds a first proportion threshold, then the predicted driving intention of the target vehicle is determined to be a cut-in, and the intention prediction result is confirmed to be stable. If the cumulative number of frames responding to lane-keeping intentions exceeds a second threshold, the predicted driving intention of the target vehicle is determined to be lane-keeping, confirming that the intention prediction result is stable.

5. The method according to claim 4, wherein, The method further includes managing decision states through a finite state machine, the decision states including a default cruise state, a warning monitoring state, an observation state, an explicit yielding state, and a resume overtaking state, including: In the default cruise mode, the control follows the conventional preset cruise logic. In response to detecting that the target vehicle enters a preset region of interest or that the intent uncertainty indicator meets the initial triggering condition, the system switches to an early warning monitoring state to continuously monitor the intent uncertainty indicator. The threshold of the initial triggering condition is less than the threshold of the preset triggering condition. In response to the intention uncertainty index meeting the preset trigger condition and the collision time being less than the preset safety threshold under the warning monitoring state, the system jumps to the observation state to activate the observation mode to constrain the longitudinal velocity. In response to the observation state where the predicted driving intention of the target vehicle is determined to be "cut in" based on the sliding window mechanism, the system jumps to the explicit yield state to execute the deceleration and yield logic. In response to the observation state where the predicted driving intention of the target vehicle is lane keeping based on the sliding window mechanism, the system jumps to the overtaking state to release the longitudinal speed constraint and perform overtaking.

6. The method according to claim 5, wherein, The method further includes: Under the observation mode, the duration of the observation mode is recorded. If the duration exceeds the preset maximum observation time, the observation mode is forcibly exited, and the system switches to overtaking mode to increase the following distance with the target vehicle.

7. The method according to claim 1, wherein, The step of constraining the longitudinal speed based on the longitudinal speed of the target vehicle includes: The relative distance to the target vehicle and the dynamic safety observation distance are obtained, wherein the dynamic safety observation distance is determined based on the intent uncertainty index; The upper limit of the longitudinal speed is calculated based on the longitudinal speed of the target vehicle, the difference between the relative distance and the dynamic safety observation distance, and the speed adjustment gain coefficient; The longitudinal speed limit is compared with the road speed limit, and the smaller value is taken as the longitudinal speed command.

8. An autonomous driving planning system with fuzzy target intent, characterized in that, The method described in claim 1 can be implemented by: The intent prediction module is used to obtain the intent prediction results of adjacent target vehicles. The intent prediction results include multiple predicted driving intentions of the target vehicles and the probability values ​​corresponding to the predicted driving intentions. The fuzzy determination module is used to calculate the intent uncertainty index of the target vehicle based on the intent prediction result. The observation module is activated in response to the intention uncertainty index meeting a preset trigger condition and the collision time with the target vehicle being less than a preset safety threshold. A speed constraint module is used to constrain the longitudinal speed based on the longitudinal speed of the target vehicle in response to activating the observation mode. Exit the observation module in response to the determination that the intention prediction result of the target vehicle is stable based on the sliding window mechanism, and then exit the observation mode.

9. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1 to 7.

10. A computer-readable medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 7.

Images