Lane change control method, electronic device, and vehicle

Abstract

The application provides a lane changing control method, an electronic device and a vehicle. The method comprises: in response to a navigation instruction prompting lane changing on a ramp, acquiring a candidate lane changing area corresponding to a driving environment of the vehicle; checking the candidate lane changing area by using a lane topology checking strategy based on a navigation map to obtain a map matching confidence; checking the candidate lane changing area by using an intention prediction checking strategy based on a historical trajectory to obtain an intention confidence; checking a feature in the candidate lane changing area by using a feature recognition checking strategy based on perception to obtain a feature risk value; determining a lane changing risk value based on the map matching confidence, the intention confidence and the feature risk value; and based on the lane changing risk value, controlling the vehicle to drive along the candidate lane changing area in a case where the candidate lane changing area is determined to be a low-risk area. The above method can improve the safety of automatic driving.

Description

Technical Field

[0001] This application belongs to the field of vehicles and relates to autonomous driving decision-making technology, and in particular to a lane change control method, electronic equipment and vehicle. Background Technology

[0002] In the field of autonomous driving decision-making and planning, when a vehicle needs to enter an exit ramp from the main road, it typically relies on navigation instructions, high-precision maps, or real-time perception information to determine whether to leave the main road. However, these methods are difficult to flexibly handle newly added temporary areas on the road, which can easily lead to vehicles mistakenly entering such areas. Summary of the Invention

[0003] This application discloses a lane change control method, electronic device, and vehicle, which solves the technical problem that related technologies cannot detect temporary areas in a timely manner to avoid mistakenly entering such areas.

[0004] This application provides a lane change control method applied to an electronic device in a vehicle. The method includes: in response to a navigation command prompting a lane change at a ramp, acquiring a candidate lane change area corresponding to the vehicle's driving environment; verifying the candidate lane change area using a lane topology verification strategy based on a navigation map to obtain a map matching confidence score; verifying the candidate lane change area using an intent prediction verification strategy based on historical trajectories to obtain an intent confidence score; verifying features in the candidate lane change area using a perception-based feature recognition verification strategy to obtain a feature risk value; determining a lane change risk value based on the map matching confidence score, the intent confidence score, and the feature risk value; and controlling the vehicle to travel along the candidate lane change area if the candidate lane change area is determined to be a low-risk area based on the lane change risk value.

[0005] In some embodiments of this application, the step of using a lane topology verification strategy based on a navigation map to verify the candidate lane change area and obtain map matching confidence includes: querying map road information within a preset range based on the vehicle's position in the navigation map; projecting the candidate information of the candidate lane change area onto the navigation map, detecting the topological connectivity, geometric feature matching degree, and attribute labels corresponding to the candidate information of the candidate lane change area; and determining the map matching confidence based on the topological connectivity, the geometric feature matching degree, and the attribute labels.

[0006] In some embodiments of this application, the step of verifying the candidate lane change area using an intent prediction verification strategy based on historical trajectories to obtain intent confidence includes: sending a data acquisition request to a server, the data acquisition request being used to instruct the server to provide the historical trajectory corresponding to the candidate lane change area within a preset time period; receiving the historical trajectory provided by the server, the historical trajectory including multiple trajectory segments; classifying the multiple trajectory segments to determine a first proportion corresponding to a first type of trajectory and a second proportion corresponding to a second type of trajectory; calculating the number of trajectories for the multiple trajectory segments; and determining the intent confidence based on the number of trajectories, the first proportion, and the second proportion.

[0007] In some embodiments of this application, the step of using a perception-based feature recognition verification strategy to verify the features in the candidate lane change area and obtain a feature risk value includes: performing potential feature recognition on the candidate lane change area and generating the feature risk value, wherein the potential features include one or more features selected from signs, markings, obstacles, road conditions, and boundaries.

[0008] In some embodiments of this application, determining the lane change risk value based on the map matching confidence, the intent confidence, and the feature risk value includes: normalizing the map matching confidence to obtain the map risk value; The intent confidence is normalized to obtain the intent risk value; the weights of the weighted fusion model are adjusted based on the type of lane the vehicle is in and / or the driving environment; the lane change risk value is determined using the weighted fusion model after weight adjustment based on the map risk value, the intent risk value, and the feature risk value.

[0009] In some embodiments of this application, determining the lane change risk value based on the map matching confidence, the intent confidence, and the feature risk value includes: normalizing the map matching confidence to obtain a map risk value; normalizing the intent confidence to obtain an intent risk value; when it is determined that the mapping relationship between the map risk value, the intent risk value, and the feature risk value does not satisfy a linear relationship, invoking a fuzzy logic fusion model; and determining the lane change risk value using the fuzzy logic fusion model based on the map risk value, the intent risk value, and the feature risk value.

[0010] In some embodiments of this application, the method further includes: if the lane change risk value is less than or equal to a preset risk value, confirming the candidate lane change area as the low-risk area; if the lane change risk value is greater than the preset risk value, determining the candidate lane change area as a high-risk area.

[0011] In some embodiments of this application, the method further includes: calculating a lateral offset based on the lane change risk value when the candidate lane change area is determined to be a high-risk area; and controlling the vehicle to travel on the currently traveling road and issuing a warning message based on the lateral offset, the warning message being used to prompt the driver of the vehicle to take over the vehicle.

[0012] This application also provides an electronic device, which includes a processor and a memory, wherein the processor is used to execute a computer program stored in the memory to implement the above-described lane change control method.

[0013] This application also provides a vehicle that includes electronic devices.

[0014] In the lane change control method provided in this application, when the vehicle's autonomous driving system perceives a lane-change ramp ahead of the vehicle, it issues a navigation command prompting the vehicle to change lanes at the ramp. To avoid the vehicle mistakenly entering a temporary area due to perceptual errors, it does not directly follow the navigation command but first enters a verification process to assess the perceived risk of the ramp. Specifically, the vehicle responds to the navigation command by obtaining candidate lane change areas corresponding to its driving environment. The process of verifying candidate lane change areas includes: verifying the candidate lane change areas using a lane topology verification strategy based on the navigation map to obtain map matching confidence; verifying the candidate lane change areas using an intent prediction verification strategy based on historical trajectories to obtain intent confidence; verifying the features in the candidate lane change areas using a perception-based feature recognition verification strategy to obtain feature risk values; and determining a lane change risk value based on the map matching confidence, intent confidence, and feature risk value. Based on the lane change risk value, if the candidate lane change area is determined to be a low-risk area, the vehicle is controlled to drive along the candidate lane change area. This helps avoid mistaking temporary areas for ramps, ensuring vehicles can enter the correct ramps and improving driving safety. Attached Figure Description

[0015] Figure 1 This is an application scenario diagram of the lane change control method provided in the embodiments of this application.

[0016] Figure 2 This is a flowchart of the lane change control method provided in the embodiments of this application.

[0017] Figure 3 This is a schematic diagram of multi-source verification of candidate lane change regions provided in an embodiment of this application.

[0018] Figure 4 This is a decision-making diagram provided in an embodiment of this application.

[0019] Figure 5This is a flowchart illustrating the determination of map matching confidence provided in an embodiment of this application.

[0020] Figure 6 This is a flowchart illustrating the determination of intent confidence provided in the embodiments of this application.

[0021] Figure 7 This is a schematic diagram of decision-making and data post-processing provided in the embodiments of this application.

[0022] Figure 8 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0023] For ease of understanding, exemplary descriptions of some concepts related to the embodiments of this application are provided for reference.

[0024] It should be noted that in this application, "at least one" means one or more, and "more than one" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and drawings of this application are used to distinguish similar objects, not to describe a specific order or sequence.

[0025] In the field of autonomous driving decision-making and planning, when a vehicle needs to enter an exit ramp from the main road, it typically relies on navigation instructions, high-precision maps, or real-time perception information to determine whether to leave the main road. However, these methods are difficult to flexibly handle newly added temporary areas on the road, which can easily lead to vehicles mistakenly entering such areas.

[0026] To address the technical problem that related technologies cannot detect temporary areas in a timely manner and thus prevent vehicles from mistakenly entering such areas, thereby improving vehicle driving safety, this application proposes a lane change control method, electronic device, and vehicle. When prompted to change lanes at a ramp, the method first verifies the candidate lane change area in front of the vehicle to prevent the vehicle from mistakenly entering the temporary area and improve vehicle driving safety.

[0027] The lane change control method of this application is applied to one or more electronic devices in a vehicle. An electronic device is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions. Its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0028] Electronic devices can be any electronic product capable of human-computer interaction with a customer. They can be in-vehicle devices built into a vehicle or external devices connected to a vehicle; this application does not limit this. Examples include personal computers, tablets, smartphones, personal digital assistants (PDAs), game consoles, interactive network television (IPTV), and smart wearable devices.

[0029] Electronic devices may also include network devices and / or client devices. The network devices include, but are not limited to, a single network server, a server group consisting of multiple network servers, or a cloud based on cloud computing consisting of a large number of hosts or network servers.

[0030] The networks in which electronic devices are located include, but are not limited to, the Internet, wide area networks, metropolitan area networks, local area networks, and virtual private networks (VPNs).

[0031] The lane change control method proposed in this application is applied to the electronic equipment of a vehicle. The application scenarios of the lane change control method are described below.

[0032] Figure 1 This is an application scenario diagram of the lane change control method provided in the embodiments of this application. For example... Figure 1 As shown, the lane change control method provided in this application can be applied to electronic device 100. Electronic device 100 can be installed in vehicle 200, and it is communicatively connected to other in-vehicle devices (not shown) in vehicle 200. For example, other in-vehicle devices may include in-vehicle camera devices, in-vehicle displays, in-vehicle audio systems, steering devices of vehicle 200, etc., and this application does not limit this. In other embodiments, electronic device 100 can also be an in-vehicle device of vehicle 200, and the lane change control method is directly applied to the in-vehicle devices in the vehicle; in practical applications, this is not limited.

[0033] like Figure 1 As shown, in the autonomous driving decision-making and planning, vehicle 200 is on the main road (e.g., Figure 1 When driving on the road indicated by A1 as shown, the system recognizes... Figure 1 The area indicated by A2 is identified as a ramp (it may be an error that a temporary area is being identified as a ramp), and a navigation command prompting a lane change ramp is issued.

[0034] To avoid misidentifying temporary areas as ramps, electronic equipment 100 can be used to sense candidate lane-changing areas (such as those in the driving environment of vehicle 200) based on the driving environment. Figure 1 The area indicated by A2 is checked to improve vehicle driving safety and prevent vehicles from mistakenly entering the temporary area.

[0035] For example, if, after verification, it is determined that... Figure 1 The area indicated by A2 is a low-risk area, meaning that... Figure 1 The area indicated by A2 is the ramp. Figure 1 As shown, the ramp indicated by A2 is the right-hand branch of the main road indicated by A1, which is an outward-curving arc-shaped lane. It is an exit ramp (exit ramp) for vehicles on the main road to change lanes and exit the highway / main road.

[0036] For example, if, after verification, it is determined that... Figure 1 The area indicated by A2 is a high-risk area, meaning that... Figure 1 The area indicated by A2 is a temporary area, which may be a temporary parking area, the end of an emergency lane, a construction area, a temporary road jurisdiction area, or a temporary traffic organization change area, etc.

[0037] Figure 2 This is a flowchart of the lane change control method provided in the embodiments of this application, which is applied to vehicle electronic equipment (such as...). Figure 1 In the electronic device 100 shown. Depending on different requirements, the order of the steps in this flowchart can be changed, and some steps can be omitted.

[0038] Step S201: In response to the navigation command prompting a lane change at the ramp, obtain the candidate lane change area corresponding to the vehicle's driving environment.

[0039] In some embodiments of this application, in the autonomous driving system, navigation is responsible for road-level planning, planning the route and ramp exits that the vehicle should travel based on the navigation map. The autonomous driving system is responsible for behavior and local planning, outputting specific control commands such as lane keeping, lane changing, and acceleration / deceleration based on high-precision maps and real-time sensor detection results. However, when the vehicle approaches a temporary construction zone or accident scene, if the navigation still requires the vehicle to continue into a certain lane based on outdated map data, or if the sensors misidentify a temporary area with features similar to a ramp (such as paved or with boundary lines) as a lane-changing ramp, the vehicle may incorrectly determine the temporary area as a ramp that allows lane changing, ultimately leading to mistakenly entering the temporary area.

[0040] Therefore, in order to avoid vehicles accidentally entering temporary areas, when a vehicle detects a navigation instruction prompting it to change lanes on a ramp, such as "change lanes to the right / exit the main road", it can obtain the drivable area determined by the vehicle as a candidate lane change area, so as to verify whether the candidate lane change area is a ramp that can be entered in the future.

[0041] In addition, to assist in verifying candidate lane change areas, the perception results output by the vehicle's perception model can be obtained, such as lane lines, traffic signs, and prohibited obstacles. The vehicle's status, such as its position and speed, can also be obtained. Furthermore, the current navigation map (such as high-precision map data) and the historical trajectory corresponding to the candidate lane change area can also be obtained. This application does not limit the data used to assist in verifying candidate lane change areas.

[0042] Step S202: The candidate lane change area is verified using a lane topology verification strategy based on the navigation map to obtain the map matching confidence.

[0043] In some embodiments of this application, a lane topology verification strategy based on a navigation map is used to determine the probability that a candidate lane-changing area belongs to a preset road in the navigation map. Based on the vehicle's position in the navigation map, map road information within a preset range is queried. For example, the preset range can be a distance in front of the vehicle; for instance, the preset range can be the area from the vehicle's position to 500 meters ahead.

[0044] A lane topology verification strategy based on the navigation map is used to project candidate lane change areas onto the navigation map. The candidate lane change areas and their corresponding projected areas on the navigation map are compared to determine the correlation. For example, this correlation is determined by comparing the topological connectivity, geometric feature matching degree, and attribute labels between the candidate lane change areas and the projected areas. The map matching confidence level is then determined based on this correlation. Alternatively, the map matching confidence level can also be determined as follows: Figure 5 The illustrated embodiment.

[0045] Step S203: Verify the candidate lane change area using an intent prediction verification strategy based on historical trajectories to obtain intent confidence.

[0046] In some embodiments of this application, the historical trajectory can be the driving trajectory of other vehicles or the vehicle itself within a preset time period corresponding to the candidate lane-changing area. The preset time period includes multiple historical time periods prior to the current time period. For example, if the current time period is from 9:00 AM to 9:10 AM on June 1st, then the preset time period can be from 9:00 AM to 9:10 AM within the week prior to June 1st. It is understood that obtaining the historical trajectory is to verify whether any vehicles have entered the candidate lane-changing area in the past period. Therefore, if the vehicle detects the navigation instruction prompting a lane change at 9:00 AM on June 2nd, the driving trajectory of other vehicles and / or the vehicle itself in the candidate lane-changing area from 9:00 AM to 9:10 AM on June 1st can be obtained as the historical trajectory.

[0047] The historical trajectory can include multiple trajectory segments, each corresponding to the driving trajectory of a historical vehicle within a preset time period. Each trajectory segment can correspond to a different type; for example, the trajectory segment type can include a traversing trajectory, a terminating trajectory, and a rare trajectory.

[0048] Historical trajectories can be obtained from servers (such as cloud servers or edge servers), or vehicles can receive historical trajectories actively sent by the server during initial route planning. This application does not restrict the method of obtaining historical trajectories. Using an intent prediction and verification strategy based on historical trajectories can determine the behavioral data of historical vehicle groups, and can infer the actual situation of candidate lane-changing areas within a preset time period, thereby obtaining intent confidence. For example, whether the candidate lane-changing area is passable within a preset time period. This can, to some extent, avoid the problem of untimely navigation map updates. Furthermore, other methods for determining intent confidence based on historical trajectories include the following... Figure 6 The illustrated embodiment.

[0049] Step S204: Use a perception-based feature recognition and verification strategy to verify the features in the candidate lane change region and obtain the feature risk value.

[0050] In some embodiments of this application, while the vehicle is in motion, the vehicle's perception module (such as a combination of various sensors) monitors the surrounding environment of the vehicle in real time, thereby providing a number of visual features.

[0051] After identifying candidate lane-change areas, a more powerful perception model and rule detector are invoked to identify potential features within and along the boundaries of these areas. A pre-defined list of features can be used for identification, and these potential features can include one or more of the following: signs, markings, obstacles, road conditions, and boundaries.

[0052] The identification of signs can include traffic signs such as "Temporary Parking", "Construction Detour", and "No Entry". The identification of road markings can include dashed lines, solid lines, grid lines, and colors (such as yellow grid lines indicating no parking).

[0053] Obstacle identification can include the identification of stationary vehicles or other objects that can prevent vehicles from moving normally. For example, if the candidate lane change area is a standard ramp, then there are no other vehicles in its lanes or dynamically moving vehicles flowing in the same direction. However, if multiple vehicles parked diagonally and facing different directions are detected within the candidate lane change area, then the potential features corresponding to the obstacle are identified.

[0054] Road condition identification can include the identification of road material or road paving marks. For example, if the road material of the candidate lane change area is detected to be different from that of the main road on which the vehicle is currently traveling, or if obvious temporary paving marks are detected on the road surface of the candidate lane change area, then the potential features corresponding to the road condition are identified.

[0055] Boundary identification can include identifying the entrances to candidate lane-changing areas. For example, if the entrance to a candidate lane-changing area is found to contain movable barriers, cones, temporary guardrails, etc., then the potential features corresponding to the identified boundary are determined.

[0056] The above are merely examples; more potential features than those described can be identified within the candidate lane change region, and this application does not limit this.

[0057] In some embodiments of this application, for each potential feature identified, its corresponding risk weight is recorded, and all risk weights are accumulated to obtain a cumulative risk weight. Based on a preset risk upper limit and the cumulative risk weight, a feature risk value is determined. A higher feature risk value indicates more sufficient visual features for the candidate lane change area to be a temporary area.

[0058] In one example, the initial risk weight is 0, and the preset risk cap is 1. If a "temporary parking" sign is detected, the risk weight increases by 0.7; if a cone / barrier is detected, the risk weight increases by 0.6; if vehicles are parked haphazardly (i.e., potential features corresponding to obstacles are identified), the risk weight increases by 0.4; if the road surface material of the main road and the candidate lane-changing area is different, the risk weight increases by 0.3. Based on the above example, the final cumulative risk weight is 2. Since the set risk cap is 1, the feature risk value corresponding to this example is determined to be 1.

[0059] In another example, if only a "Temporary Parking" sign is identified, the feature risk value is 0.7.

[0060] The risk weights of the above potential features are merely examples and can be designed according to actual applications; this application does not impose any restrictions on them.

[0061] The order of steps S202, S203, and S204 can be changed, and steps S202, S203, and S204 can be executed in parallel. This application does not restrict the execution order of steps S202, S203, and S204.

[0062] To better understand steps S202, S203, and S204 above, the following will be combined with... Figure 3 Describe it.

[0063] like Figure 3 As shown, multi-source verification of candidate lane change regions can be performed as follows: Figure 3 The parallel verification process is illustrated below. A lane topology verification strategy based on a navigation map is used to verify candidate lane-changing areas. Through navigation map querying and association, and topological and geometric matching analysis, a map confidence score is generated. An intent prediction verification strategy based on historical trajectories is used to verify candidate lane-changing areas. Historical trajectories, including multiple trajectory segments, are retrieved, and these segments are classified. The probability of passage for each trajectory category is calculated and denoted as the intent confidence score. A feature recognition verification strategy based on perception is used to verify features within candidate lane-changing areas. Through area-focused perception, latent feature detection is performed based on a feature recognition list, and weighted risk accumulation is performed to generate feature risk values.

[0064] Step S205: Determine the lane change risk value based on map matching confidence, intent confidence, and feature risk value.

[0065] In some embodiments of this application, map matching confidence, intent confidence, and feature risk values ​​are unified to the same dimension (risk dimension) and scale (0-1) in order to perform risk fusion.

[0066] Normalizing the map confidence score yields the map risk value, expressed by the formula: R_map = 1 - C_map, where R_map represents the map risk value and C_map represents the map confidence score. The map confidence score indicates the confidence that the navigation map supports candidate lane change areas as ramps; a higher map risk score indicates a higher probability that the lane is not a compliant lane provided by the navigation map, and thus a higher risk; conversely, a lower map risk score indicates a higher probability that the lane is a compliant lane provided by the navigation map, and thus a lower risk.

[0067] Normalizing the intent confidence score yields the intent risk value, expressed by the formula: R_dyn = 1 - C_dyn, where R_dyn represents the intent risk value and C_dyn represents the intent confidence score. The intent confidence score indicates the probability that historical vehicle behavior (such as historical trajectories) supports the candidate lane change area as a passable lane. A higher intent risk score indicates fewer historical vehicles will treat the candidate lane change area as a passable road, resulting in higher risk; conversely, a lower intent risk score indicates more historical vehicles will treat the candidate lane change area as a passable road, resulting in lower risk.

[0068] Among them, the map risk value R_map, the intent risk value R_dyn, and the feature risk value R_temp are all in the range [0, 1]. The larger the risk value, the higher the risk, and the smaller the risk value, the lower the risk.

[0069] In some embodiments of this application, different weights can be defined for the weighted fusion model based on the different roads the vehicle is on. The type of lane the vehicle is currently traveling in can be obtained, and this lane type can include, but is not limited to, highways and ordinary roads. Ordinary roads can include other national highways, provincial highways, and county roads, etc. For example, if the lane type is a highway, the accuracy of the navigation map is higher, and the map risk value R_map should be higher. If the lane type is an ordinary road, such as an urban expressway, due to frequent construction, the feature risk value R_temp should be higher.

[0070] Furthermore, the weights of the weighted fusion model can be dynamically adjusted based on the vehicle's driving environment. For example, if the driving environment is determined to be harsh, such as heavy rain or fog, the weight W_temp corresponding to the map risk value R_map can be reduced, while the weight W_map corresponding to the map risk value R_map can be increased. If the driving environment is determined to be an unfamiliar area, such as an area with sparse historical vehicle network data, the weight corresponding to the intent risk value R_dyn can be reduced.

[0071] In one example, after determining the map risk value R_map, the intent risk value R_dyn, and the feature risk value R_temp, and confirming that the lane type is a main road of a highway, and that the weather is good, the weights of the weighted fusion model are adjusted according to the principle of "map first, historical behavior second, and real-time features as an auxiliary factor." For example, the adjusted weights are as follows: ,in, .

[0072] A weighted fusion model is used to sum the map risk value R_map, the intent risk value R_dyn, and the feature risk value R_temp, outputting the lane change risk value. This can be expressed by the formula: ;in, This indicates the risk value for lane changes.

[0073] In other embodiments of this application, a mapping relationship is determined between map risk values, intention risk values, and feature risk values. This mapping relationship can be determined based on expert rules or data verification to assess the impact of map risk values, intention risk values, and feature risk values ​​on the final risk. If the mapping relationship does not satisfy linear superposition, it is determined that the mapping relationship does not satisfy linearity, and a fuzzy logic fusion model is invoked.

[0074] A fuzzy logic fusion model is used to define fuzzy sets (e.g., high, medium, low) for map risk values, intent risk values, and feature risk values, and a fuzzy rule base is defined to determine the output of the fuzzy logic fusion model. Based on the defined fuzzy sets and the fuzzy logic rule base, the lane change risk value is obtained using the fuzzy logic fusion model.

[0075] In one example, the fuzzy logic fusion model may include, but is not limited to: ; This indicates the risk value for lane changes.

[0076] .

[0077] .

[0078] in, ,or, If the lane change risk value is 0.1, it means keeping the lane. If the lane change risk value is 0.9, then a lane change will be performed.

[0079] In another embodiment of this application, a large amount of manually labeled driving data (scene + decision information) can be collected to train a classifier, such as a neural network or gradient boosting tree, to learn the mapping relationship between (R_map, R_dyn, R_temp, context) and lane change risk value. The trained classifier can capture complex nonlinear relationships. The trained classifier is then used to process the map risk value R_map, the intent risk value R_dyn, and the feature risk value R_temp to output the lane change risk value.

[0080] Step S206: Based on the lane change risk value, if the candidate lane change area is determined to be a low-risk area, control the vehicle to drive along the candidate lane change area.

[0081] In some embodiments of this application, a preset risk value can be set, for example, 0.6. This preset risk value can be set according to the actual situation, and this application does not limit it. If the lane change risk value is less than or equal to the preset risk value, the candidate lane change area is determined to be a low-risk area; if the lane change risk value is greater than the preset risk value, the candidate lane change area is determined to be a high-risk area.

[0082] If the candidate lane change area is determined to be a low-risk area, it means that identifying the candidate lane change area as a ramp is correct, and a sequence of path points along the candidate lane change area can be planned. Based on the planned path point sequence, a lane change trajectory is generated. The vehicle is controlled to drive according to the lane change trajectory to complete the lane change operation indicated by the navigation command and enter the ramp.

[0083] If a candidate lane-change area is determined to be a high-risk area, it indicates that the candidate lane-change area is a temporary area, and the temporary area has been mistakenly identified as a ramp. In such cases, a degraded strategy can be implemented, which may include controlling the vehicle to travel on the currently traveling road.

[0084] Specifically, taking the downgrade strategy as an example, which includes strategy A (keep passing), strategy B (driver interaction), and strategy C (continuous monitoring), strategies A, B, and C can be executed in parallel or sequentially, and this application does not restrict this.

[0085] Strategy A (Keep Moving): The vehicle continues along its current path. A lateral offset can be calculated based on the lane change risk value. This offset is used to slightly adjust the lateral position to the left (e.g., 0.2 meters) to move away from the temporary area. To prevent potential risks, this lateral offset d_offset can be introduced during fine-tuning. The calculation formula is as follows: Where k is the scaling factor, and the negative sign "-" indicates a slight adjustment to the left (away from the temporary area). The higher R_total is, the greater the lateral offset, but it does not exceed the safety boundary (such as 0.5 meters).

[0086] In one example, assuming the lane change risk value R_total = 0.926, the candidate lane change area is determined to be a high-risk area. K = 0.3 is defined, and the lateral offset is calculated according to the formula above. Based on the lateral offset, a driving trajectory offset 0.28 meters to the left relative to the current lane centerline is generated. Without leaving the current main lane, the vehicle maintains an additional safety margin of approximately 0.3 meters from the temporary area on the right.

[0087] Execution Strategy B (Driver Interaction): Prompt information is sent to the driver through the vehicle's onboard equipment, such as a voice prompt saying "The area on the right ahead is a temporary zone, not an exit. The system will maintain the current lane.", a visual prompt with a dynamic exclamation mark is displayed on the human-machine interface, and the driver is prompted to take over the vehicle, such as by controlling the steering wheel.

[0088] Strategy C (Continuous Monitoring): The system continues to assess whether a next compliant ramp exists on the navigation path; whether the lane change risk value corresponding to the currently identified high-risk area decreases due to changes in distance and angle; if the lane change risk value decreases upon approaching a candidate lane change area, the vehicle system can still attempt to prompt the user to change lanes. Furthermore, it can monitor driver intervention signals (such as turning the steering wheel, pressing the brake / accelerator). For example, if the driver does not intervene within T1 seconds (e.g., 2 seconds) after the warning, the system maintains the current "lane keeping" control. If the driver actively intervenes, the system immediately and smoothly transfers control and records it as "human intervention," which will be uploaded as high-value data.

[0089] Additionally, prompts can be issued to alert the driver to take over the vehicle. For example, a semi-transparent red shading or flashing warning box can be overlaid on the area corresponding to the candidate lane-change zone in the navigation map. The correct navigation path (straight arrow) can be displayed on the dashboard in a yellow or red color, and the candidate lane-change zone can be covered with an icon (such as a prohibition sign). A concise text prompt can be displayed on the in-vehicle device, such as, "Attention: Temporary area on the right, keep your lane." A warning tone can be broadcast via voice, such as, "Attention, there is no exit on the right ahead; the system will continue straight." Furthermore, the driver can be alerted to take over the vehicle by vibrating the steering wheel.

[0090] Combination Figure 4 As shown, given the map matching confidence, intent confidence, and feature risk value, the map matching confidence and intent confidence are quantified and normalized to generate map risk value and intent risk value. The map risk value, intent risk value, and feature risk value are then input into a fusion model to generate a lane change risk value. This fusion model can be one or more of a weighted fusion model, a fuzzy logic model, and a machine learning model.

[0091] The system determines whether the lane change risk value is greater than a preset risk value. If the risk value is less than or equal to the preset risk value, the candidate lane change area is identified as a low-risk area. Navigation instructions prompting the vehicle to change lanes at the designated ramp are executed, a trajectory for entering the candidate lane change area is generated, and the vehicle is controlled to follow this trajectory. If the risk value is greater than the preset risk value, the candidate lane change area is identified as a high-risk area, and a downgrade strategy is initiated. This downgrade strategy includes Strategy A (Keep Through): controlling the vehicle to travel along the current lane; Strategy B (Driver Interaction): issuing a prompt message to alert the driver to take over the vehicle; and Strategy C (Continuous Monitoring): searching for the next candidate lane change area. Control and interaction instructions are output based on the downgrade strategy.

[0092] To better understand, Figure 2 The illustrated embodiment will be described below through a specific example. Assume a vehicle is traveling on the main road of a highway, and the navigation system indicates that it should exit the highway to the right after 1 kilometer. The vehicle's perception module detects an opening area to the right at a distance of 300 meters, which is designated as a candidate lane-change area.

[0093] A lane topology verification strategy based on the navigation map was used to verify the candidate lane change area. A query of the navigation map revealed that the nearest ramp connection point was 950 meters away, and there were no lane elements at 300 meters, only the SHOULDER attribute. The lane topology verification strategy yielded a map matching confidence score of C_map = 0.1, indicating that the navigation map did not mark the candidate lane change area as a road.

[0094] A historical trajectory-based intent prediction verification strategy is used to verify candidate lane-changing areas. Assuming the historical trajectory includes 100 trajectory segments, 92 of these segments are terminating trajectories (vehicles enter, stop, and leave after a few minutes), and 8 segments are rare trajectories. The intent prediction verification strategy yields an intent confidence score C_dyn = 0.08, indicating that historical vehicles almost exclusively stopped at these locations rather than actually traveling through them.

[0095] A perception-based feature recognition verification strategy was used to verify the features in the candidate lane-changing area. The perception module detected multiple orange cones at the entrance edge of the candidate lane-changing area, and the two parked vehicles in the candidate lane-changing area were facing different directions. The road surface color in the candidate lane-changing area was different from that of the main road, and no standard lane arrows were detected. After the feature recognition verification strategy, the feature risk value R_temp = 0.6 (cones) + 0.4 (illegal parking) = 1.0, which reached the upper limit of risk.

[0096] After the above three verification processes, three quantitative indicators of different dimensions are obtained. Taking risk fusion using a weighted fusion model as an example, weights are assigned to the map risk value R_map, the intent risk value R_dyn, and the feature risk value R_temp according to the driving environment, road type, and / or mapping relationship. For example, the weight corresponding to the map risk value R_map is W_map=0.4, the weight corresponding to the intent risk value R_dyn is W_dyn=0.3, and the weight corresponding to the feature risk value R_temp is W_temp=0.3.

[0097] The weighted fusion model after weight adjustment is defined as follows: Substituting C_map=0.1, C_dyn=0.08, and R_temp=1.0 into the weighted fusion model after weight adjustment, we calculated R_total = 0.936.

[0098] Assuming a preset risk value R_th = 0.6, then R_total (0.936) > R_th (0.6), indicating that the candidate lane-changing area is a high-risk area, possibly a temporary one. The vehicle's path planner replans the trajectory, controlling the vehicle to travel stably along the current lane centerline, and may make slight adjustments to the left to move away from the high-risk area. To improve safety, a safe distance can be maintained from vehicles approaching from behind on the main road and stationary obstacles in the suspicious area during trajectory planning. To improve comfort, lateral acceleration and deceleration can be controlled within a reasonable range. To improve timeliness, the lane change can be completed before reaching the end of the candidate lane-changing area.

[0099] Additionally, the correct navigation route is highlighted on the vehicle's onboard equipment (such as the dashboard), and suspicious areas are marked with yellow boxes. A voice prompt is given: "Monday, the right side is a temporary parking area, keep going straight." Monitoring continues. The vehicle continues forward for a distance, for example, 650 meters. The vehicle's perception system identifies the next candidate lane-change area. At this point, C_map=0.95, C_dyn=0.90, R_temp=0.1, and the calculated lane-change risk value R_total=0.1. Since R_total=0.1 is less than the preset risk value of 0.6, the candidate lane-change area is determined to be a ramp. The vehicle is then controlled to change lanes to the right and enter the ramp.

[0100] In the above embodiments, when the vehicle's autonomous driving system detects a lane-change ramp ahead, it issues a navigation command prompting the vehicle to change lanes. To avoid the vehicle mistakenly entering a temporary area due to perceptual errors, it does not directly follow the navigation command but first enters a verification process to assess the perceived risk of the ramp. Specifically, the vehicle responds to the navigation command by obtaining candidate lane-change areas corresponding to its driving environment. The process of verifying candidate lane-change areas includes: verifying the candidate lane-change areas using a lane topology verification strategy based on the navigation map to obtain map matching confidence; verifying the candidate lane-change areas using an intent prediction verification strategy based on historical trajectories to obtain intent confidence; verifying the features in the candidate lane-change areas using a perception-based feature recognition verification strategy to obtain feature risk values; and determining the lane-change risk value based on the map matching confidence, intent confidence, and feature risk value. Based on the lane-change risk value, if the candidate lane-change area is determined to be a low-risk area, the vehicle is controlled to drive along the candidate lane-change area. This helps avoid mistaking temporary areas for ramps, ensuring vehicles can enter the correct ramps and improving driving safety.

[0101] Figure 5 This is a flowchart illustrating the determination of map matching confidence provided in an embodiment of this application. For example... Figure 5 As shown, the map matching confidence is calculated by combining the vehicle's position in the navigation map and the projection position of the candidate lane change area in the navigation map, including the following steps.

[0102] Step S501: Based on the vehicle's location on the navigation map, query the map road information within a preset range.

[0103] In some embodiments of this application, the preset range can be a distance in front of the vehicle. For example, the preset range can be the area from the vehicle's position to 500 meters ahead. Starting from the vehicle's position on the navigation map, map road information within the preset range is queried. This map road information can include all roads, the connections between roads, the connection points between the main road the vehicle is currently traveling on, and candidate lane-changing areas.

[0104] Step S502: Project the candidate information of the candidate lane change area onto the navigation map, and detect the topological connectivity, geometric feature matching degree, and attribute labels corresponding to the candidate information of the candidate lane change area.

[0105] In some embodiments of this application, the candidate information for the candidate lane change area may include the entry point, centerline, exit point, etc.

[0106] Since the entrance to a temporary area (such as a temporary parking area) may be connected to the shoulder of the main road, but does not constitute a standard "lane-to-lane" connection in the topology, the topological connectivity of candidate lane change areas can be detected. For example, it can be detected whether the entrance point of the candidate lane change area is topologically connected to the main road where the vehicle is currently traveling.

[0107] Since temporary areas (such as temporary parking areas) are often shorter, narrower, or irregularly wide, the geometric feature matching degree between candidate lane-changing areas and navigation maps can be detected. For example, the ramp closest to the candidate lane-changing area in the navigation map is obtained and denoted as the target ramp. The difference between the centerline curvature and width between the candidate lane-changing area and the target ramp is calculated to obtain the geometric feature matching degree between the candidate lane-changing area and the navigation map.

[0108] The attribute labels corresponding to the candidate information are detected. For example, the projection area of ​​the candidate lane change area onto the navigation map is obtained, and the projection area is checked to see if there are attribute labels such as road_type = RAMP (ramp), road_type = SHOULDER (shoulder), and parking_area.

[0109] Step S503: Determine the map matching confidence level based on topological connectivity, geometric feature matching degree, and attribute labels.

[0110] In some embodiments of this application, if the candidate lane change area is marked as a ramp in the navigation map and the geometric feature matching degree is greater than a preset threshold, then the map matching confidence C_map=1.0.

[0111] If no attribute labels are detected in the navigation map, or if the detected label is shoulder / parking area, then the map matching confidence C_map=0.0.

[0112] If the candidate lane change area and the ramp marked on the navigation map have different locations but similar shapes, the map matching confidence C_map will have a value between 0 and 1. For example: Map matching confidence ,in, This indicates the degree of connectivity matching between the candidate lane change area and the corresponding marked ramp. S represents the geometric matching degree between the candidate lane change area and the marked ramp, where S is the matching degree score (0-1) for each item. and For weights.

[0113] Through the above embodiments, the lane topology verification strategy based on the navigation map is used to verify the candidate lane change area, which can evaluate the degree of conformity between the candidate lane change area and the lane marked in the navigation map.

[0114] Figure 6 This is a flowchart illustrating the determination of intent confidence provided in an embodiment of this application. For example... Figure 6 As shown, the confidence level of intent is calculated by combining the historical trajectories of the candidate lane change areas within a preset time period. This includes the following steps.

[0115] Step S601: Send a data retrieval request to the server.

[0116] In some embodiments of this application, the server may be an edge server, a cloud server, etc. This application does not limit the type of server, and it is used to store relevant data of all vehicles communicating with the server. The vehicle's onboard equipment communicates with the server and can receive data from the server or actively send data to the server.

[0117] The vehicle sends a data retrieval request to the server via the vehicle-to-everything (V2X) network. This request instructs the server to provide the historical trajectory of the candidate lane-changing area within a preset time period. The preset time period can include multiple historical time periods prior to the current time period. For example, if the current time period is 9:00 AM to 9:10 AM on June 1st, then the preset time period could be 9:00 AM to 9:10 AM within the week prior to June 1st. The historical trajectory can include multiple trajectory segments, each corresponding to the driving trajectory of a historical vehicle within the preset time period.

[0118] Step S602: Receive the historical trajectory fed back by the server.

[0119] In some embodiments of this application, a historical trajectory is received from the server. This historical trajectory may include multiple trajectory segments, as well as the generation time, generation location, heading angle, etc. of each trajectory segment.

[0120] Step S603: Classify multiple trajectory segments and determine the first proportion corresponding to the first type of trajectory and the second proportion corresponding to the second type of trajectory.

[0121] In some embodiments of this application, each trajectory segment of the historical trajectory may correspond to a different type. For example, the type of trajectory segment may include a traversing trajectory, a terminating trajectory, and a rare trajectory. Multiple trajectory segments may be classified, for example, according to a first type of trajectory and a second type of trajectory. The first type of trajectory may be a traversing trajectory, and the second type of trajectory may be a terminating trajectory.

[0122] If a trajectory segment smoothly enters the candidate lane-changing area from the main road and smoothly exits, eventually merging back into the main road or leaving, then the trajectory segment is determined to be a crossing-type trajectory.

[0123] After a trajectory segment enters the candidate lane change area, the speed corresponding to the trajectory segment is queried. If the speed drops rapidly to 0 and remains there for a period of time, and then exits at a low speed or reverses out, then the trajectory segment is determined to be a terminated trajectory.

[0124] In some embodiments of this application, the number of trajectories belonging to the first type among multiple trajectory segments is calculated to obtain a first ratio, and the number of trajectories belonging to the second type among multiple trajectory segments is calculated to obtain a second ratio.

[0125] Step S604: Calculate the number of trajectories for multiple trajectory segments.

[0126] In some embodiments of this application, the number of trajectories includes the number of all trajectories within a preset time period, including the number of trajectories belonging to the first category and the number of trajectories belonging to the second category.

[0127] Step S605: Determine the confidence level of intent based on the number of trajectories, the first ratio, and the second ratio.

[0128] In some embodiments of this application, the intent confidence level represents the probability of passage in the candidate lane-changing area. If the number of trajectories is less than a preset number, it means that there are almost no historical vehicles passing through the candidate lane-changing area within a preset time period, and the intent confidence level can be recorded as the median value of 0.5, indicating uncertainty.

[0129] If the number of trajectories is greater than or equal to the preset number, the confidence level of the intent is determined according to the first ratio and the second ratio.

[0130] In one example, the first proportion is denoted as P_through, and the second proportion as P_stop. The first proportion, P_through, can be directly determined as the intent confidence C_dyn. Alternatively, the intent confidence can be calculated using a preset formula based on the first and second proportions. For example, the preset formula is: , where α > 1 is used to amplify the weight of the "parking" behavior.

[0131] By using the above embodiments and employing an intent prediction verification strategy based on historical trajectories to verify candidate lane change areas, it is possible to infer the actual use of the candidate lane change areas within a preset time period.

[0132] In other embodiments of this application, after completing the determination in the above embodiments, all the above data are encapsulated as cases to obtain a packaged scenario data packet, thereby improving the accuracy of subsequent decisions. Specifically, if the lane change risk value is greater than the data return threshold, it indicates that the above-determined scenario may be an "uncertain" or "risky" edge scenario. This data return threshold can be slightly less than the preset risk value R_th; for example, if R_th = 0.6, the data return threshold R_upload = 0.5.

[0133] The encapsulated scenario data package may include, but is not limited to, perception snapshot data, system status data, truth labels, and context information. Perception snapshot data may include camera images, LiDAR point clouds, and millimeter-wave radar data taken N seconds (e.g., 2 seconds) before and after the trigger time (when the navigation command prompting a lane change at the ramp is issued). System status data may include vehicle pose, speed, C_map, C_dyn, R_temp, R_total, and the final decision. Truth labels may include the actual path the vehicle ultimately traversed as the "actual behavior" label if the system decision is executed without driver intervention, and the result of the driver's intervention (whether it's entering or bypassing) as the "human arbitration" label if the driver takes over. Context information may include time, weather, location, and high-precision map fragments.

[0134] Before transmitting the encapsulated scene data packet to the server, the encapsulated data can be anonymized (e.g., blurred faces and license plates) and compressed to obtain the target transmission data. The vehicle can then upload the data to the server via its in-vehicle T-Box under suitable network conditions (e.g., when connected to Wi-Fi).

[0135] The server can store the target transmission data into a designated database, such as a "high-risk / edge scenario database". The server can periodically (e.g., weekly) retrain the vehicle's perception model and fusion model using data from the designated database. This perception model can be a model used to obtain verification data.

[0136] The server can transmit the trained model parameters or the trained fusion model and perception model to the vehicle via OTA (Over-The-Air) updates, completing the model upgrade for vehicle deployment. After training, the vehicle can subsequently use the perception model to better learn to distinguish the visual features (such as cones, signs, and road textures) of "ramp entrances" and "temporary parking area entrances," and the risk fusion model will also be more inclined to give higher scores to such scenarios. As a result, the vehicle's intelligent driving system gains the ability to cope with such scenarios.

[0137] Combination Figure 7 The description covers the actions taken by the vehicle to make a final decision and the subsequent data processing flow. The final decision may include a conclusion that a candidate lane-changing area is a high-risk area or a low-risk area, and instructions generated based on that conclusion.

[0138] The final decision input and execution process consists of three aspects: first, control command execution and trajectory generation; second, human-computer interaction and state management; and third, data closure and model optimization.

[0139] Specifically, the execution of control commands and trajectory generation includes: if the final decision type is determined to be a low-risk area, a smooth trajectory is generated to enter the candidate lane-change area, and the trajectory is output to the vehicle controller to complete the control action. If the final decision type is determined to be a high-risk area, a trajectory is generated to continue driving along the current lane, the lateral offset is calculated based on the lane-change risk value, the lateral offset is applied (slightly adjusting to the left to move away from the risk), and the trajectory is output to the vehicle controller to complete the control action.

[0140] The system performs human-machine interaction and state management, including: if the final decision determines a low-risk area, a green guidance message is displayed on the in-vehicle display, and a confirmation voice message is played to complete the interaction and monitoring. If the final decision determines a high-risk area, a red / yellow warning box and text are displayed on the in-vehicle display, and a warning voice message (explaining the reason and intent) is played to maintain driver control. If the system detects that the driver has taken over the vehicle, control is smoothly transferred, "manual takeover" is recorded, and the interaction and monitoring are completed. If the system detects that the driver has not taken over the vehicle, the autonomous driving system maintains control.

[0141] The process of performing data closure and model optimization includes: determining whether the lane change risk value is greater than the data feedback threshold; if the lane change risk value is less than or equal to the data feedback threshold, the process ends, and data encapsulation and transmission can be temporarily suspended. If the lane change risk value is greater than the data feedback threshold, the scene data package is encapsulated, anonymized, and compressed, and then transmitted to the cloud (server). This allows the cloud to perform cloud-based data storage and model training, updating the vehicle's stored models (such as perception and fusion models), thus completing the system's iterative evolution.

[0142] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device 100 can be a mobile phone, tablet computer, smart wearable device, augmented reality (AR) / virtual reality (VR) device, laptop computer, netbook, energy storage device, power distribution equipment, vehicle-mounted equipment, self-moving device, or other electronic device with a display device. This application embodiment does not impose any restrictions on the specific type of the electronic device 100.

[0143] like Figure 8 As shown, the electronic device 100 may include a display device 101, a communication module 102, a memory 103, a processor 104, an input / output (I / O) interface 105, and a bus 106. The processor 104 is coupled to the display device 101, the communication module 102, the memory 103, and the I / O interface 105 via the bus 106.

[0144] Display device 101 can be a touch screen, specifically a touch-sensitive liquid crystal display device. Alternatively, display device 101 can be a non-touch screen. Display device 101 is used to display navigation maps, prompts, etc.

[0145] The communication module 102 may include a wired communication module and / or a wireless communication module. The wired communication module may provide one or more wired communication solutions such as Universal Serial Bus (USB) and Controller Area Network (CAN). The wireless communication module may provide one or more wireless communication solutions such as Wireless Fidelity (Wi-Fi), Bluetooth (BT), mobile communication networks, Frequency Modulation (FM), Near Field Communication (NFC), and Infrared (IR).

[0146] The memory 103 may include one or more random access memory (RAM) and one or more non-volatile memory (NVM). The RAM can be directly read and written by the processor 104, and can be used to store executable programs (such as machine instructions) of the operating system or other running programs, as well as user and application data.

[0147] Random access memory can include static random-access memory (SRAM), dynamic random-access memory (DRAM), synchronous dynamic random-access memory (SDRAM), double data rate synchronous dynamic random-access memory (DDR SDRAM), etc.

[0148] Non-volatile memory can also store executable programs and user and application data, and can be pre-loaded into random access memory for direct reading and writing by the processor 104. Non-volatile memory can include disk storage devices and flash memory.

[0149] Memory 103 is used to store one or more computer programs. The one or more computer programs are configured to be executed by processor 104. The one or more computer programs include multiple instructions that, when executed by processor 104, can implement a lane-changing control method executed on electronic device 100.

[0150] In other embodiments, the electronic device 100 also includes an external memory interface for connecting to an external memory to expand the storage capacity of the electronic device 100.

[0151] Processor 104 may include one or more processing units, such as an application processor (AP), a modem processor, a graphics processing unit (GPU), an image signal processor (ISP), a controller, a video codec, a digital signal processor (DSP), a baseband processor, and / or a neural network processing unit (NPU). These different processing units may be independent devices or integrated into one or more processors.

[0152] The processor 104 provides computing and control capabilities. For example, the processor 104 is used to execute computer programs stored in the memory 103 to implement the lane change control method described above.

[0153] I / O interface 105 is used to provide a channel for user input or output. For example, I / O interface 105 can be used to connect various input and output devices, such as mouse, keyboard, touch device, display screen, etc., so that users can enter information or visualize information.

[0154] Bus 106 is used at least to provide a channel for communication between display device 101, communication module 102, memory 103, processor 104, and I / O interface 105 in electronic device 100.

[0155] It is understood that the structures illustrated in the embodiments of this application do not constitute a specific limitation on the electronic device 100. In other embodiments of this application, the electronic device 100 may include more or fewer components than illustrated, or combine some components, or split some components, or have different component arrangements. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.

[0156] This application also provides a computer-readable storage medium storing a computer program, which includes program instructions. When the program instructions are executed, the method implemented can refer to the methods in the above embodiments of this application.

[0157] The computer-readable storage medium can be the internal memory of the electronic device described in the above embodiments, such as the hard disk or memory of the electronic device. Alternatively, the computer-readable storage medium can be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., provided on the electronic device.

[0158] In some embodiments, a computer-readable storage medium may include a stored program area and a stored data area, wherein the stored program area may store an operating system, an application program required for at least one function, etc.; and the stored data area may store data created based on the use of the electronic device, etc.

[0159] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0160] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0161] In the embodiments provided in this application, it should be understood that the disclosed apparatus / terminal devices and methods can be implemented in other ways. For example, the apparatus / terminal device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0162] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0163] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A lane change control method, applied to vehicle electronic equipment, characterized in that, The method includes: In response to the navigation command prompting a lane change at a ramp, the candidate lane change area corresponding to the vehicle's driving environment is obtained; The candidate lane change area is verified using a lane topology verification strategy based on the navigation map to obtain the map matching confidence score. The candidate lane change region is verified using an intent prediction verification strategy based on historical trajectories to obtain intent confidence. The features in the candidate lane change region are verified using a perception-based feature recognition and verification strategy to obtain feature risk values; Based on the map matching confidence, the intent confidence, and the feature risk value, the lane change risk value is determined; Based on the lane change risk value, if the candidate lane change area is determined to be a low-risk area, the vehicle is controlled to travel along the candidate lane change area.

2. The lane change control method according to claim 1, characterized in that, The process of using a lane topology verification strategy based on a navigation map to verify the candidate lane change area and obtain map matching confidence includes: Based on the vehicle's location on the navigation map, query map road information within a preset range; The candidate information of the candidate lane change region is projected onto the navigation map, and the topological connectivity, geometric feature matching degree of the candidate lane change region and the attribute label corresponding to the candidate information are detected. The map matching confidence level is determined based on the topological connectivity, the geometric feature matching degree, and the attribute label.

3. The lane change control method according to claim 1, characterized in that, The step of using an intent prediction verification strategy based on historical trajectories to verify the candidate lane change region and obtain intent confidence includes: Send a data acquisition request to the server, the data acquisition request being used to instruct the server to provide feedback on the historical trajectory of the candidate lane change area within a preset time period; Receive the historical trajectory fed back by the server, the historical trajectory including multiple trajectory segments; The multiple trajectory segments are classified to determine a first proportion corresponding to the first type of trajectory and a second proportion corresponding to the second type of trajectory; Calculate the number of trajectories for the multiple trajectory segments; The confidence level of the intent is determined based on the number of trajectories, the first ratio, and the second ratio.

4. The lane change control method according to claim 1, characterized in that, The step of using a perception-based feature recognition and verification strategy to verify the features in the candidate lane change region and obtain feature risk values ​​includes: Potential features are identified in the candidate lane change areas to generate the feature risk values. The potential features include one or more features among signs, markings, obstacles, road conditions, and boundaries.

5. The lane change control method according to any one of claims 1 to 4, characterized in that, The process of determining the lane change risk value based on the map matching confidence, the intent confidence, and the feature risk value includes: The map matching confidence score is normalized to obtain the map risk value; The intent confidence level is normalized to obtain the intent risk value; The weights of the weighted fusion model are adjusted based on the type of lane the vehicle is in and / or the driving environment. Based on the map risk value, the intent risk value, and the feature risk value, the lane change risk value is determined using a weighted fusion model with adjusted weights.

6. The lane change control method according to any one of claims 1 to 4, characterized in that, The process of determining the lane change risk value based on the map matching confidence, the intent confidence, and the feature risk value includes: The map matching confidence score is normalized to obtain the map risk value; The intent confidence level is normalized to obtain the intent risk value; When it is determined that the mapping relationship between the map risk value, the intent risk value, and the feature risk value does not satisfy a linear relationship, the fuzzy logic fusion model is invoked. Based on the map risk value, the intent risk value, and the feature risk value, the lane change risk value is determined using the fuzzy logic fusion model.

7. The lane change control method according to claim 1, characterized in that, The method further includes: If the lane change risk value is less than or equal to the preset risk value, then the candidate lane change area is confirmed as the low-risk area. If the lane change risk value is greater than the preset risk value, the candidate lane change area is determined to be a high-risk area.

8. The lane change control method according to claim 1, characterized in that, The method further includes: Based on the lane change risk value, if the candidate lane change area is determined to be a high-risk area, the lateral offset is calculated based on the lane change risk value. Based on the lateral offset, the vehicle is controlled to travel on the current road and a prompt message is issued to prompt the driver to take over the vehicle.

9. An electronic device, characterized in that, The electronic device includes a processor and a memory, the processor being configured to execute a computer program stored in the memory to implement the lane change control method as described in any one of claims 1 to 8.

10. A vehicle, characterized in that, The vehicle includes the electronic equipment as described in claim 9.

Images