Lane type and road hypothesis determination in road models

Abstract

This document describes techniques and systems that make determinations of lane type hypotheses and road hypotheses in a road model. A road perception system can fuse various forms of evidence to determine lane type hypotheses and respective confidence qualities associated with the lane type hypotheses. The road perception system uses the confidence qualities to compute confidence parameters and plausibility parameters associated with the lane type hypotheses. The lane type hypotheses are then used to determine one or more road hypotheses. The road perception system then uses the respective confidence parameters and plausibility parameters associated with the lane type hypotheses to compute confidence parameters and plausibility parameters associated with the road hypotheses. In this way, the described techniques and systems can provide accurate and reliable road models with quantified uncertainty.

Description

Background Technology

[0001] Road perception systems provide vehicle-based systems with information about road conditions and geometry, for example, for controlling vehicles on the road. Vehicles can utilize road perception systems in various vehicle-based systems, including: Adaptive Cruise Control (ACC), Traffic Jam Assist (TJA), Lane Centering Assist (LCA), and Level 3 / Level 4 autonomous driving on highways. Some safety regulations require such vehicle-based systems to model road lanes and identify the type of each lane. Furthermore, some safety standards require road models to quantify the uncertainties associated with each lane and its type. Existing road perception systems are often inaccurate or lead to unreliable road modeling, and the uncertainties in their assumptions may not be quantifiable enough to meet such regulations. Summary of the Invention

[0002] This document describes techniques and systems for determining lane types and road assumptions in a road model. For example, this document describes a road perception system configured to define lanes that constitute part of a road. Each lane is assigned multiple lane type assumptions. The road perception system determines a corresponding belief mass associated with the multiple lane type assumptions for each lane. The road perception system then uses the corresponding belief mass to calculate a belief parameter and a plausibility parameter associated with each lane type assumption. For each lane, the multiple lane type assumptions are combined into at least one road assumption, and the belief parameter and plausibility parameter associated with each road assumption are determined. The described techniques and systems then determine whether the belief parameter and plausibility parameter associated with each lane type assumption of each of the at least one road assumptions are each greater than a corresponding threshold. In response to determining that the belief parameter and plausibility parameter of each lane type assumption of one or more of the at least one road assumptions are each greater than a corresponding threshold, an autonomous driving system or an assisted driving system uses the road assumptions to operate vehicles on the road.

[0003] This document also describes other operations of the system summarized above and other methods set forth herein, as well as the apparatus for performing these methods.

[0004] This invention presents a simplified concept for determining lane types and road assumptions in a road model, which is further described below in the detailed embodiments and accompanying drawings. This invention is not intended to identify the essential features of the claimed subject matter, nor is it intended to define the scope of the claimed subject matter. Attached Figure Description

[0005] This document describes in detail, with reference to the following figures, one or more aspects of determining lane types and road assumptions in the road model. The same numbers are used throughout the figures to refer to similar features and components:

[0006] Figure 1 An example environment is shown in which the determination of lane types and road assumptions in a road model can be realized;

[0007] Figure 2 An example configuration of the road perception system is shown;

[0008] Figure 3 An example flowchart of a road perception system configured to determine uncertainties associated with lane type assumptions and road assumptions in a road model is shown.

[0009] Figure 4 A sample flowchart of the lane type module used to generate lane type assumptions in a road model is shown;

[0010] Figure 5 An example set of lanes defined by the lane type module of the road perception system is shown;

[0011] Figure 6 An example flowchart is shown, illustrating the confidence quality assigned by the lane type module based on lane markers;

[0012] Figure 7 An example flowchart is shown, illustrating the credibility quality assigned by the lane type module based on trajectory evidence.

[0013] Figure 8 An example flowchart is shown, illustrating the credibility quality assigned by the lane type module based on road signage.

[0014] Figure 9 An example flowchart for the road type module used to generate road assumptions in a road model is shown; and

[0015] Figure 10 An example method of a road perception system configured to make determinations about lane types and road assumptions in a road model is shown. Detailed Implementation

[0016] Overview

[0017] Road perception systems are a crucial technology for driver assistance and autonomous driving systems. Some driving systems (e.g., Level 2, Level 3, or Level 4 systems) and some safety standards (e.g., Safety of Intended Functions (SOTIF) (ISO / PAS 21448:2019 "Road vehicles - Safety of Intended Functions")) require road perception systems not only to model lanes, lane types, and the road itself, but also to quantify uncertainties in the model and maintain lane type and road assumptions.

[0018] Some road perception systems define a road as a combination of one or more segments, each segment being divided into strips. These systems typically identify each strip as the area between two lane boundaries. Such systems and methods can be complex. Furthermore, these systems do not provide a clear method for updating the road topology based on sensor data acquired by vehicles.

[0019] In contrast, this document describes less complex yet accurate road perception techniques for estimating confidence and plausibility parameters associated with lane types and roads in a road model. For example, these techniques can resolve potential conflicts between different information sources, providing robust information for situational assessment and safe path planning, as well as maneuver control. Specifically, an algorithmic framework is provided for fusing information from vision, trajectories (e.g., trajectories measured using sensors), and prior knowledge to estimate lane and road types. Based on the confidence and plausibility parameters associated with lane types, the road perception system can determine the overall uncertainty associated with one or more road assumptions. In this way, the described road perception techniques and systems can provide critical information about the environment surrounding a vehicle (especially information corresponding to roads and lanes within it) to provide safe path planning and maneuver control.

[0020] For example, the described techniques can define lane types in a road model and quantify the uncertainties associated with lane types. The described road perception system merges lane segments into a set of suggested lanes. The road perception system can then use the uncertainties associated with lane types to determine confidence and plausibility parameters associated with lane type and road assumptions. In this way, the described techniques and systems can quantify uncertainties in a road model and better meet the SOTIF requirements for L3 / L4 systems. The described techniques and systems can also be scaled down to other autonomous or assisted driving systems with fewer sensors or different sensor configurations.

[0021] This section describes just one example of how the described techniques and systems can be used to determine lane types and road assumptions in a road model. Other examples and implementations are described in this document.

[0022] Operating environment

[0023] Figure 1 An example environment 100 is shown in which the determination of lane types and road assumptions in a road model can be implemented. In the depicted environment 100, a road perception system 106 is mounted on or integrated into a vehicle 102. The vehicle 102 can travel on a road 120, which includes lanes 122 (e.g., a first lane 122-1 and a second lane 122-2). In this implementation, the vehicle 102 travels in the first lane 122-1.

[0024] Although shown as a car, vehicle 102 could represent other types of motorized vehicles (e.g., motorcycles, buses, tractor-trailers, semi-trailer trucks, or construction equipment). Typically, the manufacturer can mount the road perception system 106 onto any mobile platform that can travel on the road 120.

[0025] In the depicted implementation, a portion of the road perception system 106 is mounted in the rearview mirror of the vehicle 102 to have a field of view of the road 120. The road perception system 106 can project the field of view from any external surface of the vehicle 102. For example, the vehicle manufacturer may integrate at least a portion of the road perception system 106 into a side mirror, bumper, roof, or any other internal or external location where the field of view includes the road 120. Generally, the vehicle manufacturer may design the location of the road perception system 106 to provide a specific field of view that sufficiently encompasses the road 120 on which the vehicle 102 may travel.

[0026] Vehicle 102 also includes one or more sensors 104 to provide input data to one or more processors (not shown) of road perception system 106. Figure 1 (In the middle). Sensor 104 may include a camera, radar system, Global Positioning System (GPS), Global Navigation Satellite System (GNSS), lidar system, or any combination thereof. The camera may capture still images or videos of road 120. The radar system or lidar system may use electromagnetic signals to detect objects in road 120 or features of road 120. GPS or GNSS may determine the position and / or heading of vehicle 102. Vehicle 102 may include additional sensors to provide road perception system 106 with input data relating to road 120 and its lanes 112. Road perception system 106 may also use vehicle-to-everything (V2X) or cellular communication technologies to acquire input data from external sources (e.g., in the vicinity of the vehicle, in the vicinity of infrastructure, or the Internet).

[0027] The road perception system 106 can estimate a confidence parameter and a plausibility parameter associated with one or more road assumptions for road 120 of the road model and one or more lane type assumptions for lanes 122 of the road model. The confidence parameter represents the evidence supporting the assumption (e.g., the sum of the confidence quality of subsets of the assumptions) and provides a lower bound. The confidence parameter of the lane type assumption indicates the confidence level of the road perception system 106 in identifying the lane type of a particular lane 122. The plausibility parameter represents a subtraction of the evidence not supporting the assumption (e.g., a subtraction of the sum of the confidence quality of sets that intersect with the assumption but are empty) and is an upper bound. The plausibility of the lane type assumption indicates the likelihood that the lane type is suitable for a particular lane 122. Similarly, the confidence parameter of the road assumption indicates the confidence level of the road perception system 106 in combining the lane types associated with each lane 122 of road 120. The plausibility parameter of the road assumption indicates the likelihood that a combination of lane types is suitable for road 120.

[0028] The road perception system 106 includes a lane identification module 108, a lane type module 110, and a road type module 112. The lane identification module 108 determines the topology of road 120 and lanes 122, as represented by a set of lane segments. The lane identification module 108 can identify lane segments based on various information sources, including, for example, sensors 104 (e.g., radar measurements, visual measurements from a camera system) or prior knowledge (e.g., data from a map database, previous data collected by sensors 104)). The lane identification module 108 can also calculate an uncertainty or quality value associated with each lane segment.

[0029] Lane type module 110 can merge lane segments into a set of proposed lanes in the road model. Lane type module 110 can also identify lane type assumptions for lane 122. Lane type module 110 can also use the quality value associated with lane 122 to estimate confidence parameters (e.g., confidence level) and plausibility parameters (e.g., probability) associated with one or more lane type assumptions.

[0030] The road type module 112 can combine lane types or lane type assumptions into one or more road assumptions. The road type module 112 can also use the uncertainty associated with each lane type assumption to estimate the confidence parameter and reasonableness parameter associated with each road assumption. This document refers to... Figure 2 The components and operation of the road perception system 106 are described in more detail.

[0031] Vehicle 102 also includes one or more vehicle-based systems 114 that can use data from road perception system 106 to operate vehicle 102 on road 120. Vehicle-based systems 114 may include driver assistance systems 116 and autonomous driving systems 118 (e.g., adaptive cruise control (ACC), traffic jam assist (TJA), lane centering assist (LCA), and Level 3 / Level 4 autonomous driving (L3 / L4) systems on highways). Typically, vehicle-based systems 114 use road perception data provided by road perception system 106 to perform functions. For example, driver assistance system 116 may provide adaptive cruise control and monitor the presence of objects in first lane 122-1 in which vehicle 102 is traveling (as detected by another system on vehicle 102). In this example, road perception data from road perception system 106 identifies lane 122. As another example, driver assistance system 116 may provide an alert when vehicle 102 crosses lane markings in first lane 122-1.

[0032] The autonomous driving system 118 can move the vehicle 102 to a specific location on the road 120 while avoiding collisions with objects detected by other systems on the vehicle 102 (e.g., radar systems, lidar systems). Road perception data provided by the road perception system 106 can provide information about the position of lane 122 and the uncertainty of lane 122's position, enabling the autonomous driving system 118 to perform lane changes or maneuver the vehicle 102.

[0033] Figure 2 An example configuration of a road perception system 106 is shown. The road perception system 106 may include one or more processors 202 and a computer-readable storage medium (CRM) 204.

[0034] As a non-limiting example, processor 202 may include a system-on-a-chip (SoC), an application processor (AP), a central processing unit (CPU), or a graphics processing unit (GPU). Processor 202 may be a single-core or multi-core processor implemented using a homogeneous or heterogeneous core architecture. Processor 202 may include a hardware-based processor implemented as hardware-based logic, circuitry, processing cores, etc. In some aspects, the functionality of processor 202 and other components of the road perception system 106 is provided via an integrated processing, communication, and / or control system (e.g., a system-on-a-chip) that enables various operations of the vehicle 102 embodying the system.

[0035] The CRM 204 described herein does not include propagation signals. The CRM 204 may include any suitable memory or storage device that can be used to store device data (not shown) of the road perception system 106, such as random access memory (RAM), static RAM (SRAM), dynamic RAM (DRAM), non-volatile RAM (NVRAM), read-only memory (ROM), or flash memory.

[0036] Processor 202 executes computer-executable instructions stored within CRM 204. As an example, processor 202 may execute lane recognition module 108 to define lane segments of road 120 and determine the uncertainty or quality value associated with each lane segment. Lane recognition module 108 may determine lane segments of road 120 by, for example, extracting lane segments of road 120 from a high-definition map stored in CRM 204 and / or tracking lane segments of road 120 using data from sensor 104. Lane recognition module 108 may also estimate the confidence quality associated with each lane segment.

[0037] Similarly, processor 202 may execute lane type module 110 to merge lane segments into lanes with corresponding lane types and estimate confidence and plausibility parameters associated with one or more lane type hypotheses. Processor 202 may also execute road type module 112 to merge lane type hypotheses into one or more road hypotheses and estimate confidence and plausibility parameters associated with the one or more road hypotheses. Processor 202 also generates road perception data for vehicle-based system 114.

[0038] The road type module 110 can group lane segments into one or more lane type hypotheses based on Dempster-Shafer theory. Dempster-Shafer theory provides a framework for reasoning about a set of hypotheses affected by uncertainty. It is a generalization of Bayesian probability theory that explains the lack of evidence or omissions when estimating the probability of a hypothesis being true. (See reference...) Figure 4 The generation of lane type assumptions is described in more detail.

[0039] Lane type module 110 may include lane confidence module 206 and lane rationality module 208. Lane confidence module 206 may use the quality value associated with each of the lanes 122 to estimate a confidence parameter for each lane type hypothesis. Lane rationality module 208 may use the quality value associated with each of the lanes 122 to estimate a rationality parameter for each lane type hypothesis.

[0040] The road type module 112 may include a road credibility module 210 and a road rationality module 212. The road credibility module 210 may combine the credibility quality of lane type assumptions into an estimate of a credibility parameter associated with each of the one or more road assumptions. The road rationality module 212 may use the credibility quality of the lane type assumptions to estimate a rationality parameter associated with each of the one or more road assumptions.

[0041] The communication component 214 may include a sensor interface 216 and a vehicle-based system interface 218. For example, when the various components of the road perception system 106 are integrated within the vehicle 102, the sensor interface 216 and the vehicle-based system interface 218 may transmit data via the communication bus of the vehicle 102.

[0042] The processor 202 can also receive measurement data from one or more sensors 104 via the sensor interface 216 as input to the road perception system 106. For example, the processor 202 can receive image data or video data from a camera via the sensor interface 216. Similarly, the processor 202 can send configuration data or requests to one or more sensors 104 via the sensor interface 216.

[0043] The vehicle-based system interface 218 can transmit road perception data to the vehicle-based system 114 or another component of the vehicle 102. Generally, the road perception data provided by the vehicle-based system interface 218 is in a format usable by the vehicle-based system 114. In some implementations, the vehicle-based system interface 218 may send information to the road perception system 106, which, as a non-limiting example, includes the speed or heading of the vehicle 102. The road perception system 106 can use this information to appropriately configure itself. For example, the road perception system 106 may adjust the frame rate or scan rate of one or more sensors 104 via the sensor interface 216 based on the speed of the vehicle 102 to maintain the performance of the road perception system 106 under changing driving conditions.

[0044] The operation of lane recognition module 108, lane type module 110, road type module 112 and their respective sub-components will refer to... Figures 3 to 9 It was described in more detail.

[0045] Figure 3 An example flowchart 300 is shown for a road perception system 106 configured to determine uncertainties associated with lane type and road assumptions in a road model. Flowchart 300 illustrates... Figure 1Example operation of the road perception system 106. The road perception system 106 defines road lanes (e.g., lanes 122-1 and 122-2) as basic components of the road model.

[0046] The road perception system can use various forms of evidence (e.g., sensor data from sensor 104) to identify lane type assumptions, road assumptions, and uncertainties associated with them. For example, the road perception system 106 can use prior knowledge 302, trajectory 304, and visual information 306. Prior knowledge 302 may include information from a map or database stored in CRM 204, or previous lane type information determined based on previous conditions on road 120. For example, the map may include a high-resolution map included in the CRM 204 of the road perception system 106 or the memory of the vehicle 102, a map retrieved from a map or navigation service communicating with the road perception system 106, or a map obtained from a mobile phone or other device communicatively coupled to the road perception system 106. Trajectory 304 may include radar or lidar information Z about lane 122 obtained from one or more sensors 104 at time k. T (k). Vision 306 includes camera or video information Z about lane 122 acquired from one or more sensors 104 at time k. V (k). The road perception system 106 may use various types of evidence to form lane type assumptions and road assumptions, or to estimate the uncertainties associated with lane type assumptions and road assumptions. For example, data or measurements may be acquired from sensors located outside the vehicle 102 (e.g., embedded in the road, integrated with signs or markings, or on another vehicle located near the vehicle 102).

[0047] At point 308, the road perception system 106 uses multiple sources of information and evidence to develop a road model for lane 122 of road 120. Specifically, the road perception system 106 determines one or more lane types and road hypotheses 310(R) for time k. a (k)). The road perception system 106 also determines one or more lane types and road assumptions 310 (R). a (k) related uncertainty about time k 312(L a (k)). The road perception system 106 assigns uncertainty 312 to each information source (e.g., prior knowledge 302, trajectory 304, and vision 306). As described in more detail below, the road perception system 106 can use the Dempster-Schaffer theory to resolve potential conflicts between different information sources.

[0048] At the next time step (k+1), the road perception system 106 can use the same or different forms of evidence to identify new or updated lane type assumptions, road assumptions, and uncertainties associated with the lane type assumptions and road assumptions. For example, the road perception system 106 can use prior knowledge 302 and the trajectory 314 (Z) acquired from one or more sensors 104 at time (k+1). T (k+1)), and visual 316 (Z) acquired from one or more sensors 104 at time (k+1). V (k+1)). Additionally, the road perception system 106 can use one or more lane types and road assumptions 310 (R) from a previous time k. a (k)) and uncertainty 312 (L) a (k)).

[0049] At location 318, road perception system 106 uses multiple sources of information and evidence to develop a road model for lane 122 of road 120. Specifically, road perception system 106 determines one or more lane types and road assumptions 320 (R) for time (k+1). a (k+1)). The road perception system 106 assigns an uncertainty 322 for time (k+1) to each information source (e.g., prior knowledge 302, trajectory 304, vision 306, lane type and road assumptions 310, uncertainty 312, trajectory 314, and vision 316). In this way, the road perception system 106 recursively identifies lane type assumptions, road assumptions, and corresponding uncertainties. The road perception system 106 can then fuse previous assumptions and uncertainties with new observations from trajectory 314 and vision 316.

[0050] The road perception system 106 can dynamically update lane type assumptions, road assumptions, and corresponding uncertainties at discrete time intervals. The time interval for information updates can depend on a trade-off between accuracy and computational workload. In some implementations, the vehicle-based system 114 can dynamically change the time interval based on road conditions, driving environment, vehicle speed 102, and other considerations.

[0051] Figure 4 A sample flowchart 400 is shown for the lane type module 110 used to generate lane type assumptions in a road model. Flowchart 400 shows... Figure 1 Example operation of lane type module 110. Lane type module 110 may perform fewer or additional operations to determine one or more lane type assumptions for lane 122 of road 120.

[0052] At 402, lane type hypothesis 110 determines an initial confidence quality for the lane type hypothesis. Since previous confidence quality is not initially available, lane type module 110 can determine the initial confidence quality based on evidence from sensor 104. Previous confidence quality may not be available, for example, because vehicle 102 is traveling on new road 120 or road 120 is under construction. Lane type module 110 uses this evidence to determine the confidence quality for each lane segment in the road model, which will refer to… Figures 6 to 8 To describe in more detail.

[0053] At time 404, lane type module 110 determines the confidence quality of each lane segment at subsequent time intervals. Considering that at time k, lane type module 110 has determined the confidence quality of lane segment L... i The confidence quality. Lane type module 110 can use this confidence quality to determine the confidence quality for the same lane segment at time k+1 using the following equation (this confidence quality is denoted as L). i,k+1 ):

[0054] m p (L i,k+1 )=αm(L i,k (1) Wherein, the attenuation coefficient α is a constant, and its value is between 0 and 1. The value of the attenuation coefficient can be set based on empirical research, or adjusted based on the relative confidence level of the credibility quality associated with a particular type of evidence.

[0055] At 406, if there is no evidence available for lane segmentation at time k (e.g., L...), j,k If so, the lane type module 110 can segment another lane (e.g., L) at the same time. i,k The credibility quality of ) is propagated to that lane segment (e.g., m(L)). i,k )→m p (L j,k As an example, if road 120 comprises multiple lanes 122, lane type module 110 may encounter some lanes for which there are insufficient observations to update the confidence quality. In such cases, lane type module 110 may propagate the confidence quality, for example, based on the spatial (or geometric) relationships between lanes. Considering that lane segments 1, 2, and 3 form a continuous through-lane and that lane segment 4 is adjacent to lane segment 3 but starts anew (e.g., a new highway lane), in this scenario, the confidence quality for lane segment 4 can be propagated from the other lane segments 1, 2, and 3 using the following equation:

[0056]

[0057] Where, m i(·) represents the mass value for a specific assumption, g i Let K represent the geometric center of the i-th lane segment, and K() be the kernel function describing the correlation between different lanes.

[0058] At point 408, the lane type module 110 updates the evidence as the available confidence quality for each lane segment. The confidence quality for these lane segments is determined based on the available evidence and is represented as m. e (L i ), where the subscript e denotes the quality of credibility based on evidence. The lane type module 110 then uses the Dempster-Shaffer fusion rule to update the quality for each hypothesis:

[0059]

[0060] Lane type module 110 can then use the updated quality value to determine the quality value for other lane segments without any evidence. In the example above, m p (L i The accuracy of m was improved e (L i ).

[0061] At 410, lane type module 110 uses the different types of evidence available for lane type 122 to determine a lane type hypothesis. Lane type module 110 can fuse different sources of evidence together to determine a robust lane type hypothesis using the Dempster-Shaffer fusion rule. In the following description, this document describes how lane type module 110 can fuse two sources of information. A similar procedure can be followed sequentially for additional sources of information.

[0062] Given that the road perception system 106 has two distinct sources of evidence relating to the lane type of a particular available lane, this document denotes these two sources as m1(·) and m2(·), respectively. According to the Dempster-Schafer theory, the lane type module 110 has the following information:

[0063]

[0064]

[0065] in

[0066]

[0067] In equation (5), 1–K is the normalization coefficient, and K represents the conflict between pieces of evidence.

[0068] At 412, lane type module 110 determines a confidence parameter and a reasonableness parameter for each of one or more lane type hypotheses. Lane type module 110 can use the following corresponding equations to calculate the confidence parameter bel(A) and the reasonableness parameter pl(A) for a specific lane type hypothesis A:

[0069] bel(A)=∑ B|B∈A m(B) (7)

[0070] and

[0071]

[0072] Figure 5 An example set of lanes 500 defined by the lane type module 110 of the road perception system 106 is shown. The set of lanes 500 includes lanes 502, 504, 506, 508, 510, 512, 514, 516, and 518. The set of lanes 500 constitutes a portion of the road on which the vehicle 102 is traveling. In the depicted illustration, the vehicle 102 is traveling in lane 502.

[0073] Possible lane types for lane 500 include straight lanes (e.g., lanes 502, 508, and 510), deceleration lanes (e.g., lane 518), acceleration lanes (e.g., lane 512), shoulder lanes (e.g., lanes 514 and 516), and ending lanes (e.g., lanes 504 and 506). Depending on the region or country where vehicle 102 is traveling, road perception system 106 or lane type module 110 may define additional or fewer possible states for lane type 500.

[0074] For lane 500, lane type module 110 can use the following recognition frame (FOD) to represent the status of each lane: {L t ,L a ,L d ,L s ,L e}, where L t Indicates a straight-ahead lane, L a Indicates the acceleration lane, L d Indicates a deceleration lane, L s Indicates the shoulder, and L e This indicates the end lane. The lane type module 110 can also define the end lane (e.g., lane 504) as a straight-ahead lane or a deceleration lane with a shoulder or blockage ahead (e.g., lane 506). As described above, the lane type module 110 can use map data and / or sensor measurements to define the type of each lane in the set of lanes 500.

[0075] Based on the region or country where vehicle 102 is traveling, road perception system 106 or lane type module 110 can limit additional or fewer possible states for lane type. If the road assumptions output by road perception system 106 are restricted or reduced, lane type module 110 can reduce the FOD size to {L}. t ,L d ,L s In other implementations, if the vehicle-based system 114 only needs information about whether the current driving lane is an exit lane or a straight-through lane, the lane type module 110 can further reduce the FOD size to {L}. t ,L d The size and design of the lane type FOD provided by lane type module 110 depend on the task and problem solved by vehicle-based system 114. The output of lane type module 110 can be flexibly adapted to the requirements of downstream vehicle-based system 114.

[0076] Figure 6 A sample flowchart 600 is shown, illustrating the confidence quality assigned by the lane type module 110 based on lane markings. Flowchart 600 shows... Figure 1 Lane type module 110 is based on visual evidence (e.g., visual evidence 306 (Z)). V Example operation of (k)).

[0077] At 602, lane type module 110 determines whether the boundaries of the two lane markings are available in the evidence. If neither the left nor the right lane marking is available for the current lane (e.g., only one lane marking is available), lane type module 110 assigns a confidence quality 604 accordingly. For example, if only one lane marking is available and the lane marking is a solid line, lane type module 110 assigns a high confidence quality value to L. s And assign low confidence quality values ​​to L t and L d Table 1 below provides examples of the confidence quality 604 values ​​assigned by the lane type module 110. The values ​​in Table 1 depend on the type of lane marking boundary (e.g., solid line lane marking, dashed line lane marking) and the boundary position of the lane marking (e.g., left, right). The total confidence quality 604 for a given case (e.g., solid line left lane marking) is 1.

[0078] Table 1

[0079]

[0080] At point 606, lane type module 110 compares the width between the boundaries of two lane markings with a lane width threshold range. If the lane width between the boundaries of two lane markings is greater than the lane width threshold range, then lane type module 110 assigns a confidence quality 608 accordingly. For example, lane type module 110 can assign a high confidence quality value to L. d and {L d ,L t}, and assign low confidence quality values ​​to L s If the lane width between two lane marking boundaries is less than a lane width threshold, then the lane type module 110 assigns a confidence quality 610 accordingly. For example, the lane type module 110 can assign a high confidence quality value to L. s And assign low confidence quality values ​​to L d L t and {L d ,L t}

[0081] If the lane width between the two lane marking boundaries is within the lane width threshold range, the lane type module 110 can assign a confidence quality value 612 accordingly. For example, the lane type module can assign a high confidence quality value to L. t L d 、 and {L d ,L t}, and assign low confidence quality values ​​to L s If the right lane boundary is a solid line, then the lane type module 110 can assign a high confidence quality value to L. d and {L t ,L d If the left lane boundary is a solid line, then the lane type module 110 can assign a high confidence quality value to L. t and {L t ,L d If the lane boundary marker is a solid line, then the lane type module 110 can assign a high confidence quality value to L. t .

[0082] Based on the example rules shown in flowchart 600, lane type module 110 can determine the confidence quality value corresponding to a specific observation. Table 2 below provides examples of confidence quality values ​​608, 610, and 612 assigned by lane type module 110. Lane type module 110 can implement other rules based on other flowcharts, decision trees, or lookup tables.

[0083] Table 2

[0084]

[0085] The lane type module 110 can also compare the curvature of two different lane boundaries to assign lane type L to a straight lane. t Or deceleration lane L d The quality value. Lane type module 110 can compare the curvature vector of one lane marking boundary with the curvature vector of the adjacent lane marking boundary. If the difference between the curvature vectors is greater than a threshold, this provides an indication that the lane defined by the two lane marking boundaries is a deceleration lane, not a straight-ahead lane.

[0086] Specifically, lane type module 110 can obtain discrete lane marker points for two or more lane marker boundaries. Lane type module 110 can then calculate a curvature value for each lane marker boundary based on the discrete lane marker points. Lane type module 110 can compile the curvature values ​​for each lane marker boundary into a series and compare the distance between the data series for two adjacent lane marker boundaries. If the lane defined by the two lane marker boundaries is a straight lane, the distance between the two data series should be close to zero. If the lane is a deceleration lane (e.g., an exit lane), the distance between the two data series will not be equal to or approximately equal to zero. Lane type module 110 can use a mapping function to establish a connection between the curvature difference and the confidence quality assigned to a straight lane or deceleration lane.

[0087] The lane type module 110 can also compare normal vectors from discrete lane marker points to identify curvature differences. The lane type module 110 can assume the distance between lane marker boundaries is between 0 and 10 and use the following function as the mapping function:

[0088]

[0089] The parameters of the mapping function can be calibrated and tuned based on the specific application. For example, the coefficient α can be set to a value of 0.5. The lane type module 110 can also assign L... d The mass value is defined as 1-L t In this scenario, lane type module 110 assumes only two options: L d or L t .

[0090] Figure 7 A sample flowchart 700 is shown, illustrating the credibility quality assigned by the lane type module 110 based on trajectory evidence. Flowchart 700 shows... Figure 1 Example operation of lane type module 110.

[0091] At 702, the lane type module 110 determines whether the road perception system 106 includes trajectory evidence. If there is no trajectory evidence, the lane type module 110 assigns a confidence quality of 704 accordingly.

[0092] At 706, if the road perception system 106 includes trajectory evidence for the current lane, the lane type module 110 determines whether the trajectory evidence is within the lane boundary of the current lane. If the trajectory evidence is within the lane boundary of the current lane, the lane type module 110 assigns a confidence quality 708 accordingly. For example, the lane type module 110 may assign a high confidence quality value to L. t L d 、 and {L d ,L t}, and assign low confidence quality values ​​to L s .

[0093] If the trajectory evidence is not within the lane boundary of the current lane, the lane type module 110 assigns a confidence quality 710 accordingly. For example, the lane type module 110 can assign a high confidence quality value to L. s L e Assign a medium confidence quality value to {L} s ,L d ,L t ,L e}, and assign low confidence quality values ​​to L d L t and {L t ,L d The premise is that there is no trajectory evidence in the current lane but many trajectories in other lanes of road 120. If there are almost no trajectories in any lane of road 120, the lane type module 110 can assign a high confidence quality value to {L}. s ,L d ,L t}, and assign a medium confidence quality value to L s .

[0094] Based on the example rules shown in flowchart 700, lane type module 110 can determine the quality value corresponding to a specific observation. Lane type module 110 can implement other rules based on other flowcharts, decision trees, or lookup tables. Table 3 below provides examples of the confidence quality values ​​704, 708, and 710 assigned by lane type module 110.

[0095] Table 3

[0096]

[0097] Figure 8 A sample flowchart 800 illustrates the credibility quality assigned by the lane type module 110 based on road signs. Road signs may include traffic signs and pavement markings. Flowchart 800 shows... Figure 1Example operation of lane type module 110.

[0098] At 802, the lane type module 110 determines whether a road sign has been detected by one or more of the sensors 104 (e.g., a camera system). The lane type module 110 may also detect road signs based on data included in a map or database associated with the road 120 on which the vehicle 102 is traveling.

[0099] At point 804, lane type module 110 determines whether the detected road marking is a pavement marking or a road sign. If the road marking is a pavement marking, lane type module 110 assigns a confidence quality 806 accordingly. For example, lane type module 110 may assign a high confidence quality value to L. t L d 、 and {L d ,L t}, and assign low confidence quality values ​​to L s Alternatively, the lane type module 110 can assign confidence quality 806 based on the determined marker (e.g., assigning a high confidence quality value to L based on a right-turn marker). d )

[0100] If the road sign is a road marker, then the lane type module 110 assigns a corresponding confidence quality 808. For example, if an exit sign is detected above the lane, the lane type module 110 can assign a high confidence quality value to L. d and {L t ,L d}, and assign low confidence quality values ​​to L e As another example, if a straight-ahead lane sign is detected above the lane, the lane type module 110 can assign a high confidence quality value to L. t And assign low confidence quality values ​​to L d or L s As yet another example, if a construction sign is detected, the lane type module 110 can assign a high-confidence quality value to L. e Since the detection accuracy of road markings differs from that of road signs, the lane type module 110 can use different discount coefficients when assigning confidence quality values.

[0101] Based on the example rules shown in flowchart 800, lane type module 110 can determine the confidence quality value corresponding to a specific observation. Lane type module 110 can implement other rules based on other flowcharts, decision trees, or lookup tables. Table 4 below provides examples of confidence quality values ​​806 and 808 assigned by lane type module 110.

[0102] Table 4

[0103]

[0104] Figure 9 A sample flowchart 900 is shown for the road type module 112 used to generate road assumptions in a road model. Flowchart 900 shows... Figure 1 Example operation of Road Type Module 112.

[0105] At position 902, the road type module 112 obtains the lane type assumption from the lane type module 110. The road type module 112 may assume that lane segments and lanes are independent.

[0106] At 904, the road type module 112 may use lane type assumptions to determine one or more road assumptions for two or more lanes of road 120. The road type module 112 may use one of several methods to determine the road assumptions. Similarly, the road type module 112 may use one of several methods to evaluate the corresponding uncertainty of the lane type assumptions to determine the corresponding uncertainty associated with each of the road assumptions.

[0107] Considering that road 120 includes lane segment A and lane segment B, as an example, road type module 112 can use a product-based FOD formula, a union-based FOD formula, or a formula based on the focal element to generate road assumptions.

[0108] At 906, the road type module 112 uses a product-based FOD formula method. At 908, the road type module 112 defines road FOD as the product of FOD from different lanes. At 910, for each combined road hypothesis, a confidence quality value is calculated via the product of the probabilities of each lane segment.

[0109] Consider the example of lane segment A and lane segment B. The FOD for lane segments A and B is determined by {L}. d ,L s ,L t} and {L d ,L t The following are given separately. For lane segments A and B, the road type module 112 can define the road FOD as follows:

[0110] {(L d ,L d ),(L d ,L s ),(L d ,L t ),(L t ,L d ),(Lt ,L s ),(L t ,L t )} (10)

[0111] The road type module 112 then assigns confidence quality values ​​to each hypothesis corresponding to the joint FOD. In this example, the road type module 112 can identify 64 hypotheses (e.g., hypothesis ((L...). d ,L d ),(L d ,L s To calculate the confidence and plausibility parameters corresponding to each road hypothesis, the road type module 112 uses probabilities obtained via pignistic transformations of both lane segment A and lane segment B. For example, the road type module 112 can use the following probabilities: p A (L d ),p A (L s ),p A (L t ),p B (L d ),p B (L t The confidence quality value of the hypothesis corresponding to the new FOD is obtained through a product rule:

[0112] m((L d ,L t ))=p A (L d )×p B (L t (11)

[0113] The product-based FOD formula method allows the road type module 112 to directly use probabilities for decision-making, which allows it to handle various fuzzy situations. However, the number of road FODs generated by this method can be relatively large, and some of the fuzzy information for each lane may be discarded.

[0114] At 912, the road type module 112 uses a union-based FOD formula method. At 914, the road type module 112 uses the union of FODs from different lane segments to define the road FOD. At 916, the fusion confidence quality value for each road hypothesis can be evaluated using the Dempster-Schaffer theory.

[0115] Consider the example of lane segment A and lane segment B. The FOD for lane segments A and B is determined by {L}. d ,L s ,Lt} and {L d ,L t The following are given separately. For lane segments A and B, the road type module 112 can define the road FOD as follows:

[0116] {L d ,L s ,L t}∪{L d ,L t}={L d ,L s ,L t} (12)

[0118] The road type module 112 then assigns confidence quality values ​​to each road hypothesis corresponding to the road FOD. The road type module 112 first extends the confidence quality value of each lane segment to the road. The road type module 112 can determine the confidence quality value for road 120, which includes lane segments A and B, wherein the extension is based on the confidence quality value of lane segment B.

[0119]

[0120]

[0121]

[0122] Where, p A Let represent the probability obtained by using information from A through a pignistic transformation, n be the number of elements in the road FOD through the union, and the coefficient 1 / n-1 is designed to distribute the reduced uncertainty given by equations (13) and (14) evenly among the assumptions involving individual elements (L). d ,L t ).

[0123] Similarly, the road type module 112 can perform expansion based on the confidence quality value of lane segment A. The Dempster-Schaffer fusion rule can be used to calculate the new FOD credibility quality value:

[0124]

[0125] The union-based FOD formula method generates a relatively small number of road FODs. Furthermore, union-based FOD allows the road type module 112 to ignore a portion of the ambiguity associated with lane type assumptions, but retains certain portions of the ambiguity for different lane segments. However, this method can be computationally expensive, partly because the road type module 112 calculates the probabilities used for decision-making. Additionally, a small portion of the ambiguity information for each lane segment is discarded.

[0126] At 918, the road type module 112 uses a focus element-based FOD formula method. At 920, the road type module 112 identifies focus elements for different lane segments. The road type module 112 then defines the road focus elements by multiplying the focus elements of the different lane segments. At 922, for each element in the road focus elements, the road type module 112 determines the confidence quality value by multiplying the confidence quality values ​​of each lane segment.

[0127] Consider examples of lane segment A and lane segment B. The focus element for lane segments A and B is defined by {L}. d ,L s ,L t ,(L t ,L d )} and {L d ,L t ,(L t ,L d The following are given separately. For lane segments A and B, the road type module 112 can use the product rule to generate twelve road hypotheses, including lane segments A and B:

[0128] {(L d ,L d ),(L d ,L s ),(L d ,L t ),(L d ,(L t ,L d )),(L t ,L d ),(L t ,L s ),(L t ,L t ),...}

[0129] For each road hypothesis, the road type module 112 can assign a confidence quality value using a product rule. For example, the road type module 112 can assign a confidence quality value to hypothesis (L... d ,L t The credibility quality value is calculated as follows:

[0130] m((L d ,L t ))=m A (L d )m B (L t (17)

[0131] The focus-element-based FOD formula method does not discard any ambiguity information regarding lane segmentation. However, this method results in a relatively large amount of road FOD, which may become excessive in some cases. The method also requires the road type module 112 to calculate the probabilities used for decision-making.

[0132] Depending on current road conditions, processing requirements for generating lane type and road assumptions, or other considerations, the road type module 112 may generate road assumptions using specific methods. In other implementations, the road type module 112 may also use multiple methods to verify its results.

[0133] Example Method

[0134] Figure 10 An example method 1000 for a road perception system 106 used to determine lane type assumptions and road assumptions in a road model is described. Method 1000 is shown as a set of operations (or actions) performed, but is not limited to the order or combination of operations shown herein. Furthermore, any one or more of the operations may be repeated, combined, or reorganized to provide other methods. References may be made in the sections discussed below. Figures 1 to 9 The road perception system 106 and the entities detailed therein are referred to by way of example only. The technology is not limited to being performed by one or more entities.

[0135] At point 1002, lanes constituting a portion of the road are defined. Each lane has multiple lane type assumptions. For example, road perception system 106 may define lanes 122 constituting a portion of road 120. Road perception system 106 or lane type module 110 assigns potential lane type assumptions to each of lanes 122. Road perception system 106 or lane type module 110 may use data from maps, databases, prior knowledge, or at least one of sensors 104 (including one or more visual sensors, radar sensors, or lidar sensors) to define lanes 122.

[0136] At point 1004, a corresponding confidence quality is determined for each of the multiple lane type assumptions for each lane in the lane. The corresponding confidence quality indicates the confidence level associated with the data used to define the corresponding lane type assumption. For example, the road perception system 106 or the lane type module 110 may determine the confidence quality associated with each of the multiple lane type assumptions for each lane in lane 122.

[0137] At point 1006, a confidence parameter and a reasonableness parameter are determined for each of the multiple lane type assumptions for each lane in the lane. The confidence parameter and reasonableness parameter are determined using the confidence quality associated with each of the multiple lane type assumptions for each lane. The confidence parameter indicates the confidence level of the lane type. The reasonableness parameter indicates the likelihood of the lane type existing. For example, road perception system 106 or lane type module 110 can use the confidence quality associated with the lane type assumptions for each lane in lane 122 to determine the confidence parameter and reasonableness parameter associated with each lane type assumption in lane 122.

[0138] At point 1008, multiple lane type assumptions for each lane in the lanes are merged into at least one road assumption for the road. For example, road type module 112 can merge lane type assumptions into one or more road assumptions.

[0139] At point 1010, a confidence parameter and a reasonableness parameter associated with each of at least one road hypothesis are determined. The confidence parameters and reasonableness parameters are determined using the corresponding confidence parameters and reasonableness parameters associated with multiple lane type hypotheses for each lane in the lane. For example, road perception system 106 may use the confidence parameters and reasonableness parameters associated with lane type hypotheses to determine the confidence parameters and reasonableness parameters associated with each road hypothesis.

[0140] At point 1012, it is determined whether the credibility parameter and the reasonableness parameter associated with at least one road hypothesis are each greater than or less than a corresponding threshold. For example, road perception system 106 can determine whether the credibility parameter and the reasonableness parameter associated with the road hypothesis are greater than or less than the corresponding threshold.

[0141] At point 1014, in response to determining that the credibility parameter and reasonableness parameter of at least one road hypothesis are greater than corresponding thresholds, the autonomous driving system or the driver assistance system can use at least one road hypothesis as input to operate the vehicle on the road. For example, in response to determining that the credibility parameter and reasonableness parameter are greater than corresponding thresholds, the vehicle 102 can be operated by the autonomous driving system 118 or the driver assistance system 116. The autonomous driving system 118 or the driver assistance system 116 can use at least one road hypothesis as input to operate the vehicle 102 on road 120. Alternatively, in response to determining that at least one of the credibility parameter or reasonableness parameter is less than the corresponding threshold, the road perception system 106 can interrupt the operation of the vehicle using the autonomous driving system 118 or the driver assistance system 116 and switch the operation of the vehicle 102 to driver control.

[0142] Example

[0143] Examples are provided in the following sections.

[0144] Example 1: A method comprising: defining lanes by a road perception system of a vehicle, the lanes constituting part of a road, each of the lanes having a plurality of lane type hypotheses; determining, by the road perception system, a corresponding confidence quality associated with the plurality of lane type hypotheses for each of the lanes, the corresponding confidence quality indicating a confidence level associated with data used to define the corresponding lane type hypothesis; and determining, by the road perception system and using the corresponding confidence quality associated with the plurality of lane type hypotheses for each of the lanes, a corresponding confidence parameter and a reasonableness parameter associated with each lane type hypothesis, the confidence parameter indicating a confidence level of the lane type for each of the lanes, the reasonableness parameter indicating the applicability of the lane type to each of the lanes. The possibility; the road perception system merges the plurality of lane type assumptions for each lane in the lanes into at least one road assumption for the road; the road perception system and using the corresponding confidence parameter and reasonableness parameter associated with the plurality of lane type assumptions for each lane in the lanes determine the confidence parameter and reasonableness parameter associated with each of the at least one road assumption; determines whether the confidence parameter and reasonableness parameter associated with the at least one road assumption are each greater than or less than a corresponding threshold; and in response to determining that the confidence parameter and reasonableness parameter of the at least one road assumption are each greater than the corresponding threshold, uses the at least one road assumption as input to an autonomous driving system or assisted driving system operating the vehicle on the road.

[0145] Example 2: The method of Example 1, wherein the lane is defined using data from at least one of the following: one or more visual sensors, a map, a database, one or more radar sensors, or one or more lidar sensors.

[0146] Example 3: The method of Example 2, wherein each road assumption and each lane type assumption is dynamically updated based on the data.

[0147] Example 4: A method of any of the preceding examples, wherein merging the plurality of lane type assumptions for each of the lanes into the at least one road assumption for the road comprises: determining a proposed set of lanes that satisfies the properties of the Dempster-Shaffer theory and the Dempster-Shaffer fusion rule.

[0148] Example 5: A method of any of the preceding examples, wherein determining the corresponding confidence quality associated with the plurality of lane type assumptions for each of the lanes comprises: determining whether two lane marking boundaries are available for each of the lanes; in response to determining that the two lane marking boundaries are unavailable for one of the lanes, determining the corresponding confidence quality associated with the plurality of lane type assumptions for the one of the lanes based on the type of the lane marking boundaries and the boundary position of the lane marking boundaries; and in response to determining that the two lane marking boundaries are available for the other lane of the lane, comparing the lane width of the other lane of the lane with a lane width threshold range; in response to determining that the lane width of the other lane is greater than or less than the lane width threshold range, assigning the corresponding confidence quality associated with the plurality of lane type assumptions for the other lane of the lane; and in response to determining that the lane width is within the lane width threshold range, determining the corresponding confidence quality associated with the plurality of lane type assumptions for the other lane of the lane based on the type of the lane marking boundaries for the other lane of the lane and the boundary position of the lane marking boundaries.

[0149] Example 6: A method of any of the preceding examples, wherein determining the corresponding confidence quality associated with the plurality of lane type assumptions for each of the lanes includes: comparing the curvature of two lane marking boundaries for each of the lanes.

[0150] Example 7: The method of Example 6, wherein comparing the curvature of the two lane marking boundaries for each of the lanes includes: for each of the lanes, obtaining discrete lane marking points of the two lane marking boundaries; using the discrete lane marking points, calculating a curvature value for each of the lane marking boundaries; compiling the curvature value of each of the lane marking boundaries into a series of curvature values; and comparing the distance between adjacent lane marking boundaries in the series of curvature values.

[0151] Example 8: A method of any of the preceding examples, wherein determining the corresponding confidence quality associated with the plurality of lane type assumptions for each of the lanes comprises: determining whether trajectory evidence is available for the lanes of the road; in response to determining that the trajectory evidence is not available for the lanes of the road, determining the corresponding confidence quality associated with the plurality of lane type assumptions for each of the lanes; and in response to determining that the trajectory evidence is available for the lanes of the road, determining whether the trajectory evidence is available for the current lane in which the vehicle is traveling; in response to determining that the trajectory evidence is available for the current lane, determining the corresponding confidence quality associated with the plurality of lane category assumptions for the current lane; and in response to determining that the trajectory evidence is not available for the current lane, determining the corresponding confidence quality associated with the plurality of lane category assumptions for each of the lanes.

[0152] Example 9: A method of any of the preceding examples, wherein determining the corresponding confidence quality associated with the plurality of lane type assumptions for each of the lanes comprises: determining whether the road sign is detected; in response to determining that the road sign is detected, determining whether the road sign includes pavement markings or road signs; in response to determining that the road sign includes pavement markings, determining the corresponding confidence quality associated with the plurality of lane type assumptions for each of the lanes; and in response to determining that the road sign includes road signs, determining the corresponding confidence quality associated with the plurality of lane type assumptions for each of the lanes.

[0153] Example 10: A method of any of the preceding examples, wherein: combining the plurality of lane type assumptions for each lane in the lanes into the at least one road assumption for the road comprises: determining a product of each identification frame for the plurality of lane type assumptions; and determining the confidence parameter and the reasonableness parameter associated with each of the at least one road assumption comprises: determining a product of the corresponding probabilities for each lane type for each of the at least one road assumptions.

[0154] Example 11: A method of any of the preceding examples, wherein: merging the plurality of lane type assumptions for each lane in the lanes into the at least one road assumption for the road comprises: determining the union of each identification frame for the plurality of lane type assumptions; and determining the confidence parameter and the reasonableness parameter associated with each of the at least one road assumption comprises: using Dempster-Schafer theory to determine a fused confidence quality value for each of the at least one road assumptions.

[0155] Example 12: A method of any of the preceding examples, wherein: combining the plurality of lane type assumptions for each of the lanes into the at least one road assumption for the road comprises: determining the product of the focal elements of each of the plurality of lane type assumptions for each lane; and determining the credibility parameter and the reasonableness parameter associated with each of the at least one road assumption comprises: determining the product of the corresponding credibility quality values ​​for each of the at least one road assumption.

[0156] Example 13: A method of any of the foregoing examples, the method further comprising: in response to determining that at least one of the credibility parameter or the reasonableness parameter of the at least one road hypothesis is less than the corresponding threshold, causing an interruption of operation of the vehicle using the autonomous driving system or the driver assistance system.

[0157] Example 14: A method of any of the foregoing examples, the method further comprising: in response to determining that at least one of the credibility parameter or the reasonableness parameter of the at least one road hypothesis is less than the corresponding threshold, causing the operation of the vehicle to be switched to driver control.

[0158] Example 15: A method of any of the preceding examples, wherein the autonomous driving system or the assisted driving system includes at least one of the following: an automatic cruise control system, a traffic jam assist system, a lane centering assist system, or an L3 / L4 autonomous driving system on a highway.

[0159] Example 16: A computer-readable storage medium including computer-executable instructions that, when executed, cause a processor in a vehicle to perform any of the methods described in the preceding examples.

[0160] Example 17: A system comprising a processor configured to perform a method of any one of Examples 1 to 15.

[0161] Example 18: A computer-readable storage medium comprising computer-executable instructions, which, when executed, cause a processor in a vehicle to: define lanes, the lanes constituting part of a road, each of the lanes having a plurality of lane type hypotheses; determine a corresponding confidence quality associated with the plurality of lane type hypotheses for each of the lanes, the corresponding confidence quality indicating a confidence level associated with data used to define the corresponding lane type hypothesis; and use the corresponding confidence quality associated with the plurality of lane type hypotheses for each of the lanes to determine a corresponding confidence parameter and a reasonableness parameter associated with each lane type hypothesis, the confidence parameter indicating a confidence level of the lane type for each of the lanes, the reasonableness parameter indicating the lane class The possibility of a type being applicable to each of the lanes; merging the plurality of lane type assumptions for each of the lanes into at least one road assumption for the road; using the corresponding confidence and reasonableness parameters associated with the plurality of lane type assumptions for each of the lanes to determine the confidence and reasonableness parameters associated with each of the at least one road assumption; determining whether the confidence and reasonableness parameters associated with the at least one road assumption are each greater than or less than a corresponding threshold; and in response to determining that the confidence and reasonableness parameters of the at least one road assumption are each greater than the corresponding threshold, using the at least one road assumption as input to an autonomous driving system or assisted driving system operating the vehicle on the road.

[0162] Example 19: A computer-readable storage medium of Example 18, wherein data from at least one of the following is used to define the lane: one or more visual sensors, a map, a database, one or more radar sensors, or one or more lidar sensors.

[0163] Example 20: A computer-readable storage medium of Example 19, wherein each road hypothesis and each lane type hypothesis is dynamically updated based on the data.

[0164] Example 21: A computer-readable storage medium of any of Examples 18 to 20, wherein determining the corresponding confidence quality associated with the plurality of lane type assumptions for each of the lanes includes: comparing the curvature of two lane marking boundaries for each of the lanes.

[0165] Example 22: A computer-readable storage medium of Example 21, wherein comparing the curvature of the two lane marking boundaries for each of the lanes comprises: for each of the lanes, obtaining discrete lane marking points of the two lane marking boundaries; using the discrete lane marking points, calculating a curvature value for each of the lane marking boundaries; compiling the curvature value for each of the lane marking boundaries into a series of curvature values; and comparing the distances between adjacent lane marking boundaries in the series of curvature values.

[0166] Conclusion

[0167] While various embodiments of the present disclosure have been described in the foregoing description and illustrated in the accompanying drawings, it should be understood that the present disclosure is not limited thereto, but can be practiced in various ways within the scope of the following claims. It will be apparent from the foregoing description that various modifications can be made without departing from the spirit and scope of the present disclosure as defined by the following claims.

Claims

1. A method for a means of transport, the method comprising: Lanes are defined by a road perception system of a vehicle, the lanes forming part of a road, each of the lanes having multiple lane type assumptions, the possible lane types of the multiple lane type assumptions including: straight lane, deceleration lane, shoulder, acceleration lane or end lane; The road perception system determines a corresponding confidence quality associated with the plurality of lane type assumptions for each of the lanes, the corresponding confidence quality indicating the confidence level associated with the data used to define the corresponding lane type assumption; The road perception system determines, using the corresponding confidence quality associated with the plurality of lane type assumptions for each of the lanes, a corresponding confidence parameter and a reasonableness parameter associated with each lane type assumption, the confidence parameter indicating the confidence level of the lane type for each of the lanes, and the reasonableness parameter indicating the likelihood that the lane type is applicable to each of the lanes. The road perception system uses one of the product-based recognition framework FOD formula method, the union-based FOD formula method, or the focal element-based FOD formula method to merge the multiple lane type hypotheses for each lane in the lane into at least one road hypothesis for the road. Each road hypothesis represents a unique set of lane type hypotheses for each lane in the lane, and each road hypothesis satisfies the properties of the Dempster-Shaffer theory and the Dempster-Shaffer fusion rule. The road perception system determines the confidence parameter and reasonableness parameter associated with each of the at least one road hypothesis by using the corresponding confidence parameter and reasonableness parameter associated with the plurality of lane type hypotheses for each of the lanes. Determine whether the credibility parameter and the reasonableness parameter associated with at least one road hypothesis are each greater than or less than a corresponding threshold; and In response to determining that the credibility parameter and the reasonableness parameter of the at least one road hypothesis are each greater than the corresponding threshold, the at least one road hypothesis is used as input to an autonomous driving system or assisted driving system that operates the vehicle on the road.

2. The method as described in claim 1, characterized in that, The lane is defined using data from at least one of the following: one or more visual sensors, maps, databases, one or more radar sensors, or one or more lidar sensors.

3. The method as described in claim 2, characterized in that, Each road hypothesis and each lane type hypothesis is updated dynamically based on the data.

4. The method as described in claim 1, characterized in that, Determining the corresponding confidence quality associated with the plurality of lane type assumptions for each of the lanes includes: Determine whether the two lane marking boundaries are available for each of the lanes; In response to determining that the two lane marking boundaries are unavailable for one of the lanes, the corresponding confidence quality associated with the plurality of lane type assumptions for the one lane in the lane is determined based on the type of the lane marking boundaries and the boundary position of the lane marking boundaries; and In response to determining that the two lane marking boundaries are available for another lane in the lane, the lane width of the other lane in the lane is compared with a lane width threshold range: In response to determining that the lane width of the other lane is greater than or less than the lane width threshold range, a corresponding confidence quality is assigned to the plurality of lane type assumptions for the other lane in the lane; and In response to determining that the lane width is within the lane width threshold range, the corresponding confidence quality associated with the plurality of lane type assumptions for the other lane in the lane is determined based on the type of the lane marking boundary for the other lane in the lane and the boundary position of the lane marking boundary.

5. The method as described in claim 1, characterized in that, Determining the corresponding confidence quality associated with the plurality of lane type assumptions for each of the lanes includes: comparing the curvature of two lane marking boundaries for each of the lanes.

6. The method as described in claim 5, characterized in that, For each of the lanes, comparing the curvature of the two lane marking boundaries includes: For each of the lanes, obtain discrete lane marker points of the two lane marker boundaries; Using the discrete lane marker points, calculate the curvature value for each of the lane marker boundaries; The curvature value of each of the lane marking boundaries is compiled into a series of curvature values; and Compare the distances between adjacent lane marker boundaries in the series of curvature values.

7. The method as described in claim 1, characterized in that, Determining the corresponding confidence quality associated with the plurality of lane type assumptions for each of the lanes includes: Determine whether trajectory evidence is available for the lane of the road; In response to determining that the trajectory evidence is not available for the lanes of the road, the corresponding confidence quality associated with the plurality of lane type assumptions for each lane is determined; and In response to determining that the trajectory evidence is available for the lane of the road, determine whether the trajectory evidence is available for the current lane in which the vehicle is traveling: In response to determining that the trajectory evidence is available for the current lane, the corresponding confidence quality associated with the plurality of lane type assumptions for the current lane is determined; and In response to determining that the trajectory evidence is not available for the current lane, the corresponding confidence quality associated with the plurality of lane type assumptions for each lane in the lane is determined.

8. The method as described in claim 1, characterized in that, Determining the corresponding confidence quality associated with the plurality of lane type assumptions for each of the lanes includes: Determine whether the road sign has been detected; In response to determining that the road sign is detected, determine whether the road sign includes pavement markings or road signs; In response to determining that the road signage includes the pavement markings, a corresponding confidence quality is determined associated with the plurality of lane type assumptions for each of the plurality of lanes; and In response to determining that the road signage includes the road markings, a corresponding confidence quality is determined associated with the plurality of lane type assumptions for each of the lanes.

9. The method as described in claim 1, characterized in that: The road perception system uses a product-based FOD formula method; Combining the plurality of lane type assumptions for each lane in the lanes into the at least one road assumption for the road includes: determining the product of each identification frame for the plurality of lane type assumptions; and Determining the credibility parameter and the reasonableness parameter associated with each of the at least one road hypothesis includes: determining the product of the corresponding probabilities for each lane type for each of the at least one road hypothesis.

10. The method as described in claim 1, characterized in that: The road perception system uses a union-based FOD formula method; Merging the plurality of lane type assumptions for each lane in the lanes into the at least one road assumption for the road includes: determining the union of each identification frame for the plurality of lane type assumptions; and Determining the credibility parameter and the plausibility parameter associated with each of the at least one road hypothesis includes: using the Dempster-Schafer theory to determine a fused credibility quality value for each of the at least one road hypothesis.

11. The method as described in claim 1, characterized in that: The road perception system uses the FOD formula method based on focal elements; Combining the plurality of lane type assumptions for each lane in the lanes into the at least one road assumption for the road includes: determining the product of the focal elements of each lane for the plurality of lane type assumptions; and Determining the credibility parameter and the reasonableness parameter associated with each of the at least one road hypothesis includes: determining the product of the corresponding credibility quality values ​​for each of the at least one road hypothesis.

12. The method as described in claim 1, characterized in that, The method further includes: In response to determining that at least one of the credibility parameter or the reasonableness parameter of the at least one road hypothesis is less than the corresponding threshold, operation of the vehicle using the autonomous driving system or the driver assistance system is interrupted.

13. The method as described in claim 1, characterized in that, The method further includes: In response to determining that at least one of the credibility parameter or the reasonableness parameter of the at least one road hypothesis is less than the corresponding threshold, the operation of the vehicle is switched to driver control.

14. The method as described in claim 1, characterized in that, The autonomous driving system or the assisted driving system includes at least one of the following: automatic cruise control system, traffic jam assist system, lane centering assist system, or L3 / L4 autonomous driving system on highways.

15. A computer-readable storage medium comprising computer-executable instructions, which, when executed, cause a processor in a vehicle to: Defined lanes, which constitute part of a road, each of the lanes having multiple lane type assumptions, the possible lane types of which include: Straight lane, deceleration lane, shoulder, acceleration lane, or end lane; Determine the corresponding confidence quality associated with the plurality of lane type assumptions for each of the lanes, the corresponding confidence quality indicating the confidence level associated with the data used to define the corresponding lane type assumption; Using the corresponding confidence quality associated with the plurality of lane type assumptions for each of the lanes, a corresponding confidence parameter and a reasonableness parameter associated with each lane type assumption are determined, the confidence parameter indicating the confidence level of the lane type for each of the lanes, and the reasonableness parameter indicating the likelihood that the lane type is applicable to each of the lanes; The plurality of lane type assumptions for each lane in the lanes are merged into at least one road assumption for the road using one of the product-based FOD formula method, the union-based FOD formula method, or the focus element-based FOD formula method, wherein each road assumption represents a unique set of lane type assumptions for each lane in the lanes, and each road assumption satisfies the properties of the Dempster-Shaffer theory and the Dempster-Shaffer fusion rule. The confidence parameter and reasonableness parameter associated with each of the at least one road hypothesis are determined using the corresponding confidence parameter and reasonableness parameter associated with the plurality of lane type assumptions for each of the lanes. Determine whether the credibility parameter and the reasonableness parameter associated with at least one road hypothesis are each greater than or less than a corresponding threshold; as well as In response to determining that the credibility parameter and the reasonableness parameter of the at least one road hypothesis are each greater than the corresponding threshold, the at least one road hypothesis is used as input to an autonomous driving system or assisted driving system that operates the vehicle on the road.

16. The computer-readable storage medium as claimed in claim 15, characterized in that, The lane is defined using data from at least one of the following: one or more visual sensors, maps, databases, one or more radar sensors, or one or more lidar sensors.

17. The computer-readable storage medium as claimed in claim 16, characterized in that, Each road hypothesis and each lane type hypothesis is updated dynamically based on the data.

18. The computer-readable storage medium as claimed in claim 15, characterized in that, Determining the corresponding confidence quality associated with the plurality of lane type assumptions for each of the lanes includes: comparing the curvature of two lane marking boundaries for each of the lanes.

19. The computer-readable storage medium as claimed in claim 18, characterized in that, For each of the lanes, comparing the curvature of the two lane marking boundaries includes: For each of the lanes, obtain discrete lane marker points of the two lane marker boundaries; Using the discrete lane marker points, calculate the curvature value for each of the lane marker boundaries; The curvature value of each of the lane marking boundaries is compiled into a series of curvature values; and Compare the distances between adjacent lane marking boundaries in the series of curvature values.

20. The computer-readable storage medium as claimed in claim 15, characterized in that, This enables the processor to use a product-based FOD formula method to: The plurality of lane type assumptions for each of the lanes are combined into at least one road assumption for the road in the following manner: Determine the product of each recognition frame for the multiple lane type assumptions and Determining the credibility parameter and the reasonableness parameter associated with each of the at least one road hypothesis includes: determining the product of the corresponding probabilities for each lane type for each of the at least one road hypothesis.

21. The computer-readable storage medium as claimed in claim 15, characterized in that, This enables the processor to use the FOD formula method based on union: The plurality of lane type assumptions for each of the lanes are combined into at least one road assumption for the road in the following manner: Determine the union of each identification frame for the plurality of lane type assumptions; and Determining the credibility parameter and the plausibility parameter associated with each of the at least one road hypothesis includes: using the Dempster-Schafer theory to determine a fused credibility quality value for each of the at least one road hypothesis.

22. The computer-readable storage medium as claimed in claim 15, characterized in that, This enables the processor to use the FOD formula method based on the focus element: The plurality of lane type assumptions for each of the lanes are combined into at least one road assumption for the road in the following manner: Determine the product of the focal elements for each lane for the multiple lane type assumptions; and Determining the credibility parameter and the reasonableness parameter associated with each of the at least one road hypothesis includes: determining the product of the corresponding credibility quality values ​​for each of the at least one road hypothesis.

Images