Autonomous traveling route construction method using sensor recognition information, and autonomous traveling support device

Abstract

To provide an autonomous traveling route construction method using sensor recognition information, and an autonomous traveling support device.SOLUTION: A construction method includes the steps of: acquiring, on the basis of recognition information acquired from a moving body having an observation sensor and a positioning sensor, a movement route by tracking the moving body and at least one target object among surrounding dynamic objects around the moving body; generating a predicted route whose movement is estimated from the movement route on the basis of the movement route; generating route information comprising the movement route and the predicted route to incorporate the route information into map information; updating the route information on the basis of actual information when verification of the map information reveals a deviation of a predetermined range or more between the route information and the actual information in which the target object has actually moved at least in an area corresponding to the predicted route; and incorporating the updated route information into the map information.SELECTED DRAWING: Figure 6

Description

Technical Field

[0001] The present disclosure relates to a method for constructing an autonomous driving route using sensor recognition information and an autonomous driving support device. More specifically, when constructing and updating route information on map information for autonomous driving, recognition information collected from an autonomous driving mobility device is utilized to precisely and efficiently generate route information similar to an actual driving route. The present disclosure relates to a method for constructing an autonomous driving route and an autonomous driving support device.

Background Art

[0002] Autonomous driving mobile bodies have been developed in various mobility fields such as vehicles, robots, unmanned mobile devices, and drones, and commercialization is being explored. Through various tests, algorithms and data utilized by autonomous driving mobile bodies, such as cognitive / judgment / control SW, precise maps, and learning data, have been developed and constructed, and services are provided by installing the algorithms and the data in autonomous driving devices.

[0003] Precise maps for autonomous driving have high accuracy and a large amount of information, but have the drawback of high construction and update costs. In addition, the driving routes on precise maps may be virtual information representing the routes along which vehicles travel, rather than physically installed facilities such as lanes, signboards, and traffic signals. As a result, the virtual driving routes may differ from the actual driving routes depending on the algorithm or producer that creates the virtual driving routes.

[0004] Autonomous driving mobile bodies perform driving plans and judgment processes with reference to the driving routes provided in precise maps, and also refer to the driving route data for predicting the movement routes of surrounding objects. However, when the actual driving route differs from the virtual driving route set on the map, there is a high possibility of malfunction or accident of the autonomous driving mobile body.

Summary of the Invention

Problems to be Solved by the Invention

[0005] The technical problem of the present disclosure is to accurately and efficiently generate route information similar to the actual driving route by utilizing the recognition information collected from the autonomous driving mobility device when constructing and updating the route information on the map for autonomous driving, and to provide a method for constructing an autonomous driving route and an autonomous driving support device.

[0006] Another technical problem of the present disclosure is to provide a method for constructing an autonomous driving route and an autonomous driving support device that provide optimal driving route information by utilizing data that is time-series and has multiple viewpoints collected from a large number of autonomous driving mobilities.

[0007] The technical problem to be solved by the present disclosure is not limited to the above-described technical problems, and other technical problems not described above will be clearly understood by those having ordinary knowledge in the technical field to which the present disclosure pertains from the following description.

Means for Solving the Problem

[0008] According to one aspect of the present disclosure, a method for constructing an autonomous driving route using sensor recognition information is provided. The construction method includes: tracking at least one target object among the moving object and the surrounding dynamic objects around the moving object based on the recognition information obtained from a moving object having an observation sensor and a positioning sensor to obtain a moving route; generating a predicted route whose movement is estimated from the moving route based on the moving route; generating route information including the moving route and the predicted route, and incorporating the route information into map information; updating the route information based on the actual information when there is a deviation of a predetermined range or more between the actual information that the target object has actually moved in at least the area corresponding to the predicted route and the route information by verifying the map information; and incorporating the updated route information into the map information.

[0009] According to another embodiment of the present disclosure, the step of obtaining the movement path may include tracking the target object using the positioning information of the target object estimated from the positioning sensor and a trajectory based on the optimal position of the target object. The recognition information is collected as a plurality of recognition information so as to have a plurality of viewpoints (multiview) in a time series with respect to the target object. Features are extracted for each of the plurality of recognition information, and relative displacement information between the observation sensor and the features is generated by matching between the features. The optimal position may be generated so as to minimize the amount of interval in the relative displacement information.

[0010] According to another embodiment of the present disclosure, the extraction of the features and the matching between the features are performed with reference to the movement information and the observation state information of the moving body, and the observation state information can indicate a state in which the observation sensor of the moving body recognizes the target object.

[0011] According to another embodiment of the present disclosure, it may further include the step of providing the optimal position of the target object to the positioning information and the map information.

[0012] According to another embodiment of the present disclosure, the step of generating the predicted path may include generating the predicted path based on the displacement information of the target object, the speed information of the target object, the movement path, the environmental information of the path on which the target object travels, the regulation information applied to the path on which the target object travels, the movement pattern information of a path identical or similar to the path on which the target object travels, and the cumulative path information of a path identical or similar to the path on which the target object travels.

[0013] According to another embodiment of the present disclosure, the displacement information of the target object and the velocity information of the target object are provided in a time series, and the step of generating the predicted path generates the predicted path using trajectory modeling including a non-linear state transition method based on the time-series displacement information and the time-series velocity information, and the trajectory modeling may be configured to be fed back by an error between a predicted trajectory derived from the trajectory modeling and a trajectory derived based on the recognition information.

[0014] According to another embodiment of the present disclosure, the generation of the path information includes determining, as the path information, a single object movement information derived by matching a plurality of object movement information of the target object, and the plurality of object movement information may be transmitted from each of a plurality of moving bodies that generate a movement path and a predicted path of the target object.

[0015] According to another embodiment of the present disclosure, the matching of the plurality of object movement information includes performing geometric matching between the object movement information, clustering similar object movement information based on statistical information according to the degree of weighting of the object movement information, and generating the single object movement information based on a path model based on the clustered object movement information and adopting the single object movement information as the path information.

[0016] According to another embodiment of the present disclosure, the step of incorporating the path information into the map information may include generating connection relationship information that associates at least one of road information of a path on which the target object travels, infrastructure information provided on the traveling path, weather information of the traveling path, and regulation information applied to the traveling path with the path information.

[0017] According to another embodiment of the present disclosure, the verification of the map information includes performing an initial verification to check at least one of the integrity, continuity, and regularity of the route information, and performing a simulation verification to confirm the presence or absence of interference and collision between the plurality of moving objects through a simulation that virtualizes the running of the plurality of moving objects. Based on the errors generated in the initial verification and the simulation verification, it can include correcting at least one of the objects, metadata, and the route information included in the map information.

[0018] According to another aspect of the present disclosure, there is provided an autonomous driving support device that constructs an autonomous driving route using sensor recognition information. The device includes a communication unit that exchanges data with a moving object, a memory that stores at least one instruction, and a processor that executes the at least one instruction stored in the memory using the data. The moving object acquires a moving route by tracking at least one target object among the moving object and the surrounding dynamic objects around the moving object based on the recognition information acquired from a moving object having an observation sensor and a positioning sensor, and generates a predicted route from which movement is estimated based on the moving route. The processor generates route information consisting of the moving route and the predicted route, incorporates the route information into map information, and when there is a deviation of a predetermined range or more between the actual information that the target object actually moves in at least the area corresponding to the predicted route and the route information through the verification of the map information, updates the route information based on the actual information and incorporates the updated route information into the map information.

[0019] The features briefly described above about the present disclosure are merely exemplary aspects of the detailed description of the present disclosure to be described later, and do not limit the scope of the present disclosure.

Effects of the Invention

[0020] According to the present disclosure, when constructing and updating route information on map information for autonomous driving, recognition information collected from an autonomous driving mobility device is utilized to accurately and efficiently generate route information similar to an actual driving route, and a method for constructing an autonomous driving route and an autonomous driving support device can be provided.

[0021] Further, according to the present disclosure, by reflecting an actual driving route on map information, the accuracy and safety of the recognition, planning, and judgment processes of an autonomous driving mobile body can be improved, and the accuracy and productivity of map information for autonomous driving and a normal navigation system can be improved.

[0022] According to the present disclosure, by being useful for understanding and analyzing general traffic flow, it can be utilized for traffic management and operation, and can contribute to reducing traffic congestion and optimizing traffic flow.

[0023] The effects obtained in the present disclosure are not limited to the above-described effects, and other effects not described above will be clearly understood by those having ordinary knowledge in the technical field to which the present disclosure pertains from the following description.

Brief Description of the Drawings

[0024]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figures 12 - 14

Figure 15

Best Mode for Carrying Out the Invention

[0025] Hereinafter, with reference to the accompanying drawings, embodiments of the present disclosure will be described in detail so that those having ordinary knowledge in the technical field to which the present disclosure belongs can easily implement them. However, the present disclosure can be realized in various different forms and is not limited to the embodiments described herein.

[0026] When explaining the embodiments of the present disclosure, if it is determined that a specific explanation of a known configuration or function may obscure the gist of the present disclosure, the detailed explanation thereof will be omitted. In the drawings, parts not related to the description of the present disclosure are omitted, and the same reference numerals are given to the same parts.

[0027] In the present disclosure, when it is stated that a certain component is "connected", "coupled", or "connected" to another component, this can include not only a direct connection relationship but also an indirect connection relationship in which another component exists between them. Also, when it is stated that a certain component "includes" or "has" another component, this means that, unless otherwise stated to the contrary, it does not exclude other components but can further include other components.

[0028] In the present disclosure, terms such as "first" and "second" are used only for the purpose of distinguishing one component from another, and do not limit the order or importance between components unless otherwise specifically mentioned. Therefore, within the scope of the present disclosure, the first component in one embodiment may be referred to as the second component in another embodiment, and similarly, the second component in one embodiment may be referred to as the first component in another embodiment.

[0029] In the present disclosure, components that are distinguished from each other are for the purpose of clearly explaining their respective features, and do not necessarily mean that the components are separated. That is, a plurality of components may be integrated and configured as one hardware or software unit, or one component may be distributed and configured as a plurality of hardware or software units. Therefore, even without separate mention, such integrated or distributed embodiments are also included in the scope of the present disclosure.

[0030] In the present disclosure, the components described in various embodiments do not necessarily mean essential components, and some may be optional components. Therefore, embodiments constituted by a subset of the components described in one embodiment are also included in the scope of the present disclosure. Also, embodiments that further include other components in addition to the components described in various embodiments are included in the scope of the present disclosure.

[0031] In the present disclosure, each of the phrases such as "A or B", "at least one of A and B", "at least one of A or B", "A, B or C", "at least one of A, B and C", and "at least one of A, B, or C (at least one of A, B, C or combination thereof)" can include any one of the items listed together in the corresponding phrase, or all possible combinations of these.

[0032] The advantages, features, and the methods to achieve them of the present disclosure will become clear by referring to the embodiments described in detail hereinafter together with the accompanying drawings. However, the present disclosure is not limited to the embodiments disclosed below and can be realized in various different forms. However, these embodiments are merely provided to make the disclosure of the present invention complete and to fully inform those with ordinary knowledge in the technical field to which the present invention pertains of the scope of the invention.

[0033] Hereinafter, with reference to FIGS. 1 to 3, a vehicle and an autonomous driving support device for realizing a process for constructing an autonomous driving route using sensor recognition information will be described. FIG. 1 is a diagram illustrating an example in which a moving body communicates with a server and other devices to transmit and receive data. FIG. 2 is a block diagram of a moving body according to an embodiment of the present disclosure. FIG. 3 is a block diagram of a server according to another embodiment of the present disclosure.

[0034] Referring to FIG. 1, the moving body 100 can be a mobility device utilized for a specific purpose while moving on the ground, in the air, or on the sea. The moving body 100 can be, for example, a vehicle, a robot, a drone, or a ship. As an example, the moving body 100 can be a mobility device that communicates with a server 200 and other devices 300, 400 to realize autonomous movement. The moving body 100 transmits various information acquired during travel, such as recognition information based on multiple sensors, positioning information, and environmental information related to the moving route, to the server 200, and the server 200 can transmit route information, map information, driving support information, and software to the moving body 100 based on the above-described information. As another example, the moving body 100 can communicate with the server 200 and other devices 300, 400 to exchange the above-described information and acquire navigation information for guiding the moving route. In the present disclosure, the server 200 can function as an autonomous driving support device that constructs autonomous driving route information based on the behavior of the moving body 100 and the recognition information and positioning information of surrounding objects collected from the moving body 100.

[0035] In the present disclosure, an example in which the moving body 100 is a vehicle will be mainly described, but it can also be applied to other types of moving bodies described above. Hereinafter, for the sake of convenience of explanation, the moving body 100 and the vehicle may be described interchangeably.

[0036] When the moving body 100 is a vehicle, the moving body 100 can be driven based on electric energy or fossil energy. In the case of electric energy, the moving body 100 can be, for example, a pure battery-based vehicle driven only by a high-voltage battery or can adopt a gas-based fuel cell as an energy source. Also, the fuel cell can use various forms of gas that can generate electric energy, and the gas can be, for example, hydrogen. However, it is not limited to this, and various gases are applicable. In the case of fossil energy, the moving body 100 is driven based on fuel such as gasoline, light oil, or liquefied gas, and can be equipped with an engine that drives the wheel drive unit 114 by burning the fuel. The engine can be included in the energy generation unit 112 from the viewpoint of providing the driving rotational force of the wheel to the wheel drive unit 114.

[0037] The moving body 100 can be controlled and driven in autonomous driving, and the autonomous driving can be realized by semi-autonomous driving or fully autonomous driving. The fully autonomous driving can be provided as an autonomous movement in which the control unit 120 of the moving body 100 completely controls the control right without user intervention even when the driving situation is uncertain. The semi-autonomous driving can be provided as an autonomous movement in which driver intervention is required according to a specific driving situation. The semi-autonomous driving can be realized by the control unit 120 switching the control right to the user while deactivating the autonomous driving when the situation occurs, so that the user can perform manual driving.

[0038] On the one hand, the mobile body 100 can communicate with other devices 200, 300 or other vehicles 400. The other devices can include, for example, a server 200 that supports various controls, status management, and driving of the mobile body 100, an ITS (Intelligent Transportation System) device 300 for receiving information from ITS, and various types of user devices. The server 200 can transmit various information and software modules used for controlling the mobile body 100 to the mobile body 100 in response to requests and data transmitted from the mobile body 100 and user devices to support the autonomous driving and various services of the mobile body 100.

[0039] The ITS device 300 is, for example, a Road Side Unit (RSU). The ITS device 300 can exchange vehicle perception data, driving control and status data, environmental data around the vehicle, map data, etc. with the mobile body 100 via V2I to assist the user's driving of their own vehicle or support the autonomous driving of the mobile body 100. The mobile body 100 can exchange the previously listed data with other vehicles 400 via V2V to support its own driving or autonomous driving.

[0040] The mobile body 100 can communicate with other vehicles or other devices based on cellular communication, WAVE (Wireless Access in Vehicular Environment) communication, DSRC (Dedicated Short Range Communication), short-range communication, or other communication methods.

[0041] For example, for communication between the mobile body 100, the server 200, the ITS device 300, and other vehicles 400, communication networks such as LTE and 5G as cellular communication networks, a WiFi communication network, a WAVE communication network, etc. can be used. As another example, DSRC or the like used in the mobile body 100 can also be used for vehicle-to-vehicle communication. The communication methods between the mobile body 100, the server 200, the ITS device 300, other vehicles 400, and the user device are not limited to the above-described embodiments.

[0042] Referring to FIG. 2, the mobile body 100 can include a sensor unit 102, a transceiver unit 106, and a display 108.

[0043] The sensor unit 102 can include various types of detectors that sense various states and situations occurring in the external and internal environments of the mobile body 100 and grasp the positioning information of the mobile body 100. That is, the sensor unit 102 is composed of a multi-sensor module including heterogeneous sensors, and can acquire sensing data detected from each sensor.

[0044] Specifically, the sensor unit 102 can be provided with an observation sensor so as to recognize dynamic and static objects existing around the mobile body 100, and can have a positioning sensor 104d that acquires the position information and direction information of the vehicle. The observation sensor can be composed of a multi-sensor having a LiDAR sensor 104a, a camera 104b functioning as an image sensor, and a radar sensor 104c. The sensor unit 102 can acquire sensor data including recognition information and positioning information, etc. by the above-described sensors. The recognition information can include LiDAR data including three-dimensional recognition data of surrounding objects acquired by the LiDAR sensor 104a, two-dimensional image data of surrounding objects acquired by the camera 104b, and radar data for detecting the presence and movement states of surrounding objects.

[0045] The lidar sensor 104a can be a type of three-dimensional perception sensor according to the present disclosure. The lidar sensor 104a can be a sensor that observes the surrounding environment based on laser scanning and perceives the three-dimensional form of an object. Specifically, the lidar sensor 104a can irradiate a laser around the moving body 100 to obtain three-dimensional recognition data for the surrounding environment and objects. The three-dimensional recognition data can include a point cloud representing the three-dimensional form of the object, that is, detection data, and observation image data that visually shows the surrounding environment. The detection data can be provided, for example, to identify each object by showing the three-dimensional contour form of the object and the arrangement of the objects. The image data can be provided, for example, to identify the object and the surrounding environment through an image of the object and the surrounding environment.

[0046] The camera 104b can obtain two-dimensional image data for the environment around the moving body 100, objects, or image data having depth information. The radar sensor 104c can irradiate radio waves of a predetermined wavelength around, for example, and detect the behavior of an object based on the radio waves reflected from the object. The behavior of the object can include, for example, the presence of the object and the presence or absence of movement of the object, the distance between the moving body 100 and the object, the speed of the object, the moving direction, and the like.

[0047] The positioning sensor 104d can be composed of a GNSS (Global Navigation Satellite System), an IMU (Inertial Measurement Unit), an INS (Inertial Navigation System), a wheel encoder, a steering sensor, etc. to confirm positioning information including its own position, traveling attitude, and speed.

[0048] In the present disclosure, only the sensor of the sensor unit 102 referred to in the description of the embodiments will be described, and sensors for detecting various situations not listed herein can be further included.

[0049] The transceiver unit 106 can assist in mutual communication with the server 200, the ITS device 300, the surrounding vehicles 400, and the like. In the present disclosure, the transceiver unit 106 can transmit data generated or stored during travel to the server 200 and receive data and software modules transmitted from the server 200. In the present disclosure, the moving body 100 can transmit and receive data used in the method according to the present disclosure to and from the outside via the transceiver unit 106.

[0050] The display 108 can function as a user interface. The display 108 can be caused by the control unit 120 to display the operating state, control state, route / traffic information, remaining energy amount information, content requested by the driver, etc. of the moving body 100. The display 108 can display route information, map information, and various information related to the driving route transmitted from the server 200. The display 108 is composed of a touch screen capable of detecting the driver's input and can receive the driver's request for instructing the control unit 120.

[0051] On the other hand, the moving body 100 can include an actuating unit 110, an energy generation unit 112, a wheel drive unit 114, and a load device 116.

[0052] The actuating unit 110 includes at least one module for realizing a driving operation and can perform at least one driving operation such as longitudinal control like acceleration and deceleration and lateral control like steering. The actuating unit 110 can include various operating modules for generating the driving operation according to the request for the wheel drive unit 114, including pedals and steering wheels for receiving the user's request for the control.

[0053] The energy generation unit 112 can generate and supply the power and electricity used for the traveling power system such as the wheel drive unit 114 and the load device 114. When the moving body 100 is driven on an electric energy basis, the energy generation unit 112 can be composed of, for example, an electric battery, or a combination of an electric battery and a fuel cell for charging this battery. When the moving body 100 is driven on a fossil energy basis, the energy generation unit 112 can be composed of an internal combustion engine.

[0054] The wheel drive unit 114 can include a plurality of wheels, a driving force transmission module for generating a driving force and applying it to the wheels or transmitting the driving force, a braking module for decelerating the driving of the wheels, and a steering module for realizing the lateral control of the wheels. When the moving body 100 is driven on an electric energy basis, the driving force transmission module can be composed of a motor module that generates a driving force based on the power output from the electric battery. When the moving body 100 is operated on a fossil energy basis, the driving force transmission module can be provided with a transmission and a gear module for transmitting the power of the internal combustion engine.

[0055] The load device 116 is mounted on the moving body 100 and can be an auxiliary device that consumes the power supplied from the energy generation unit 112 by the use of a passenger or a user, or the power obtained by converting the output of the energy generation unit 112. In the present disclosure, the load device 116 can be a type of non-traveling electric device excluding the traveling power system such as the wheel drive unit 114. The load device 114 can be, for example, an air conditioning system, a lighting system, a seat system, and various devices installed in the moving body 100.

[0056] Also, the moving body 100 can include a storage unit 118 or a control unit 120.

[0057] The memory unit 118 stores applications and various data for the control of the mobile body 100, and can load the application and read and write data according to the request of the control unit 120. In the present disclosure, the memory unit 118 can store an application for constructing route information in the server 200 and generating object movement information related to the mobile body itself and surrounding dynamic objects for generating map information. Specifically, the memory unit 118 tracks at least one target object among the mobile body 100 and the surrounding dynamic objects around the mobile body based on the recognition information obtained from the mobile body 100 to obtain a movement route, generates a predicted route based on the movement route, and stores an application and at least one instruction for transmitting object movement information having the movement route and the predicted route to the server 200. Here, the surrounding dynamic object is a mobile object that moves on the road around the traveling road of the mobile body 100, and can be, for example, a vehicle moving on a surrounding vehicle in the same direction or the opposite direction, a vehicle traveling on each road connected to an intersection, and the like. Although a vehicle is described as an example of the surrounding dynamic object, it is not limited thereto, and various types of ground mobility moving on a road or a detailed road of a road can correspond to the surrounding dynamic object.

[0058] On the other hand, the memory unit 118 can store and manage map information including route information and various information related to the traveling route from the server 200. The map information can be used to generate a traveling route set for the mobile body 100 according to the request of the user or the control unit 120. Further, the map information can be utilized for autonomous driving and can include a low-precision map or can include a high-precision map together with the map. The map information can be provided to have various information and data including the above-described objects and environment.

[0059] The control unit 120 can perform overall control of the mobile body 100. The control unit 120 can be configured to execute applications and instructions stored in the storage unit 118. The control unit 120 can activate autonomous driving in response to an autonomous driving request based on user or mobile body 100 itself settings, and control the mobile body 100. Note that the control unit 120 can deactivate autonomous driving according to a request by user cancellation or automatic cancellation, and control the mobile body 100 to perform manual driving.

[0060] In relation to the present disclosure, the control unit 120 can track at least one target object among the mobile body 100 and surrounding dynamic objects around the mobile body based on recognition information acquired from the mobile body 100 having observation sensors 104a to 104c and a positioning sensor 104d, using applications, instructions, and data stored in the storage unit 118, and acquire a movement route. The control unit 120 can generate a predicted route where movement is estimated from the movement route based on the movement route, and transmit object movement information having the movement route and the predicted route to the server 200.

[0061] In the present disclosure, the control unit 120 can be realized by a single processing module as an example.

[0062] As another example, as shown in FIG. 4, the processing according to the above matters can be distributedly processed by a plurality of processing modules. FIG. 4 is a diagram showing modules according to the detailed functions of the mobile body and the server. A sensor module corresponds to the sensor unit 102 of the present disclosure, includes observation sensors 104a to 104c and a positioning sensor 104d, and can be composed of functional modules of a recognition information acquisition and analysis unit 122. The sensor module can perform sensor fusion and data processing, and communicate with the server 200 as necessary.

[0063] The recognition information acquisition and analysis unit 122 can further include a localization module and a perception module in addition to the sensor module. The localization module can accurately estimate and correct the position of an object using recognition information and positioning information. The localization model can handle the conversion between global and local coordinate systems. The perception module can recognize and analyze the surrounding environment and objects, grasp characteristics such as the size, shape, and speed of the objects, and obtain the trajectories or movement paths of the surrounding dynamic objects moving on the road. The prediction module corresponds to the object prediction unit 124 and can estimate the future trajectories or predicted paths of the recognized objects, such as the surrounding dynamic objects. The predicted path can be utilized for driving routes and traffic flow analysis, etc. These modules can correspond to a plurality of processing modules that constitute the control unit 120. The control unit 120 can also be collectively referred to as a plurality of processing modules in the present disclosure.

[0064] Referring to FIG. 3, as described above, the server 200 can function as an autonomous driving support device according to the present disclosure and can include a communication unit 202, a memory 204, and a processor 206.

[0065] The communication unit 202 can support mutual communication with the mobile body 100, the ITS device 300, the surrounding vehicles 400, and the like. In the present disclosure, the communication unit 202 can receive data generated or stored during the running of the mobile body 100 and the surrounding vehicles 400, and can receive data and software modules from the mobile body 100 and the surrounding vehicles 400. As illustrated in FIG. 4, a plurality of the mobile bodies 100 shown in FIGS. 1 and 2 can exist, and the communication unit 202 can receive object movement information from the plurality of mobile bodies 100. The plurality of mobile bodies are denoted as Mobility Device#1 to 4 in FIG. 4, and the mobile bodies can also be called mobile devices capable of traveling. Each of the plurality of mobile bodies 100 can track a target object based on the recognition information and the positioning information of at least one target object among the mobile body itself and the surrounding moving objects, obtain a movement route, generate a predicted route based on the movement route, and configure object movement information including the movement route and the predicted route. Further, real-time situation recognition, route update, emergency commands, etc. generated by the processor 206 based on the object movement information and various information can be transmitted to the plurality of mobile bodies 100 via the communication unit 202.

[0066] The memory 204 can store applications and various data for controlling the server 200, and can load the applications and read and write the data according to the requests of the processor 206. In the present disclosure, the memory 204 can store an application for constructing map information including route information based on the object movement information transmitted from the mobile body 100. Specifically, the memory 204 can generate route information based on the movement route and the predicted route, incorporate it into the map information, and by verifying the map information, reflect the actual movement route of the target object, that is, the actual route, in the map information, and store an application and at least one instruction for processing corrections that meet various characteristics required for the route information.

[0067] On the one hand, the memory 204 manages map information constructed to include route information and various information for planning a driving route, and can transmit the above information to a plurality of moving bodies 100 to assist autonomous driving. In the present disclosure, construction can be meant to include all of generation of map information based on route information generated primarily, and update of map information by route information reflecting an actual route, etc. In the present disclosure, the route information can provide not only real-time movement trajectories of the moving body 100 and surrounding dynamic objects, but also movement trajectories of the moving body 100 expected for each detailed lane of a road in the map information. The real-time movement trajectories can be used to establish real-time route plans of the moving body 100 executing autonomous driving and other moving bodies 100. The movement trajectories expected on the detailed lanes can correspond to estimated trajectories that are registered in the map information and are normally realized on the detailed lanes. The estimated trajectories can be provided to virtually guide the route of the moving body 100 moving in the area including the trajectories, or to establish route plans of a plurality of moving bodies 100 so that interference does not occur between the plurality of moving bodies 100. Also, the estimated trajectories can be utilized for overall traffic analysis of the area.

[0068] The processor 206 can perform overall control of the server 100. The processor 206 can be configured to execute applications and instructions stored in the memory 204.

[0069] In relation to the present disclosure, the processor 206 receives object movement information of target objects including movement routes and predicted routes from a plurality of moving bodies 100 using applications, instructions, and data stored in the memory 204, and can generate route information based on the object movement information and incorporate it into the map information. The processor 206 can update the map information with the actual route by verification of the map information, and process corrections that meet various characteristics required for the route information.

[0070] In the present disclosure, the processor 206 can be implemented, for example, as a single processing module.

[0071] As another example, as shown in FIG. 4, the processing based on the above matters can be distributed among a plurality of processing modules. FIG. 4 is a diagram showing modules according to the detailed functions of the mobile body and the server. The functions of the above-described processor 206 can be combined and executed by a map creation module, an evaluation (or verification) module, a map update module, a machine learning module, and an algorithm module. These modules can correspond to the route information construction unit 208. The map creation module can generate a precise map including route information based on static objects, recognition information of dynamic objects, and positioning information of these objects. The evaluation module can evaluate and verify the accuracy and efficiency of all data and processes of the map information. The map update module can continuously update and update the map information based on the real-time data received from the mobile body 100 and user feedback. The machine learning module can learn and optimize a model for assisting autonomous driving based on object movement information and various data, for example, a model for constructing route information and map information. This can improve operations such as object recognition and trajectory prediction. The algorithm module can control and optimize mathematical operations for performing specific operations for assisting autonomous driving, such as route planning and traffic analysis.

[0072] The functional modules of the server 200 can include a map database and a map service module in addition to the modules that make up the processor 206. The map database corresponds to the map storage unit 210 and can be built into the memory 204. The map database can hold the route information and map information constructed by the server 200. The map service module corresponds to the map information providing unit 212 and can share the map constructed by the map information server 200 with other systems, such as a plurality of moving objects, and provide various map-based services.

[0073] In the present disclosure, it is described that a plurality of moving objects 100 generate object movement information of a target object and transmit it to the server 200, and the server 200 constructs map information including route information based on the plurality of object movement information. As another example, a plurality of moving objects 100 can transmit the recognition information and positioning information detected from surrounding moving objects to the server 200, and the server 200 can construct map information based on these information. As another example, the moving object 100 can also receive recognition information, positioning information, and various information from other moving objects and the server 200, and construct map information including route information. Hereinafter, for the sake of convenience of explanation, the process of constructing an autonomous driving route according to the examples shown in FIGS. 2 and 3 will be described. Specifically, the processing of the control unit 120 of the moving object and the processor 206 of the server 200 described above will be described in detail with reference to FIGS. 5 to 13.

[0074] The method for constructing an autonomous driving route according to the present disclosure can be executed by combining the processes shown in FIGS. 5 and 6. FIG. 5 is a flowchart related to the process executed by a moving object in the method for constructing an autonomous driving route according to another embodiment of the present disclosure. FIG. 6 is a flowchart related to the process executed by a server in the method for constructing an autonomous driving route according to another embodiment of the present disclosure. Hereinafter, for the sake of convenience of explanation, the control unit 120 of the moving object 100 and the processor 206 of the server 200 that process the processes in FIGS. 5 and 6 can be abbreviated as the moving object 100 and the server 200 respectively, or these terms can be used interchangeably in the description.

[0075] Referring to FIG. 5, each of the plurality of moving objects 100 illustrated in FIG. 4 can acquire recognition information and positioning information of at least one target object among the moving object and the surrounding dynamic objects around the moving object by using observation sensors 104a to 104c and a positioning sensor 104d (S105). Since the plurality of moving objects 100 perform the process of FIG. 5 substantially in the same manner, hereinafter, for convenience of explanation, the plurality of moving objects will be referred to as the moving object 100.

[0076] Next, the moving object 100 can estimate the position of the target object based on the recognition information (S110).

[0077] The recognition information can be collected as a plurality of recognition information so as to have a plurality of viewpoints (multiview) in a time series with respect to the target object. Features are extracted for each of the plurality of recognition information, and relative displacement information between the features and the observation sensors 104a to 104c can be generated by matching between the features. The features can be, for example, lines of the target object, edges, surfaces of a predetermined shape, or geometric forms having similarity to a previously specified form. The extraction of the features and the matching between the features can be performed with reference to the movement information and the observation state information of the moving object 100. The observation state information can indicate the state in which the observation sensors 104a to 104c of the moving object 100 recognize the target object. The movement information of the moving object 100 can be generated based on, for example, the positioning information and the recognition information of the moving object 100 equipped with the observation sensors 104a to 104c that recognize the surrounding dynamic objects and move. The moving object 100 can analyze the change data of the positioning information and the change data of the recognition information to grasp its own movement trajectory. The observation state information can include the direction and posture of each sensor constituting the observation sensors 104a to 104c that observe the target object according to the movement trajectory of the moving object 100 during traveling.

[0078] The relative displacement information can be, for example, the relative distances between the target object measured by the observation sensors 104a to 104c of the moving body 100 and the observation sensors 104a to 104c, or the direction of the target object. The moving body 100 can also communicate with other moving bodies that are peripheral dynamic objects, and calculate the relative displacement information based on the positioning information of the other moving bodies and the positioning information of the moving body itself.

[0079] The position of the target object can be estimated by being generated so as to minimize the interval amount in the relative displacement information.

[0080] FIG. 7 is a diagram illustrating the result of estimating the position of an object. While moving along the travel route 506, the moving body 502 can observe the target object 504 by an observation sensor. The travel route 506 is data included in the movement information of the moving body 502, and the observation relationship 508 of the moving body 502 with respect to the target object 504 can be data included in the observation state information. The moving body 502 collects a plurality of recognition information along the travel route 506, and features can be extracted for each of the plurality of recognition information with reference to the travel route 506 and the observation relationship 508 of the moving body 502. The moving body 502 can generate relative displacement information between the observation sensor and the features through feature matching based on the travel route 506 and the observation relationship 508.

[0081] The recognition information of the target object 504 derived by the observation sensor can have recognition prediction errors 510 and 512 due to various causes. The recognition prediction errors 510 and 512 can be caused, for example, by the observation sensor of the moving body 502 and the observation recognition of the target object 504. Accordingly, the moving body 502 can take into account the recognition prediction errors 510 and 512, generate an optimal position that minimizes the separation amount in the relative displacement information, and estimate the position of the target object 504.

[0082] On the other hand, at this stage, the mobile object 100 can further include a process of providing the optimal position of the target object to the positioning information and the map information. Specifically, the estimated positions represented by the optimal positions of a plurality of target objects and the collected information of the mobile object 100 can form a geometric network. The position information of the target object can be corrected through a process of minimizing the interval amount or error between the estimated positions in the network. Integrating the above-mentioned matters, precisely corrected position information is estimated, and the corrected position information can be combined with the positioning information of the target object and the map information managed by the memory 204 so as to contribute to the operation of other systems and the generation of maps. In addition, the feature-related information used for the estimation of the optimal position and the minimization in the network can be provided to the map information. Thereby, when performing the position estimation and route planning of the mobile object 100 and other mobile objects based on the recognition information and map information of the mobile object 100, the processes related to the features due to the position change of the sensor and the position identification of the object can refer to the feature-related information. For example, the position information is estimated by a combination of the position information estimated from the position change of the sensor, and the calculation result of the position change of the feature in the sensor data can be used for the extraction and integration of the feature information.

[0083] Next, the mobile object 100 determines the target object whose position is estimated as the target object for generating object movement information, and can continuously detect the target object using the observation sensors 104a to 104c and the positioning information to generate object information (S115). The recognition by detecting the target object can be processed to grasp the form and attributes of the target object. The details of the object information will be described later in step S120.

[0084] Next, the mobile object 100 can obtain the movement route of the target object by tracking the target object using the positioning information of the target object estimated from the positioning sensor 104d and the trajectory based on the optimal position of the target object.

[0085] The trajectory of the target object based on the optimal position is the trajectory along which the continuously observed target object has moved, and the moving trajectory of the target object can be estimated using the optimal position according to the recognition information of the observation sensors 104a to 104c continuously acquired by the moving body 100 that is traveling.

[0086] As described above, the object information of the target object to be tracked basically has, for example, the position and direction of the target object, and can have the shape, color, class, identifier, moving speed, angular velocity, observation time, etc. of the object. In the case of the position information of the target object, the moving position of the center point of the target object can be adopted as the position information. The position information can be expressed in three-dimensional or two-dimensional coordinates in a predetermined coordinate system, and different coordinate systems can be used according to the situation. The relative coordinates can indicate the relative relationship between the target object and the sensor according to the coordinate system of the observation sensor. The absolute coordinates can indicate a fixed position related to the earth or a specific map. The position information is important for the accurate tracking and analysis of the target object, and can be used together with other data to model the movement and behavior of the object throughout the system. The position information of the object can be selectively converted from the relative coordinates of the sensor to the map absolute coordinates. This conversion helps to grasp the accurate distance and direction relationship between the object and the vehicle.

[0087] The orientation of the target object can indicate the direction in which the target object is facing. The direction can be expressed, for example, by representing the heading of the target object, or by the rotation values of each axis in two-dimensional or three-dimensional space, quaternion, rotation matrix, etc. The heading can simply represent the front of the object, and complex rotation information can be expressed by the rotation values for each axis in 2D or 3D. Quaternion and rotation matrix can depict rotation in three-dimensional space.

[0088] The object identifier, class, and speed information can be used for re-identifying an object in the tracking of a target object. When the same target object is tracked through multiple frames or observations of lidar data and / or image data, the object-specific identifier unique to the target object can be used to consistently identify the target object. The class indicates the type or classification of the target object and can be classified, for example, into vehicles, pedestrians, bicycles, etc. The speed information indicates the moving speed and direction of the object, which can be utilized as an important variable for predicting the future position and behavior of the object.

[0089] On the other hand, the recognition information of the observation sensors 104a to 104c can be acquired from multiple viewpoints, for example, t, t + 1, t + 2. The same target object can be repeatedly detected with the recognition information from multiple viewpoints. The moving body 100 applies a tracking algorithm and / or filtering technique (e.g., Kalman filter) that adopts related techniques to identify the same target object within the recognition information from multiple viewpoints and can track the trajectory of the identified target object, i.e., the moving target object. Also, multi-viewpoint tracking can mitigate temporary sensor noise or detection errors and can be used for accurate tracking of an object.

[0090] Based on the above, the moving body 100 can generate the position and attribute information of the target object from the recognition information of the observation sensors 104a to 104c, concatenate the object information detected from multiple viewpoints, and track the trajectory of each object. In summary, as time passes, the detected object information is concatenated, the moving body 100 tracks the motion and trajectory of each target object, and based on the motion, trajectory, and positioning information, the moving route of the target object can be obtained. Thereby, the movement pattern and behavior of the object can be analyzed. This can contribute to the safe driving of the vehicle.

[0091] Next, the moving object 100 can generate a predicted path from which movement is estimated based on the movement path traveled by the target object (S125).

[0092] The moving object 100 can generate a predicted path of the target object based on at least one of displacement information of the target object, speed information of the target object, the movement path, environmental information of the path on which the target object travels, regulation information applied to the path on which it travels, movement pattern information of a path identical or similar to the path on which it travels, and cumulative path information of a path identical or similar to the path on which it travels. The displacement information includes the position and direction of the target object based on the movement path, and the position and direction are substantially the same as those described in step S120. The environmental information can include, for example, road form related to the path on which it travels, traffic flow state, weather, road event information, etc. The environmental information can be utilized as a driving scenario composed of a combination of the exemplary data listed above, and the driving scenario can be provided in a traffic dictionary for route planning and traffic analysis. The driving scenario corresponding to the path on which it travels can be provided for the generation of the predicted path.

[0093] The regulation information can include information related to, for example, speed limits, operation caution areas, no parking or stopping areas, etc. The movement pattern information can be received from the server 200 and can include a behavior pattern analyzed based on the current travel path and / or the trajectory traveled by a past moving object on a path similar to the travel path. The cumulative path information can be a set of trajectories accumulated on a path identical or similar to the current travel path.

[0094] The displacement information of the target object and the speed information of the target object can be provided in time series. The moving object 100 can generate a predicted path using trajectory modeling including a non-linear state transition method based on the time-series displacement information and the time-series speed information. The trajectory modeling can be configured to be feedback based on the error between the predicted trajectory derived from the trajectory modeling and the trajectory derived based on the recognition information.

[0095] Specifically, when explaining trajectory modeling, the movement path of the target object based on multi-viewpoint observation information can be represented as the changes in the position, speed, and direction of the target object over time. The data related to the movement path is used to form the state space model of the dynamic system and can explain the transition from the current state to the future state of the system. Since non-linear filtering techniques, such as the Extended Kalman Filter (EKF) or the Particle Filter, are used, the above techniques can model the data.

[0096] TIFF2025081183000002.tif53170

[0097] Next, the mobile body 100 can merge the movement path and the predicted path generated for each target object to generate object movement information and transmit the object movement information to the server 200 (S130). As illustrated in FIG. 4, a plurality of mobile bodies 100 can transmit the object movement information generated by each mobile body recognizing the target object.

[0098] Referring to FIG. 6, the server 200 can acquire object movement information from a plurality of mobile bodies 100 as illustrated in FIG. 4 (S205).

[0099] Subsequently, the server 200 can integrate the plurality of object movement information of the target object to derive a single object movement information (S210).

[0100] Server 200 can perform geometric alignment between object movement information for the alignment of object movement information. Next, Server 200 can cluster the same object movement information based on statistical universe information according to the weight of the object movement information. Server 200 can generate a single object movement information based on a path model based on the clustered object movement information.

[0101] The object movement information transmitted from each mobile body 100 can be aligned using filtering, correction, and optimization techniques, and finally accurate and reliable final object movement information can be generated. Such a process plays an important role in enabling an autonomous driving mobile body to recognize and understand the environment in real time. Thereby, the mobile body 100 can have a detailed insight into the surrounding environment and can determine autonomous driving control safely and effectively. Furthermore, the ability to respond quickly and accurately to complex road conditions and unexpected situations is improved.

[0102] Regarding the above alignment, when object movement information is collected from the driving routes from multiple viewpoints of multiple mobile bodies, the object movement information from multiple mobile bodies can have uncertainty due to position estimation errors, detection, tracking, and prediction errors. Multiple object movement information with uncertainty can be aligned into a single object movement information shown as an optimal estimated value through an alignment process using mathematical and statistical methods. The uncertainty generated during the collection process of object movement information can be derived from various causes. For example, sensor accuracy limitations, environmental noise, shielding phenomena from other objects, etc. can cause position estimation errors. Also, complex situation judgments in the detection, tracking, and prediction processes can increase uncertainty.

[0103] The process of integrating a large number of object movement information with uncertainties into a single object movement information can be performed using techniques such as data fusion, filtering, and smoothing. By considering the correlation and continuity between the object movement information collected at each multi-viewpoint, the uncertainty can be reduced.

[0104] The integration process can be performed by statistical modeling and optimization. For example, Bayesian filtering such as Kalman filter and particle filter can be used to estimate the state change over time and fuse various measurement values to minimize uncertainty. The optimal single object movement path can be derived as illustrated in FIG. 8. FIG. 8 is a diagram showing an example of path information generated by integrating object movement information collected at multiple viewpoints by a large number of devices. FIG. 8 shows the trajectories of two target objects, and a plurality of object movement information 514 corresponding to each trajectory is collected. The server 200 can generate single object movement information for each trajectory through the integration.

[0105] As an example of integrating a large number of object movement information collected at multiple viewpoints in a large number of moving objects, there may be an integration process that considers the geometric characteristics between the data. The geometric integration process can perform integration through coordinate information and statistical information, and the information can be selectively utilized.

[0106] When dealing with coordinate information, each object movement information can also be recorded as relative coordinates. The object movement information needs to be converted into a unified coordinate system. In this process, by converting the relative coordinates into a common reference system, the consistency between different data can be ensured. Based on the converted coordinate information, the position data is analyzed, and this process can analyze the common information between position information, the entire set of position information, and the similarity between position information. Through such analysis, related data can be grouped together, or clustering operations can be performed to grasp the connectivity. This provides a basis for more clearly understanding the relevance between data and appropriately processing complex spatial information.

[0107] Statistical information reflects the weight and reliability of the collected data and can affect the accuracy of position estimation and object recognition. Statistical information plays an important role in evaluating the quality and accuracy of data and can be used to grasp the characteristics or patterns of object movement information under specific conditions. In the clustering process, statistical information is utilized to group or classify data with similar characteristics together.

[0108] The clustering of data is an important process for understanding and interpreting the complex patterns of travel routes, and it can be performed by selecting an appropriate geometric model of the travel route. Geometric models can be represented in various forms such as points, lines, and planes, but generally, the form of a line can be used. The linear model can be represented by polynomials, straight lines, curves, etc. according to the geometric characteristics of the travel route, and the coefficients can be estimated according to the selected mathematical model to clearly represent the route.

[0109] When estimating the coefficients of the model from a large number of clustered movement routes, the error between each movement route and the estimated model can be measured, and coefficients that minimize the error can be obtained by utilizing optimization techniques such as the least squares method. This can obtain a more accurate and consistent travel route model from the clustered data.

[0110] The estimation of model coefficients can also use a specific approach for computational efficiency. It may be an important matter to generate an initial model through random sampling and evaluate its consistency with the entire data or the sampled data. By selecting the most precise model among those estimated through iterative sampling and the estimation process, the complexity of the data can be managed, and an optimal driving route model can be effectively derived.

[0111] Next, the server 200 can generate route information based on the single object movement information determined in the above-described process. (S215).

[0112] The single object movement information inferred by the optimal driving route linear model in step S210 can be expressed in various ways to generate route information. This can fully reflect the complexity and diversity of the route information. From a geometric perspective, the route information can be expressed as linear route information 516 with vertices, as illustrated in FIG. 8, or as polygon-shaped route information 518, as illustrated in FIG. 9. FIG. 9 is a diagram showing another example of the route information. The linear route information with vertices can be expressed as a driving route 526 where a node 528 intersecting a road sign object, for example, a stop line 524, as illustrated in FIG. 10.

[0113] As another example, the route information can be composed of routes that appear by utilizing a specific mathematical model and the coefficients applied thereto. Such a representation method can accurately reflect the form and structure of the line. Note that the representation method can define various attribute information of the driving route. For example, it can include the ID of the route, the moving direction, the road type, the speed limit, the lane number, etc. Also, since it can include complex association information such as the connection relationship between driving routes and the relationship with road facilities, it can contribute to improving the consistency and efficiency with the entire traffic system. Such an integrated approach can improve the performance of the driving system by constructing the route information flexibly and accurately.

[0114] Next, the server 200 can edit the route information (S220).

[0115] The editing of the route information can involve a noise removal process and a linear representation optimization process. These processes can remove unnecessary noise on the estimated route information and convert the route into a smooth and continuous linear form.

[0116] For this purpose, various techniques and procedures can be applied, and each procedure can be adjusted according to a specific purpose. In addition to initial stages such as random sampling and model estimation, there are methods for processing data in detail. For example, various filtering techniques can be applied. Such filtering is useful for reducing noise from the original data and emphasizing the main features. Filtering techniques can be, for example, moving average, Gaussian Filter, Median Filter, Spline Interpolation, Kalman Filter, Fourier Transform, Contour Tracing, Deep Learning-based Filtering, Morphological Operations, Probabilisti Modeling, etc. within a specific window (window).

[0117] Since the moving average calculates the average of data within a specified window size, the window can move along the dataset. The moving average can smooth out short fluctuations on the path and reduce noise. The Gaussian filter can calculate the weighted average of surrounding points using the Gaussian distribution. The weights of points closer to the center are given more than those far away, and the Gaussian filter can effectively remove noise while preserving the details of the signal. The median filter can use the median value of data points within a given window. Since the value located in the center among all the values within the window is selected, it may be free from the influence of extreme values or outliers.

[0118] Spline interpolation can smoothly connect a series of points on the driving path to generate a more natural driving path. Spline interpolation can form a long and extended curve to remove unnecessary noise and angles. The Kalman filter can be used to estimate the uncertainty of a dynamic system and is useful for reducing uncertainties such as sensor noise. Fourier transform can separate noise by analyzing the frequency domain of the driving path and maintain only the necessary frequency components. This method is particularly effective for removing periodic noise from complex paths.

[0119] Contour tracing can be used when drawing lines according to specific features on the path and can form linear or curved paths with corrections as needed. Deep learning-based filtering can utilize artificial neural networks to learn and remove complex noise patterns. This method can be particularly effective for separating noise and signals in especially complex environments. Morphological operations can be used to manipulate the geometric shape of the path and are used for noise removal, hole filling, contour extraction, etc. Probabilistic modeling can use probability models to quantify the uncertainty of path information and derive the optimal path. Techniques such as Bayesian filtering fall under probabilistic modeling.

[0120] In addition, through the sophisticated combination of measured values and model values, noise can be removed more effectively. Such an overall editing process improves the accuracy of the route and enhances the applicability in the actual driving environment. The results processed in this way can provide more precise route information and guarantee robust performance in various driving scenarios and environments.

[0121] Next, the server 200 can generate connection relationship information that associates at least one of the static objects, quasi-static objects, and environmental information of the road related to the route traveled by the target object with the route information (S225).

[0122] The static objects of the road can be, for example, road information of the route traveled by the target object and infrastructure information provided on the traveled route. The road information is information that indicates or guides the route of the moving body on the road and can include, for example, the lanes of the road, stop lines, road branch points, merge points, U-turns, driving direction signs (e.g., straight ahead, left turn), etc. The infrastructure information is facilities installed on the road that affect the driving route and can be, for example, traffic signal lights, road sign boards, etc. The quasi-static objects are not objects that physically exist on the road but are factors that affect driving or the route and can include, for example, regulatory information given to the road, such as speed limits, operation caution areas, etc. The environmental information can include, for example, traffic flow conditions, weather, road event information, etc.

[0123] Specifically explaining the generation of the connection relationship information, not only the interaction between the driving routes based on the route information of the target object but also the complex connection relationship between the driving route and the related information can be defined and analyzed. The connection relationship can comprehensively consider various elements such as traffic flow, road type, safety regulations, etc. Thereby, it can be utilized in application fields such as efficient driving route planning, traffic management, and safety analysis. The definition and analysis of the connection relationship can consider complex interactions in cooperation with the topology of the road network, not just geometric and statistical analyses.

[0124] Figure 10 is a diagram showing an example of connection relationship information. Figure 10 is an illustration of the result of setting the connection relationship between driving routes based on the center line and the driving direction. The driving routes are divided into an east (E) group and a west (W) group according to the driving direction, and each lane can be classified and grouped based on the divided groups. The classification plays an important role when the autonomous driving vehicle plans a driving route and moves between lanes. Based on the center line, driving routes can be sequentially assigned to each lane, such as E-1, E-2, E-3. The driving route enables more accurate and efficient decisions regarding lane changes, left / right turns, speed adjustments, etc. of the moving object. Such a structured expression can also cooperate with a traffic management system and can be utilized for traffic flow prediction, safety analysis, etc.

[0125] The driving route can play an important role in defining the connection relationship between road structures such as stop lines, lanes, traffic signals, and signboards. As illustrated in Figure 11, through geometric intersection relationship analysis of the connection relationship between the stop line 524 and the driving route 526 of a road having a center line 520 and a lane 522, a node object 528 is generated at the intersection, and the connection relationship can be recorded in the attribute information of the driving route 526, the stop line 524, and the node object 528. Figure 11 is a diagram showing another example of connection relationship information. According to this, the driving plan and judgment logic of the autonomous driving vehicle due to changes in traffic signals, events on crosswalks, etc. can be designed more precisely and diversely. Also, the connection relationship can be utilized to improve the response ability of the autonomous driving vehicle to changes in real-time traffic conditions, road work information, emergencies, etc.

[0126] To illustrate an additional example of the cooperation between route information and related information for generating connection relationship information, there can be cooperation between the driving route and speed limit signboard information. Complicated connection relationships such as indicating the speed limit for each driving section, restrictions on driving directions, lane change information, etc. can also be precisely defined. As a result, the autonomous driving vehicle can recognize and understand the road conditions in real time and determine safe and efficient driving. Information related to the speed limit plays an important role particularly in the speed adjustment of autonomous driving vehicles and compliance with traffic regulations.

[0127] As another example, there can be cooperation between the driving route and the traffic signal system. When the state and timing of traffic signals are coordinated with the driving route, the autonomous driving vehicle can predict changes in traffic signal lights in advance and establish an efficient driving plan.

[0128] As another example, there can be cooperation between the driving route and emergency facilities. When the location information of emergency facilities such as fire stations and hospitals located within the driving route is coordinated with the driving route, more rapid and accurate responses can be made in the event of an emergency. The above cooperation can be utilized not only for route optimization of emergency vehicles but also for indicating evacuation routes of vehicles.

[0129] As another example, there can be cooperation between the driving route and weather information. By coordinating the weather information with the driving route, it is possible to anticipate and respond to road conditions associated with weather changes in advance. For example, when weather conditions such as rain, snow, fog, etc. are predicted, appropriate measures such as adjusting the driving speed in the relevant section or changing the route can be taken.

[0130] As another example, there can be cooperation between the driving route and parking facilities. When the information on currently available parking spaces in a parking lot is coordinated with the driving route, the autonomous driving vehicle can search for the nearest available parking lot in real time.

[0131] As described above, the driving route of the route information can be expressed in polygon form as shown in FIG. 9. FIG. 9 shows a driving route 518 expressed in polygon form. Each road section is composed of continuous vertices, and these vertices are combined so that the width, curvature, lane structure, etc. of the road can be accurately expressed.

[0132] The method of expressing the driving route as a polygon can accurately reflect the actual width and form of the road. The polygon model that defines the driving route in polygon form can represent the form of the road in more detail than the linear model represented by a plurality of vertices that define the outline of the road. For example, in a general linear model, only the center line of the road can be considered, but when using polygon representation, complex forms such as the boundaries on both sides of the road, lane divisions, median strips, sidewalks, etc. can also be accurately modeled.

[0133] Such a representation method helps an autonomous driving vehicle to establish an accurate driving route plan even in a complex road environment and realize safer and more efficient driving. Also, since the cooperation information with other traffic structures can be included as an attribute within the polygon, the overall characteristics of the road can be comprehensively grasped.

[0134] Next, the server 200 can incorporate the route information with which the connection relationship information is coordinated into the map information to generate map information including the route information (S230), and verify the map information to correct or update the map information according to predetermined conditions (S235).

[0135] Server 200 can correct or update the map information according to predetermined conditions while checking the map information through initial verification, simulation verification, and correction. The server 200 can confirm the presence or absence of errors in the estimated driving route through inspection, and if there is an error, the server 200 can register the correction items in the map database of the memory 204 by performing the error correction process. The inspection and correction process of the estimated driving route is necessary to ensure the accuracy and reliability of the data. The inspection and correction process is important to improve the quality of the estimated driving route and provide a driving route that matches the actual driving environment. In particular, for the safe driving of an autonomous mobile body, the above-described process is required and continuous update and management are necessary.

[0136] Initial verification can check at least one of the integrity, continuity, and regularity of the route information. Initial verification can automatically inspect, for example, the geometric integrity, continuity, regularity, etc. of the route information through a predetermined algorithm. By identifying clear errors or inconsistencies in this verification, initial filtering can be realized.

[0137] Operator inspection can be added after initial verification. Operator inspection can be performed on the route information and map information that have passed the automatic verification. In this process, the operator can confirm the accuracy of the data using reference data such as maps and satellite images. Also, complex elements such as the logical connectivity of the driving route, the relationship with adjacent roads, and the consistency of road signs and signals can be considered.

[0138] Simulation verification can be performed to check for interference and collisions between multiple mobile bodies 100 through a simulation that virtualizes the driving of the multiple mobile bodies 100. Specifically, the server 200 can virtualize the driving of at least one autonomous mobile body, examine whether there is interference or collision between the mobile bodies, and analyze whether it moves at an appropriate speed and direction. Depending on the case, the size, height, field of view angle of the sensor, weather, road conditions, etc. of the mobile body can be included as simulation input values.

[0139] Next, when a condition occurs in which there is an error in at least one of the initial verification, operator acceptance inspection, and simulation verification, the server 200 can correct at least one of the objects, metadata, and route information included in the map information based on the error.

[0140] When an error is detected, the server 200 can correct the error-related data according to a correction algorithm or an operator's request. In this process, the form, connectivity, attributes, etc. of the driving route can be corrected. Among the above-mentioned correction items, when the driving route belonging to the route information of the target object is different from the actual route information, an example in which the route information of the target object is corrected and updated will be described.

[0141] When there is a deviation of more than a predetermined range between the actual information and the route information in which the target object has actually moved in at least the area corresponding to the predicted route of the route information, the server 200 can update the route information based on the actual information. The actual route can be grasped based on the subsequent object movement information transmitted by the moving body 100 after going through the process of FIG. 5 after the route information is generated. More specifically, while the target object is actually traveling along the predicted route of the route information, the moving body 100 that acquires the moving route actually traveled on the predicted route of the target object can transmit the subsequent object movement information corresponding to the actual moving route to the server 200 in chronological order. The server 200 can generate subsequent route information based on the subsequent object movement information, at least by steps S205 to S220. The server 200 can compare the previous route information before the target object moves to the predicted route with the subsequent route information to determine whether there is a deviation of more than a predetermined range. If there is a deviation of more than a predetermined range, the server 200 can update the route information by replacing the previous route information with the subsequent route information.

[0142] FIGS. 12 to 14 are diagrams showing an example of verifying map information by updating route information.

[0143] As shown in FIG. 12, the moving object and the route information related to the surrounding moving objects can be generated. The moving object and the surrounding moving objects can be target objects related to reference numerals 530 and 532, respectively, in FIG. 12. The route information 534 and 536 illustrated in FIG. 12 includes moving routes 530 and 532 having the positions and directions of the moving object and the surrounding moving objects at times t and t + 1, and can have connection relation information in which the stop line and the traveling routes of the respective moving objects are coordinated. The route information at times t and t + 1 can correspond to the moving routes along which the moving object and the surrounding moving objects have moved until times t and t + 1.

[0144] As shown in FIG. 13, the route information 534 and 536 includes predicted routes 538 and 540 for the road sections where the moving object and the surrounding moving objects have not traveled. The route information illustrated in FIG. 13 includes the predicted routes 538 and 540 of the moving object and the surrounding moving objects at times t + 2 and t + 3, together with the moving routes 530 and 532 at times t and t + 1. The predicted routes 538 and 540 at times t + 2 and t + 3 can be estimated to include the positions and directions of the moving object and the surrounding moving objects, similar to the moving routes 530 and 532.

[0145] As shown in FIG. 14, when subsequent route information 542 and 544 including the actual routes of the moving object and the surrounding moving objects traveling on the predicted routes 538 and 540 is acquired, the server 200 can determine whether there is a deviation of a predetermined range or more between the predicted routes 538 and 540 and the actual routes via the preceding route information 534 and 536 and the subsequent route information 542 and 544. Since the deviation between the predicted route 538 of the moving object and the actual route of the moving object is less than the predetermined range, the server 200 maintains the predicted route 538 and does not update the preceding route information 534. In contrast, since the deviation between the predicted route 540 of the surrounding moving object and the actual route of the surrounding moving object is equal to or more than the predetermined range as shown in FIGS. 14 and 15, the server 200 corrects the predicted route 540 to the actual route, changes the preceding route information 536 to the subsequent route information 544, and thereby updates the route information. FIG. 15 is a diagram illustrating the update by verification of the route information.

[0146] On the one hand, the server 200 can incorporate the updated route information into the map information according to the above-mentioned matters, update the map information, and register it in the map database. The server 200 transmits the map information including the updated route information to the moving body, and the moving body can control autonomous driving based on the updated map information.

[0147] The exemplary method of the present disclosure described above is presented in a series of operations for the sake of clarity of explanation, but this is not for restricting the order in which the steps are executed. If necessary, each step may be executed simultaneously or in a different order. To implement the method according to the present disclosure, it may further include other steps in addition to the exemplified steps, or include the remaining steps excluding some steps, or include additional other steps excluding some steps.

[0148] The various embodiments of the present disclosure do not list all possible combinations, but are for explaining representative aspects of the present disclosure. The matters described in the various embodiments may be applied independently or in combinations of two or more.

[0149] Furthermore, the various embodiments of the present disclosure can be realized by hardware, firmware, software, or combinations thereof, etc. In the case of realization by hardware, it can be realized by one or more Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DSPDs), Programmable Logic Devices (PLDs), Field Programmable Gate Arrays (FPGAs), general processors, controllers, microcontrollers, microprocessors, etc.

[0150] The scope of the present disclosure includes software or machine-executable instructions (e.g., operating systems, applications, firmware, programs, etc.) that cause operations according to the methods of various embodiments to be executed on a device or computer, and non-transitory computer-readable media on which such software or instructions are stored and executable on a device or computer.

Claims

1. A method for constructing an autonomous driving route using sensor recognition information, comprising: tracking at least one target object among the moving object and the surrounding dynamic objects around the moving object based on the recognition information obtained from the moving object having an observation sensor and a positioning sensor, and obtaining a movement route; generating a predicted route where movement is estimated from the movement route based on the movement route; generating route information consisting of the movement route and the predicted route, and incorporating the route information into map information; when there is a deviation of a predetermined range or more between the actual information that the target object actually moves in at least the area corresponding to the predicted route and the route information by verification of the map information, updating the route information based on the actual information; incorporating the updated route information into the map information.

2. The step of obtaining the movement route includes tracking the target object using the positioning information of the target object estimated from the positioning sensor and a trajectory based on the optimal position of the target object. The recognition information is collected as a plurality of recognition information so as to have a multi-view in time series with respect to the target object, features are extracted for each of the plurality of recognition information, and relative displacement information between the observation sensor and the features is generated by matching between the features. The optimal position is generated so as to minimize the interval amount in the relative displacement information. The construction method according to Claim 1.

3. The extraction of the features and the matching between the features are performed with reference to the movement information and the observation state information of the moving object. The observation state information indicates the state in which the observation sensor of the moving object recognizes the target object. The construction method according to Claim 2.

4. The construction method according to Claim 2, further comprising the step of providing the optimal position of the target object to the positioning information and the map information.

5. The step of generating the predicted path includes generating the predicted path based on displacement information of the target object, velocity information of the target object, the moving path, environmental information of the path on which the target object travels, regulation information applied to the traveling path, movement pattern information of a path identical or similar to the traveling path, and cumulative path information of a path identical or similar to the traveling path. The construction method according to claim 1.

6. The displacement information of the target object and the velocity information of the target object are provided in a time series. The step of generating the predicted path includes generating the predicted path using trajectory modeling including a non-linear state transition method based on the time-series displacement information and the time-series velocity information. The trajectory modeling is configured to be feedback by an error between a predicted trajectory derived from the trajectory modeling and a trajectory derived based on the recognition information. The construction method according to claim 5.

7. The generation of the path information includes determining, as the path information, single object movement information derived by matching a plurality of object movement information of the target object. The plurality of object movement information is transmitted from each of a plurality of moving bodies that generate a moving path and a predicted path of the target object. The construction method according to claim 1.

8. The matching of the plurality of object movement information includes performing geometric matching between the object movement information, clustering similar object movement information based on statistical information according to the degree of weighting of the object movement information, and generating the single object movement information based on a path model based on the clustered object movement information and adopting the single object movement information as the path information. The construction method according to claim 7.

9. The step of incorporating the path information into the map information includes generating connection relationship information that associates at least one of road information of the path on which the target object travels, infrastructure information provided on the traveling path, weather information of the traveling path, and regulation information applied to the traveling path with the path information. The construction method according to claim 1.

10. The verification of the map information includes performing an initial verification to check at least one of the integrity, continuity, and regularity of the route information, and performing a simulation verification to confirm the presence or absence of interference and collision between the plurality of moving objects through a simulation that virtualizes the travel of the plurality of moving objects. Based on the errors generated in the initial verification and the simulation verification, the method for construction according to claim 1 includes correcting at least one of the objects, metadata, and the route information included in the map information.

11. An autonomous driving support device that constructs a route for autonomous driving using sensor recognition information, a communication unit that exchanges data with a moving object, a memory that stores at least one instruction, a processor that executes the at least one instruction stored in the memory using the data, and the moving object acquires a moving route by tracking at least one target object among the moving object and the surrounding dynamic objects around the moving object based on the recognition information acquired from a moving object having an observation sensor and a positioning sensor, and generates a predicted route from which movement is estimated from the moving route. Based on the moving route, the processor generates route information consisting of the moving route and the predicted route, and incorporates the route information into map information, when there is a deviation of a predetermined range or more between the actual information that the target object has actually moved in at least the area corresponding to the predicted route and the route information through the verification of the map information, the route information is updated based on the actual information, An autonomous driving support device configured to incorporate the updated route information into the map information.

Images