WO2021016920A1 - 可通行性的识别方法、系统、设备及计算机可读存储介质 - Google Patents

可通行性的识别方法、系统、设备及计算机可读存储介质 Download PDF

Info

Publication number
WO2021016920A1
WO2021016920A1 PCT/CN2019/098586 CN2019098586W WO2021016920A1 WO 2021016920 A1 WO2021016920 A1 WO 2021016920A1 CN 2019098586 W CN2019098586 W CN 2019098586W WO 2021016920 A1 WO2021016920 A1 WO 2021016920A1
Authority
WO
WIPO (PCT)
Prior art keywords
grid
simulation unit
access frequency
moving object
simulation
Prior art date
Application number
PCT/CN2019/098586
Other languages
English (en)
French (fr)
Inventor
李星河
邱凡
刘寒颖
Original Assignee
深圳市大疆创新科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳市大疆创新科技有限公司 filed Critical 深圳市大疆创新科技有限公司
Priority to PCT/CN2019/098586 priority Critical patent/WO2021016920A1/zh
Priority to CN201980034105.XA priority patent/CN112166446A/zh
Publication of WO2021016920A1 publication Critical patent/WO2021016920A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G06Q10/047—Optimisation of routes or paths, e.g. travelling salesman problem
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/02—Control of position or course in two dimensions
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/15—Vehicle, aircraft or watercraft design
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/25—Design optimisation, verification or simulation using particle-based methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/10—Terrestrial scenes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle

Definitions

  • the embodiments of the present invention relate to the field of unmanned driving, and in particular, to a method, system, device, and computer-readable storage medium for identifying accessibility.
  • sensors are mainly used to obtain surrounding environment information, and then the description of the surrounding environment is calculated and analyzed, that is, the accessibility of the surrounding environment, so as to provide decision-making for automatic driving planning.
  • the analysis of the accessibility is mainly to analyze the local accessibility of the 2.5D map, determine the accessibility of each grid in a small range, and then determine the accessibility of the "accessibility" in the global scope.
  • “Passable” grids are clustered, and finally the smaller fragments after clustering are filtered out, and their attributes are marked as “unknown” or “unpassable”. The remaining clustered “passable” grids are assigned their attributes Mark as "Passable”.
  • this processing method treats the 2.5D map as a binary image, and only considers whether the passable area is connected, but in fact, the autonomous vehicle has the minimum pass size, which is smaller than the minimum pass size of the opening, even if Physically connected, vehicles cannot pass.
  • the results obtained from the analysis of local accessibility are binarized, the "cost of arrival" of a connected area cannot be predicted. In summary, the recognition accuracy of the accessibility in the prior art is not high.
  • the embodiments of the present invention provide a method, system, device, and computer-readable storage medium for recognizing passability, so as to improve the recognition accuracy of passability.
  • the first aspect of the embodiments of the present invention is to provide a method for identification of accessibility, including: setting a plurality of simulation units, each of the simulation units can simulate the movement of a moving object; According to the access frequency of each grid in the pre-generated grid map during the movement of the moving object, access frequency information is obtained; wherein, the grid map is used to describe information about the surrounding environment of the moving object; Access frequency information to determine the trafficability of the surrounding environment of the moving object.
  • the second aspect of the embodiments of the present invention is to provide a viability identification system, including: a memory and a processor; the memory is used to store program code; the processor calls the program code, when the program When the code is executed, it is used to perform the following operations: set multiple simulation units, each of which can simulate the motion of the moving object; record multiple simulation units in the process of simulating the motion of the moving object
  • the access frequency of each grid in the grid map is used to obtain access frequency information; wherein, the grid map is used to describe information about the surrounding environment of the moving object; according to the access frequency information, the surrounding area of the moving object is determined Accessibility of the environment.
  • the third aspect of the embodiments of the present invention is to provide a movable platform, including:
  • the power system is installed on the fuselage to provide mobile power
  • the fourth aspect of the embodiments of the present invention is to provide a computer-readable storage medium having a computer program stored thereon, and the computer program is executed by a processor to implement the method described in the first aspect.
  • each simulation unit can simulate the movement of the moving object; recording multiple simulation units in the simulation of the movement of the moving object During the process, the access frequency information of each grid in the pre-generated grid map is obtained; the grid map is used to describe the information of the surrounding environment of the moving object; according to the access frequency information, the information of the surrounding environment of the moving object is determined Accessibility. Since the set simulation unit can simulate the movement process of the moving object on the ground, and the simulated movement process of the simulation unit is represented by the access frequency information, it can intuitively understand the trafficability of the vehicle in the surrounding environment, thereby providing automatic driving Provide basis for path planning decision.
  • Figure 1 is a schematic diagram of an application scenario provided by an embodiment of the present invention.
  • FIG. 2 is a flowchart of a method for identifying accessibility provided by an embodiment of the present invention
  • FIG. 3 is a schematic diagram of the size attributes of an analog unit provided by an embodiment of the present invention.
  • FIG. 4 is a schematic diagram of a preset step size provided by an embodiment of the present invention.
  • FIG. 5 is a schematic diagram of a rebound direction after a collision according to an embodiment of the present invention.
  • Figure 6 is a schematic diagram of a neighborhood grid provided by an embodiment of the present invention.
  • Figure 7 is a schematic diagram of exercise mileage provided by an embodiment of the present invention.
  • FIG. 8 is a flowchart of a method for identifying accessibility according to another embodiment of the present invention.
  • FIG. 9 is a diagram of the recognition result of the accessibility in the prior art.
  • Fig. 10 is the access frequency information obtained by using the method of the embodiment of the present invention.
  • FIG. 11 is a diagram of the result of accessibility recognition obtained by using the method of the embodiment of the present invention.
  • Fig. 12 is a structural diagram of a passability identification system provided by an embodiment of the present invention.
  • 121 memory; 122: processor; 123: detection device.
  • a component when a component is said to be “fixed to” another component, it can be directly on the other component or a central component may also exist. When a component is considered to be “connected” to another component, it can be directly connected to another component or there may be a centered component at the same time.
  • the embodiment of the present invention provides a method for identifying accessibility.
  • the accessibility identification method described in this embodiment can be applied to a movable platform, the movable platform is provided with a detection device, and the detection device is used to detect objects around the movable platform to obtain two-dimensional images and three-dimensional points. cloud.
  • the movable platform includes: a movable robot or a vehicle.
  • the vehicle may be an unmanned vehicle, or a vehicle equipped with an Advanced Driver Assistance Systems (ADAS) system.
  • ADAS Advanced Driver Assistance Systems
  • the method for identifying the accessibility can also be applied to a mobile robot, for example, a mobile robot equipped with a detection device that acquires information about its surrounding environment.
  • the application scenario includes at least one vehicle 11, and the vehicle 11 is a carrier equipped with a detection device.
  • the detection device may specifically be a binocular stereo camera, a time of flight (TOF) camera, and / Or Lidar.
  • TOF time of flight
  • Lidar Or Lidar
  • a binocular stereo camera and a time of flight (TOF) camera are used to collect two-dimensional images
  • a lidar is used to collect a three-dimensional point cloud.
  • the detection device may also be a monocular camera, a millimeter wave radar, etc. This embodiment is only an exemplary description and is not limited herein.
  • FIG. 2 is a flowchart of a method for identifying accessibility provided by an embodiment of the present invention. As shown in Figure 2, the method in this embodiment may include:
  • Step S201 Set multiple simulation units, each of which can simulate the movement of the moving object.
  • the moving object may be an unmanned vehicle as shown in FIG. 1.
  • the execution subject of the method in this embodiment may be an unmanned driving control device, and the unmanned control device may be a control device for unmanned driving control of an unmanned vehicle, specifically, an unmanned vehicle Driving controller.
  • the accessibility identification method can be executed by the on-board processor, or by other devices with data processing functions other than the on-board processor.
  • the server 12 shown in FIG. 1 the vehicle 11 and the server 12 can perform wireless communication or wired communication, and the vehicle 11 can send a two-dimensional image and a three-dimensional point cloud to the server 12, and the server 12 performs the identification of the accessibility. method.
  • the following will take a vehicle as an example to illustrate the method for identifying the accessibility provided by the embodiment of the present invention.
  • the simulation unit may be a particle with random walk performance, and by assigning the attribute of the moving object to the particle, the particle can simulate the movement of the moving object.
  • the number of simulation units can be set according to the computing power of the control device.
  • the simulation units are scattered in the grid map.
  • the number of simulation units can cover the entire grid map to the greatest extent in the process of simulating the movement of moving objects. Better.
  • Step S202 Record the access frequency of each grid in the pre-generated grid map by the multiple simulation units in the process of simulating the motion of the moving object to obtain access frequency information; wherein the grid map is used to describe the surrounding environment of the moving object information.
  • the pre-generated grid map can be obtained by fusion of information collected by various detection devices on the unmanned vehicle.
  • binocular stereo cameras and TOF cameras collect information about the surrounding environment of an unmanned vehicle to obtain a two-dimensional image
  • lidar collects information about the surrounding environment of an unmanned vehicle to obtain a three-dimensional point cloud.
  • the two-dimensional image and the three-dimensional point cloud Fusion can get an environmental map of the surrounding environment of the unmanned vehicle.
  • a grid map can be obtained, and the grid map includes multiple grids. For example, divide the environment map according to the square size of 0.2*0.2m, and obtain a grid map with a grid size of 0.2*0.2m.
  • the position of the unmanned vehicle on the grid map is set as the initial position of the simulation unit.
  • the position of the center of the unmanned vehicle is the origin
  • the opposite direction of the gravity direction is the Z axis
  • the front of the vehicle travel direction is the X axis
  • the left of the vehicle travel direction is the Y axis to create a grid.
  • the map coordinate system. All simulation units start from the initial position set above and simulate the movement of the unmanned vehicle on the ground in the grid map. When the simulation unit simulates the motion of the moving object, it will pass through some grids in the grid map. For a grid, each time a simulation unit passes through the grid, the access frequency of the grid is increased by 1.
  • each grid in the grid map will correspond to an access frequency, thereby forming access frequency information. If some grids have not been accessed by the analog unit, the access frequency of this grid in the access frequency information is the initial access frequency.
  • the access frequency information may be an access frequency graph. It can be understood that in other implementation manners, the access frequency information may also be in other suitable manners, such as an access frequency table, an access frequency value array, and the like.
  • Step S203 Determine the trafficability of the surrounding environment of the moving object according to the access frequency information.
  • the access frequency Information can be used to determine the passability of the surrounding environment of the moving object, or the size of the passability, which can be used as a basis for autonomous driving to provide path planning decisions.
  • each simulation unit can simulate the motion of the moving object; the access frequency of the multiple simulation units to each grid in the pre-generated grid map during the process of simulating the movement of the moving object is recorded, Obtain access frequency information; among them, the grid map is used to describe the information of the surrounding environment of the moving object; according to the access frequency information, determine the accessibility of the surrounding environment of the moving object. Since the set simulation unit can simulate the movement process of the moving object on the ground, and the simulated movement process of the simulation unit is represented by the access frequency information, it can intuitively understand the trafficability of the vehicle in the surrounding environment, thereby providing automatic driving Provide basis for path planning decision.
  • setting a plurality of simulation units includes: respectively setting the plurality of simulation units to have attributes of a moving object, and the attributes of the moving object include inherent attributes and motion attributes.
  • the inherent attributes include at least size attributes.
  • setting the multiple simulation units to have the attribute of the moving object at least includes: setting the multiple simulation units to have the size attribute respectively.
  • the size attribute means that the size of the simulation unit is not less than the size of the moving object.
  • respectively setting the multiple simulation units to have the attributes of the moving object includes: respectively setting the size of the multiple simulation units to be not smaller than the size of the moving object.
  • the size attribute of the moving object may be the width of the unmanned vehicle, that is, the size of the simulation unit is set to be no less than the width of the unmanned vehicle.
  • the width of the unmanned vehicle refers to the maximum distance of the unmanned vehicle in the Y direction on the grid coordinate system.
  • the size attribute can be set according to the maximum distance L between the two faces A and B that are arranged oppositely on the unmanned vehicle.
  • the size attribute of the analog unit is not specifically limited in this embodiment.
  • the inherent attributes of the moving object can be given to the simulation unit.
  • the size of different simulation units can be set to different sizes, but it is necessary to ensure that the size of all simulation units is not smaller than the size of the moving object.
  • the diameter of each particle is set to be not less than the vehicle width L.
  • the motion attribute includes at least one of an initial motion rate, an initial motion direction, an initial motion energy, a climbing ability, and a traffic cost.
  • setting the multiple simulation units to have the attributes of the moving object respectively includes: respectively setting the multiple simulation units to have at least one of an initial movement rate, an initial movement direction, an initial movement energy, a climbing ability, and a traffic cost.
  • the motion attribute of the moving object can be assigned to the simulation unit, so that the simulation unit can simulate the movement of the moving object. Specifically, all simulation units start from the initial position set above, and simulate the movement of the unmanned vehicle on the ground in the grid map. This means that all simulation units start from the initial position set above and follow the set initial position.
  • the movement rate and the initial movement direction move in the grid map, and a certain amount of energy will be consumed during the movement.
  • the simulation unit stops performing the simulation movement.
  • the climbing ability refers to the ability of the simulation unit to simulate a moving object when climbing on the ground.
  • different simulation units can be set to different initial movement speed, initial movement direction, initial movement energy, climbing ability, and traffic cost.
  • the inherent attributes and motion attributes of the autonomous vehicle are assigned to the simulation units, so that each simulation unit has corresponding traffic capacity.
  • the area beyond the traffic capacity of the simulation unit will not be reached by the simulation unit. Therefore, it can be well Simulate the movement of moving objects. That is, by setting the inherent attributes and motion attributes of the simulation unit, to simulate the driving restriction and passing ability of the vehicle, to ensure that the unreachable area of the vehicle will not be reached by the simulation unit, and finally the passability description of each grid in the grid map is obtained.
  • the passability not only depends on the passability of the area itself, but also depends on the passing cost of the vehicle's location to reach here. The higher the passing cost, the lower the passability of the area.
  • the access frequency of each grid in the pre-generated grid map is 0.
  • the access frequency of each grid can also be It is set to other initial values other than 0, which is not specifically limited in the present invention.
  • the access frequency of each grid in the pre-generated grid map is recorded in the process of multiple simulation units simulating the motion of the moving object, and the access frequency information is obtained, including: when the simulation unit uses the moving object in the grid map The position is the initial position, and the access frequency of the grid corresponding to the current position of the simulation unit is updated when the step length is preset based on the motion attribute; when the number of updates reaches the preset maximum number of iterations, the access frequency information is obtained.
  • updating the access frequency of the grid corresponding to the current position of the simulation unit includes: When the simulation unit takes the position of the moving object in the grid map as the initial position and presets the step length for each movement based on the motion attributes, mark the grid corresponding to the current location of the simulation unit on the pre-generated grid map, and Add 1 to the access frequency of the grid corresponding to the current location of the simulation unit.
  • a preset step length is preset for all simulation units, so that all simulation units move in the grid map according to the preset step length.
  • the access frequency of the grid corresponding to the current location of the simulation unit is updated once. For example, before an iterative movement, the access frequency of a certain grid S1 is X1, then after all simulation units have performed an iterative movement, X2 simulation units have all moved to grid S1, then the grid S1 is accessed The frequency is X1+X2.
  • the preset step size refers to the displacement of an iterative movement of the simulation unit.
  • the movement time ⁇ t required for the simulation unit to perform one iterative movement can be used to represent the prediction.
  • the position coordinates of the simulation unit before performing an iterative motion for example, the coordinates of the center of the particle
  • the position coordinates (the coordinates of the center of the particle) after the iterative motion is A 0 1
  • the displacement A 1 A 0 between the two points A 0 and A 1 is the preset step length.
  • the embodiment of the present invention may also determine the preset step length according to other position coordinates of the particles, for example, the position coordinates of a certain point on the circumference of the particle.
  • the number of updates and the number of iterations of the simulation unit may be set to be unequal, for example, the number of updates is greater than the maximum number of iterations of the simulation unit.
  • the size of the simulation unit is not less than the size of a single grid; the grid corresponding to the current location of the simulation unit refers to the grid covered by the simulation unit at the current location. Since the size of the simulation unit is not less than the size of a single grid, when the simulation unit moves to a certain grid, the grid will be overwritten, and each simulation unit will update the simulation unit every time it moves. Position coordinates. If the updated position coordinates are in a certain grid, the grid is considered to be the grid corresponding to the current location of the simulation unit.
  • the position coordinates of the simulation unit may be the center point position coordinates of the simulation unit.
  • the simulation unit since the simulation unit has a preset size attribute, the size of the simulation unit is not less than the width of the vehicle. Therefore, for the connected area that the vehicle cannot pass, the simulation unit will also be affected by the same influence during movement and cannot pass. In this way, it is possible to avoid the practically inaccessible drivable area due to the passing size caused by the connected domain method.
  • the simulation unit when the simulation unit takes the position of the moving object in the grid map as the initial position and presets the step length for each movement based on the motion attribute, after updating the access frequency of the grid corresponding to the current position of the simulation unit, this The method of the embodiment further includes: when the motion of the moving object meets the preset condition, updating the state information of the simulation unit, wherein the state information is represented by the traffic cost.
  • the traffic cost includes the collision cost; when the motion of the moving object meets the preset condition, the state information of the simulation unit is updated, including: when there is an obstacle grid in the grid corresponding to the current position of the simulation unit Or when the position coordinates of the current position of the simulation unit are the same as the position coordinates of the edge of the grid map, the movement direction of the simulation unit is updated as the rebound direction, and the collision cost of the simulation unit is calculated. Specifically, if there is an obstacle in the grid corresponding to the current location of the simulation unit or the location coordinates of the current location of the simulation unit are the same as the location coordinates of the edge of the grid map, it is considered that the simulation unit has a collision. After the collision, There will be a collision cost.
  • the collision cost of each collision can be set to a constant. Further, you can also set the single-shot generated when the grid corresponding to the current location of the simulation unit has an obstacle, and the location coordinates of the current location of the simulation unit are the same as the location coordinates of the edge of the grid map. The collision costs are different values.
  • the local slope of the grid map is greater than the maximum climbing ability, that is, if the local slope of the grid map is greater than or If it is equal to the maximum climbing ability, it is considered that the grid corresponding to the current location of the simulation unit has an obstacle grid. If the local slope of the grid map is less than the maximum climbing ability, it is considered that the grid corresponding to the current location of the simulation unit does not exist Obstacle grid.
  • the movement direction of the simulation unit is updated to the rebound direction.
  • the circle in the figure represents the simulation unit
  • the direction indicated by the V1 arrow is the movement direction of the simulation unit before the collision
  • the small square at the OC point represents In the impassable area
  • the tangent of the analog unit at the OC position is L.
  • the rebound direction is the direction of movement of the analog unit before the collision
  • V1 is the mirror vector V2 arrow of the tangent L Point to the direction.
  • the traffic cost includes the climbing cost; when the motion of the moving object meets the preset condition, the state information of the simulation unit is updated, including: when the local slope of the grid corresponding to the current position of the simulation unit is at When the projection of the motion direction of the simulation unit exceeds the preset maximum slope, the motion direction of the simulation unit is updated as the rebound direction, and the collision cost of the simulation unit is calculated; when the local slope of the grid corresponding to the current location of the simulation unit is at When the projection in the moving direction of the simulation unit does not exceed the preset maximum slope, the climbing cost of the simulation unit is calculated, where the preset maximum slope may be the climbing ability set in the initial state.
  • the climbing cost is proportional to the local slope.
  • the local slope refers to the tangent of the angle between the local normal vector of the grid map and the ground normal vector. 6, assuming a grid location corresponding to the current analog units S 23, the selected grid 4 or 8 neighborhood grid raster S 23 as its neighbor grid, In this embodiment, The 8-neighborhood is taken as an example for description.
  • grid S 23 The local slope refers to the normal vector of the local fitting surface formed by grid S 12 , S 13 , S 14 , S 22 , S 24 , S 32 , S 33 , S 34 and grid S 23 and the ground normal vector
  • calculating the collision cost of the simulation unit includes: adding a preset single collision cost on the basis of the current collision cost of the simulation unit.
  • the current collision cost of the simulation unit is P1
  • the single collision cost is P0.
  • the collision cost of the simulation unit is P1+P0.
  • the traffic cost includes motion cost; when the simulation unit takes the position of the moving object in the grid map as the initial position and presets the step length for each movement based on the motion attribute, the current position of the simulation unit is updated After the access frequency of the corresponding grid, the method of the embodiment of the present invention further includes: calculating the movement mileage of the simulation unit; and calculating the movement cost based on the movement mileage.
  • the movement mileage of the simulation unit refers to the distance between position a and position b plus position
  • the distance between b and position c, the movement mileage of the simulation unit is L1+L2, where L1 refers to the path length between position a and position b, and L2 refers to the path length between position b and position c.
  • the simulation unit when the motion of the moving object meets preset conditions, update the state information of the simulation unit, including: when the motion mileage of the simulation unit reaches the preset motion mileage, and there are no obstacles in the passing grid
  • the simulation unit is split into multiple sub-simulation units, and the multiple sub-simulation units continue to move along the random movement direction. For example, if the accumulated movement mileage of a certain simulation unit reaches the preset movement mileage, and before that, the simulation unit has passed through N grids, and none of these N grids has obstacles or the simulation unit is in When no collision occurs during the movement, the area corresponding to the N grids is considered to be an open area.
  • updating the state information of the simulation unit includes: calculating the cumulative traffic cost of the simulation unit; when the cumulative traffic cost of the simulation unit exceeds the initial energy, the simulation unit stops moving.
  • calculating the cumulative travel cost of the simulation unit includes: when the simulation unit presets the step length for each movement based on the motion attribute, calculating the travel cost of the simulation unit; from the current movement onwards, accumulate the travel costs of all sub-motions, Get the cumulative traffic cost.
  • the traffic cost includes at least one of collision cost, climbing cost and motion cost.
  • the cumulative transit cost refers to the cumulative value of the cumulative collision cost, the cumulative climb cost, and the cumulative motion cost.
  • the embodiment of the present invention explores the area in the grid map by using the random movement of particles.
  • the particles with random movement direction and speed start from the location of the vehicle.
  • the terrain features in the grid map (obstacle collision, slope, distance traveled by the particles) ) Will have a certain travel cost on the particles, and each particle has a certain initial energy to resist the external travel cost.
  • the travel cost to reach a certain area is high, it is possible to exhaust the set initial energy before reaching the area and fail to reach it. From the perspective of access frequency information, the number of particles that can reach the area is only Less, on the contrary, for the area with lower travel cost, more particles can reach, and finally, the travel frequency of the particles to a certain area is used to calculate the accessibility of the area.
  • determining the passability of the surrounding environment of the moving object according to the access frequency information includes: marking the passability and/or passability probability of the grid in the access frequency information.
  • marking the passability and/or passability of the grid includes: marking a grid with an access frequency greater than a first preset access frequency in the access frequency information as being passable, and / Or mark the pass probability greater than the first preset probability threshold; mark the grids with access frequency less than the second preset access frequency in the access frequency information as impassable, and/or pass the probability less than the second preset probability threshold;
  • a grid with an access frequency greater than or equal to the second preset access frequency and less than or equal to the first preset access frequency is marked as passable and/or a calculated pass probability, where the first preset probability threshold Can be set to be less than 100%, for example, greater than or equal to 50% and less than 100%; the second preset threshold can be set to be greater than 0, for example, greater than 0 and less than 10%, and the first preset probability threshold is greater than the second preset probability Threshold.
  • the grid is considered to be passable, and the passable passing probability can reach 100% at the maximum; if the passing probability is less than the second preset probability threshold, the grid is considered to be impassable. The lowest probability of passing can reach zero.
  • a grid with a passing probability greater than or equal to the second preset probability threshold and less than or equal to the first preset probability threshold may also be marked as possibly passable. For grids that may be able to pass, a specific pass probability can be given.
  • the passing probability is calculated according to the access frequency of the grid, the first preset access frequency, and the second preset access frequency. Specifically, it can be calculated according to the following formula:
  • P probfree is the passing probability
  • f is the access frequency of the grid after the update is completed
  • TH non-free is the second preset access frequency
  • TH free is the first preset access frequency
  • the embodiment of the present invention may also set a probability threshold, marking grids greater than or equal to the probability threshold as passable, and grids less than the probability threshold as impassable.
  • Fig. 8 is a flow chart of another method for identifying accessibility provided by an embodiment of the present invention.
  • the method of the embodiment of the present invention further includes the following step:
  • Step S801 Obtain a 2.5D grid map.
  • a 2.5D raster map describing the surrounding environment of an autonomous vehicle can be obtained by fusion of the two-dimensional image and three-dimensional point cloud collected by the detection equipment on the vehicle.
  • the horizontal plane is described by a grid of equal resolution, and the height in the grid is the maximum height value of the corresponding position of the grid.
  • the grid corresponding position refers to the grid corresponding to the actual position with the autonomous vehicle as the origin, and the grid position coordinates are (P x , P y ),
  • x res is the resolution of the 2.5D raster map in the length direction; x is the position coordinate of the vehicle's surrounding environment information on the x-axis in the vehicle coordinate system, the unit is physical length, which can be in meters unit; Means right Rounding;
  • y res is the resolution of the 2.5D raster map in the width direction; y is the position coordinate of the vehicle surrounding environment information on the y axis in the vehicle coordinate system, the unit is physical length, which can be in meters unit; Means right Rounding.
  • Step S802 For each grid in the 2.5D grid map, calculate the grid height of the grid and the corresponding neighborhood grid.
  • the grid heights of the grid and the corresponding 4 neighborhood grids, or the grid heights of 8 neighborhood grids are calculated.
  • the grid height refers to the height of the grid to the ground, that is, the height relative to the ground.
  • Step S803 Determine feature information of the grid based on the grid height of the grid and the corresponding neighborhood grid, where the feature information includes at least the maximum grid height, the minimum grid height, the average grid height, and the local slope.
  • the maximum grid height, minimum grid height, average grid height, and local slope are determined from the grid heights of 4 neighborhood grids or 8 neighborhood grids, and It is bound to the grid as the characteristic information of the grid.
  • Step S804 Determine whether the grid is a passable grid based on the characteristic information, and obtain a grid map.
  • the grid is a passable grid to obtain the grid map of the foregoing embodiment, on the basis of this grid map, further access frequency information can be obtained.
  • FIG. 9 is a diagram of the recognition result of the accessibility in the prior art.
  • FIG. 10 is access frequency information obtained by using the method of an embodiment of the present invention, where the access frequency information is in the form of an access frequency graph.
  • FIG. 11 is a diagram of the result of the accessibility recognition obtained by the method of the embodiment of the present invention.
  • Figure 9, Figure 10, and Figure 11 show the accessibility of the same environment.
  • the gray area in the figure represents the recognized passable area.
  • R the long narrow band area
  • R the width of R
  • W the width of W
  • W the width is greater than the width of the vehicle, but the area under W is narrower, therefore, the vehicle can not actually reach the W area, and the W area is actually an impassable area.
  • the lighter color area in the figure is the identified passable area.
  • the lighter the color the higher the passability.
  • the darker the color the lower the passability.
  • FIG. 12 is a structural diagram of a passability identification system provided by an embodiment of the present invention.
  • the passability identification system 120 includes: a memory 121, a processor 122, and a detection device 123; the detection device uses The two-dimensional image and three-dimensional point cloud are obtained by detecting the surrounding environment information of the movable platform; the memory 121 is used to store program code; the processor 122 calls the program code, and when the program code is executed, it is used to execute the following Operation: Based on the fusion of the two-dimensional image and the three-dimensional point cloud, a grid map is obtained; multiple simulation units are set up, each of which can simulate the movement of a moving object; and multiple simulation units are recorded According to the access frequency of each grid in the pre-generated grid map during the movement of the moving object, access frequency information is obtained; wherein, the grid map is used to describe information about the surrounding environment of the moving object; Access frequency information to determine the trafficability of the surrounding environment of
  • the processor 122 is specifically configured to: respectively set multiple simulation units to have attributes of the moving object, and the attributes of the moving object include inherent attributes and motion attributes.
  • the inherent attributes include at least a size attribute; when the processor 122 respectively sets a plurality of the simulation units to have the attributes of the moving object, it is at least configured to: respectively set a plurality of the simulation units to have the size Attributes.
  • the size attribute means that the size of the simulation unit is not less than the size of the moving object; when the processor 122 separately sets that multiple simulation units have the attribute of the moving object, it is specifically used to: The size of the multiple simulation units is set to be not smaller than the size of the moving object.
  • the motion attribute includes at least one of the initial motion rate, the initial motion direction, the initial motion energy, the climbing ability, and the traffic cost; the processor 122 has the When the attributes of the moving object are used, it is specifically used to: respectively set the plurality of simulation units to have the initial movement rate, the initial movement direction, the initial movement energy, the climbing ability, and the traffic cost. At least one of.
  • the access frequency of each grid in the pre-generated grid map is 0; the processor 122 is recording multiple
  • the simulation unit simulates the access frequency of each grid in the pre-generated grid map during the process of simulating the movement of the moving object, and obtains the access frequency information, it is specifically used for: when the simulation unit uses the moving object in The location in the grid map is the initial location, and the access frequency of the grid corresponding to the current location of the simulation unit is updated every time a preset step length is based on the motion attribute; when the number of updates reaches the preset The maximum number of iterations to obtain the access frequency information.
  • the processor 122 updates the The access frequency of the grid corresponding to the current location of the simulation unit is specifically used to: when the simulation unit takes the position of the moving object in the grid map as the initial position, and moves based on the motion attribute
  • the step size is preset, the grid corresponding to the current location of the simulation unit is marked on the pre-generated grid map, and the location of the grid corresponding to the current location of the simulation unit is added. Add 1 to the frequency of access.
  • the size of the simulation unit is not less than the size of a single grid; the grid corresponding to the current location of the simulation unit refers to the coverage of the simulation unit at the current location Grid.
  • the processor 122 updates the After the access frequency of the grid corresponding to the current position of the simulation unit, it is also used to update the state information of the simulation unit when the motion of the moving object meets a preset condition, and the state information is determined by the traffic cost. Said.
  • the traffic cost includes a collision cost
  • the processor 122 updates the state information of the simulation unit when the motion of the moving object meets a preset condition, it is specifically configured to: When the grid corresponding to the current position has an obstacle grid or the position coordinates of the current position of the simulation unit are the same as the position coordinates of the edge of the grid map, the movement direction of the simulation unit is updated as the rebound direction , And calculate the collision cost of the simulation unit.
  • the traffic cost includes a climbing cost
  • the processor 122 updates the state information of the simulation unit when the motion of the moving object meets a preset condition, it is specifically configured to: When the projection of the local slope of the grid corresponding to the current position of the unit on the motion direction of the simulation unit exceeds the preset maximum slope, the motion direction of the simulation unit is updated as the rebound direction, and the simulation unit is calculated When the projection of the local slope of the grid corresponding to the current position of the simulation unit in the motion direction of the simulation unit does not exceed the preset maximum slope, the climbing cost of the simulation unit is calculated .
  • the processor 122 calculates the collision cost of the simulation unit, it is specifically configured to: add a preset single collision cost on the basis of the current collision cost of the simulation unit.
  • the climbing cost is proportional to the slope.
  • the traffic cost includes a motion cost
  • the processor 122 uses the position of the moving object in the grid map as an initial position when the simulation unit is used, and performs a motion prediction based on the motion attribute.
  • the step size is set, after updating the access frequency of the grid corresponding to the current position of the simulation unit, it is also used to: calculate the movement mileage of the simulation unit; and calculate the movement cost based on the movement mileage.
  • the processor 122 updates the state information of the simulation unit when the movement of the moving object meets a preset condition, it is specifically configured to: when the movement mileage of the simulation unit reaches a preset movement When the mileage and no obstacle grid exists in the marked grid, the simulation unit is split into multiple sub-simulation units, and the multiple sub-simulation units continue to move in a random direction of movement.
  • the processor 122 updates the state information of the simulation unit when the motion of the moving object satisfies a preset condition, it is specifically configured to: calculate the cumulative traffic cost of the simulation unit; When the accumulated traffic cost of the simulation unit exceeds the initial energy, the simulation unit stops moving.
  • the processor 122 calculates the cumulative travel cost of the simulation unit, it is specifically configured to: when the simulation unit presets the step size per exercise based on the motion attribute, calculate the Travel cost; from the current movement onwards, accumulate the transit cost of all sub-sports to obtain the cumulative transit cost.
  • the traffic cost includes at least one of collision cost, climbing cost, and motion cost.
  • the processor 122 determines the accessibility of the surrounding environment of the moving object according to the access frequency information, it is specifically configured to: in the access frequency information, mark the accessibility of the grid Sex and/or passability.
  • the processor 122 marks the passability and/or passability probability of the grid in the access frequency information, it is specifically configured to: in the access frequency information, the access frequency is greater than the first A grid with a preset access frequency is marked as passable and/or a pass probability is greater than the first preset probability threshold; the grid with an access frequency less than the second preset access frequency in the access frequency information is marked as unavailable Access, and/or access probability is less than a second preset probability threshold; the access frequency in the access frequency information is greater than or equal to the second preset access frequency and the grid is less than or equal to the first preset access frequency , Marked as passable and/or calculated passing probability, wherein the first preset probability threshold is greater than the second preset probability threshold.
  • the passing probability is calculated according to the access frequency of the grid, the first preset access frequency, and the second preset access frequency.
  • the moving object is a movable robot or a vehicle.
  • the vehicle may be an unmanned vehicle or a vehicle equipped with an ADAS system.
  • the simulation unit is a particle with random walk performance.
  • the processor 122 records the access frequency of each grid in the pre-generated grid map during the process of recording the multiple simulation units to simulate the motion of the moving object, and before obtaining the access frequency information, it also uses In: Obtain a 2.5D grid map; for each grid in the 2.5D grid map, calculate the grid height of the grid and the corresponding neighborhood grid; based on the grid and the corresponding neighborhood The grid height of the domain grid is used to determine the feature information of the grid, the feature information includes at least the maximum grid height, the minimum grid height, the average grid height, and the local slope; the judgment is based on the feature information Whether the grid is a passable grid, the pre-generated grid map is obtained.
  • each simulation unit can simulate the motion of the moving object; the access frequency of the multiple simulation units to each grid in the pre-generated grid map during the process of simulating the movement of the moving object is recorded, Obtain access frequency information; among them, the grid map is used to describe the information of the surrounding environment of the moving object; according to the access frequency information, determine the accessibility of the surrounding environment of the moving object. Since the set simulation unit can simulate the movement process of the moving object on the ground, and the simulated movement process of the simulation unit is represented by the access frequency information, it can intuitively understand the trafficability of the vehicle in the surrounding environment, thereby providing automatic driving Provide basis for path planning decision.
  • the embodiment of the present invention provides a movable platform.
  • the movable platform includes: a fuselage, a power system, and the accessibility identification system described in the above-mentioned embodiments; wherein the power system is installed on the fuselage to provide moving power.
  • the movable platform includes: a movable robot or a vehicle.
  • the fuselage may include the vehicle body, chassis and other bearing parts.
  • the accessibility identification system When the accessibility identification system is a separate device, it can be integrated on the movable platform such as a vehicle in the form of front or rear installation; when the accessibility identification system is a distributed system, its Various parts such as detection equipment, processor and memory can be installed in the same or different positions of the movable platform.
  • this embodiment also provides a computer-readable storage medium on which a computer program is stored, and the computer program is executed by a processor to implement the accessibility identification method described in the foregoing embodiment.
  • the disclosed device and method may be implemented in other ways.
  • the device embodiments described above are only illustrative.
  • the division of the units is only a logical function division, and there may be other divisions in actual implementation, for example, multiple units or components can be combined or It can be integrated into another system, or some features can be ignored or not implemented.
  • the displayed or discussed mutual coupling or direct coupling or communication connection may be indirect coupling or communication connection through some interfaces, devices or units, and may be in electrical, mechanical or other forms.
  • the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the objectives of the solutions of the embodiments.
  • the functional units in the various embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit.
  • the above-mentioned integrated unit may be implemented in the form of hardware, or may be implemented in the form of hardware plus software functional units.
  • the above-mentioned integrated unit implemented in the form of a software functional unit may be stored in a computer readable storage medium.
  • the above-mentioned software functional unit is stored in a storage medium and includes several instructions to make a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor execute the method described in the various embodiments of the present invention. Part of the steps.
  • the aforementioned storage media include: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disk and other media that can store program code .

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Geometry (AREA)
  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Automation & Control Theory (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Computer Hardware Design (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Pure & Applied Mathematics (AREA)
  • Development Economics (AREA)
  • Mathematical Optimization (AREA)
  • Game Theory and Decision Science (AREA)
  • Mathematical Analysis (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Computational Mathematics (AREA)
  • Multimedia (AREA)
  • Traffic Control Systems (AREA)

Abstract

提供可通行性的识别方法、系统、设备及计算机可读存储介质，该方法包括：(S201)设置多个模拟单元，每个模拟单元能够模拟运动对象的运动；(S202)记录多个模拟单元在模拟运动对象运动的过程中对预先生成的栅格地图中每个栅格的访问频率，得到访问频率信息；其中，栅格地图用于描述运动对象周围环境的信息；(S203)根据访问频率信息，确定运动对象周围环境的可通行性。该方法能够提高可通行性的识别精度。

Description

可通行性的识别方法、系统、设备及计算机可读存储介质 技术领域
本发明实施例涉及无人驾驶领域，尤其涉及一种可通行性的识别方法、系统、设备及计算机可读存储介质。
背景技术
在自动驾驶中，主要是利用传感器获取周边环境信息，进而通过计算分析出周围环境的描述，即周围环境的可通行性，从而为自动驾驶规划提供决策。
目前，可通行性的分析主要是对2.5D地图进行局部可通行性分析，确定每个栅格在小范围内二值化描述的可通行性，然后在全局范围内对可通行性为“可通行”的栅格做连通聚类，最后过滤掉聚类后尺寸较小的碎片，标记其属性为“未知”或“不可通行”，其余聚类所得的“可通行”栅格，将其属性标记为“可通行”。这种处理方法在连通阶段，是将2.5D地图作为一幅二值图来处理，只考虑了可通行区域是否连通，但是实际上自动驾驶车辆具有最小通过尺寸，小于最小通过尺寸的开口，即使物理上连通，车辆也无法通过。另外，由于局部可通行性分析得到的结果是二值化的，也无法预知某一连通区域的“到达代价”。综上，现有技术中对于可通行性的识别精度不高。
发明内容
本发明实施例提供一种可通行性的识别方法、系统、设备及计算机可读存储介质，以提高可通行性的识别精度。
本发明实施例的第一方面是提供一种可通行性的识别方法，包括：设置多个模拟单元，每个所述模拟单元能够模拟运动对象的运动；记录多个所述模拟单元在模拟所述运动对象运动的过程中对预先生成的栅格地图中每个栅格的访问频率，得到访问频率信息；其中，所述栅格地图用于描述所述运动对象周围环境的信息；根据所述访问频率信息，确定所述运动 对象周围环境的可通行性。
本发明实施例的第二方面是提供一种可通行性的识别系统，包括：包括：存储器和处理器；所述存储器用于存储程序代码；所述处理器，调用所述程序代码，当程序代码被执行时，用于执行以下操作：设置多个模拟单元，每个所述模拟单元能够模拟运动对象的运动；记录多个所述模拟单元在模拟所述运动对象运动的过程中对预先生成的栅格地图中每个栅格的访问频率，得到访问频率信息；其中，所述栅格地图用于描述所述运动对象周围环境的信息；根据所述访问频率信息，确定所述运动对象周围环境的可通行性。
本发明实施例的第三方面是提供一种可移动平台，包括：
机身；
动力系统，安装在所述机身，用于提供移动动力；
以及权利要求第一方面所述的方法。
本发明实施例的第四方面是提供一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行以实现第一方面所述的方法。
本实施例提供的可通行性的识别方法、系统、设备及计算机可读存储介质，通过设置多个模拟单元，每个模拟单元能够模拟运动对象的运动；记录多个模拟单元在模拟运动对象运动的过程中对预先生成的栅格地图中每个栅格的访问频率，得到访问频率信息；其中，栅格地图用于描述运动对象周围环境的信息；根据访问频率信息，确定运动对象周围环境的可通行性。由于设置的模拟单元能够模拟运动对象在地面上的运动过程，且模拟单元的模拟运动过程采用访问频率信息来表示，因此，可以直观地了解车辆在周围环境中的可通行性，从而为自动驾驶路径规划决策提供依据。
附图说明
为了更清楚地说明本发明实施例中的技术方案，下面将对实施例描述中所需要使用的附图作一简单地介绍，显而易见地，下面描述中的附图是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动性的前提下，还可以根据这些附图获得其他的附图。
图1为本发明实施例提供的应用场景示意图；
图2为本发明实施例提供的可通行性的识别方法的流程图；
图3为本发明实施例提供的模拟单元尺寸属性示意图；
图4为本发明实施例提供的预设步长示意图；
图5为本发明实施例提供的碰撞后反弹方向示意图；
图6为本发明实施例提供的邻域栅格示意图；
图7为本发明实施例提供的运动里程示意图；
图8为本发明另一实施例提供的可通行性的识别方法的流程图；
图9为现有技术中可通行性识别结果图。
图10为采用本发明实施例的方法得到的访问频率信息。
图11为采用本发明实施例的方法得到的可通行性识别结果图；
图12为本发明实施例提供的可通行性的识别系统的结构图。
附图标记：
11：车辆；      12：服务器；
120：可通行性的识别系统；
121：存储器；   122：处理器；  123：探测设备。
具体实施方式
下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚地描述，显然，所描述的实施例仅仅是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。
需要说明的是，当组件被称为“固定于”另一个组件，它可以直接在另一个组件上或者也可以存在居中的组件。当一个组件被认为是“连接”另一个组件，它可以是直接连接到另一个组件或者可能同时存在居中组件。
除非另有定义，本文所使用的所有的技术和科学术语与属于本发明的技术领域的技术人员通常理解的含义相同。本文中在本发明的说明书中所使用的术语只是为了描述具体的实施例的目的，不是旨在于限制本发明。本文所使用的术语“及/或”包括一个或多个相关的所列项目的任意的和所有的组合。
下面结合附图，对本发明的一些实施方式作详细说明。在不冲突的情 况下，下述的实施例及实施例中的特征可以相互组合。
本发明实施例提供一种可通行性的识别方法。本实施例所述的可通行性的识别方法可以应用于可移动平台，所述可移动平台设置有探测设备，所述探测设备用于探测所述可移动平台周围物体得到二维图像和三维点云。可选的，所述可移动平台包括：可移动机器人或车辆。
以可移动平台是车辆为例，该车辆可以是无人驾驶车辆，或者是搭载有高级辅助驾驶(Advanced Driver Assistance Systems，ADAS)系统的车辆等。可以理解的是，可通行性的识别方法还可以应用于可移动机器人上，例如，搭载有获取其周围环境的信息的探测设备的可移动机器人。如图1所示，该应用场景包括至少一个车辆11，车辆11为搭载有探测设备的载体，该探测设备具体可以是双目立体相机、飞行时间测距法(Time of flight，TOF)相机和/或激光雷达。车辆11在行驶的过程中，探测设备实时探测车辆11周围环境得到二维图像和三维点云。其中，双目立体相机、飞行时间测距法(Time of flight，TOF)相机用于采集二维图像，激光雷达用于采集三维点云。可以理解，在其他实施例中，探测设备也可以是单目相机、毫米波雷达等，本实施例仅为示例性说明，在此不作限定。
本发明实施例提供一种可通行性的识别方法。图2为本发明实施例提供的可通行性的识别方法的流程图。如图2所示，本实施例中的方法，可以包括：
步骤S201、设置多个模拟单元，每个模拟单元能够模拟运动对象的运动。
本实施例中，运动对象可以是如图1所示的无人驾驶车辆。可选的，本实施例方法的执行主体可以是无人驾驶控制设备，该无人驾驶控制设备可以是对无人驾驶车辆进行无人驾驶控制的控制设备，具体地，可以是无人驾驶车辆的驾驶控制器。本实施例并不限定可通行性的识别方法的执行主体，该可通行性的识别方法可以由车载的处理器执行，也可以由车载的处理器之外的其他具有数据处理功能的设备执行，例如，如图1所示的服务器12，车辆11和服务器12可进行无线通信或有线通信，车辆11可以将二维图像和三维点云发送给服务器12，由服务器12执行该可通行性的 识别方法。下面将以车辆为例说明本发明实施例提供的可通行性的识别方法。
其中，模拟单元能够模拟运动对象的运动是指模拟单元能够模拟无人驾驶车辆在地面上的运动。例如，模拟无人驾驶车辆在地面上朝着某一方向以某一速度和加速度行驶。
可选的，模拟单元可以是具有随机游走性能的粒子，通过对粒子赋予运动对象的属性，使粒子来模拟运动对象的运动。
本实施例中，模拟单元的数量可以根据控制设备的计算能力来设定，模拟单元在栅格地图中分散分布，模拟单元的数量以在模拟运动对象运动的过程中最大程度覆盖整个栅格地图为佳。
步骤S202、记录多个模拟单元在模拟运动对象运动的过程中对预先生成的栅格地图中每个栅格的访问频率，得到访问频率信息；其中，栅格地图用于描述运动对象周围环境的信息。
如图1所示，预先生成的栅格地图可以由无人驾驶车辆上的各个探测设备采集的信息融合得到。例如，双目立体相机和TOF相机对无人驾驶车辆周围环境信息进行采集得到二维图像，激光雷达对无人驾驶车辆周围环境信息进行采集得到三维点云，通过将二维图像和三维点云融合，可以得到无人驾驶车辆周围环境的环境地图。将环境地图按照预先设定的栅格大小划分后可以得到栅格地图，栅格地图中包括多个栅格。例如，按照0.2*0.2m的正方形大小划分环境地图，得到栅格大小为0.2*0.2m的栅格地图。
在一个具体的实施方式中，无人驾驶车辆在栅格地图中所处的位置设定为模拟单元的初始位置。在栅格地图中，以无人驾驶车辆的中心所处的位置为原点，重力方向的反方向为Z轴，车辆行驶方向的前方为X轴，车辆行驶方向的左方为Y轴建立栅格地图坐标系。所有模拟单元从上述设定的初始位置出发，在栅格地图中模拟无人驾驶车辆在地面上的运动。在模拟单元模拟运动对象运动的过程中，会经过栅格地图中的一些栅格，对一个栅格来说，每有一个模拟单元经过该栅格时，则该栅格的访问频率加1。在模拟单元模拟运动对象运动的过程结束时，栅格地图中每个栅格会对应有一个访问频率，从而形成访问频率信息。若一些栅格未被模拟单元访问 过，则访问频率信息中该栅格的访问频率为初始访问频率。在一种实施例中，该访问频率信息可以为访问频率图，可以理解，在其他实施方式中，该访问频率信息也可以是其他合适的方式，例如，访问频率表、访问频率数值阵列等。
步骤S203、根据访问频率信息，确定运动对象周围环境的可通行性。
具体地，通过访问频率信息，可以了解所有模拟单元对栅格地图中各个栅格的访问情况，而每个栅格的访问频率与该栅格的可通行性大小正相关，因此，根据访问频率信息，可以确定运动对象周围环境的可通行性，或者可通行性的大小，从而作为自动驾驶提供路径规划决策的依据。
本实施例通过设置多个模拟单元，每个模拟单元能够模拟运动对象的运动；记录多个模拟单元在模拟运动对象运动的过程中对预先生成的栅格地图中每个栅格的访问频率，得到访问频率信息；其中，栅格地图用于描述运动对象周围环境的信息；根据访问频率信息，确定运动对象周围环境的可通行性。由于设置的模拟单元能够模拟运动对象在地面上的运动过程，且模拟单元的模拟运动过程采用访问频率信息来表示，因此，可以直观地了解车辆在周围环境中的可通行性，从而为自动驾驶路径规划决策提供依据。
具体的，设置多个模拟单元，包括：分别设置多个模拟单元具有运动对象的属性，运动对象的属性包括固有属性和运动属性。
可选的，固有属性至少包括尺寸属性。相应的，分别设置多个模拟单元具有运动对象的属性，至少包括：分别设置多个模拟单元具有尺寸属性。可选的，尺寸属性是指模拟单元的尺寸不小于运动对象的尺寸。相应的，分别设置多个模拟单元具有运动对象的属性，包括：分别设置多个模拟单元的尺寸为不小于运动对象的尺寸。本实施例中，运动对象的尺寸属性可以是无人驾驶车辆的宽度，即设置模拟单元的尺寸为不小于无人驾驶车辆的宽度。具体的，无人驾驶车辆的宽度是指无人驾驶车辆在栅格坐标系上Y方向上的最大距离。如图3所示，可以根据无人驾驶车辆上相对设置的两个面A、B之间的最大距离L来设置尺寸属性，当然，本领域技术人员也可以根据车辆上的其他参照物来设置模拟单元的尺寸属性，本实施例对 此不做具体限定。通过本实施例，能够为模拟单元赋予运动对象的固有属性。可选的，不同模拟单元的尺寸可以设置为不同的大小，但需保证所有模拟单元的尺寸均不小于运动对象的尺寸。例如，将每个粒子的直径设置为不小于车辆宽度L。
可选的，运动属性包括初始运动速率、初始运动方向、初始运动能量、爬坡能力和通行代价中的至少一项。相应的，分别设置多个模拟单元具有运动对象的属性，包括：分别设置多个模拟单元具有初始运动速率、初始运动方向、初始运动能量、爬坡能力和通行代价中的至少一项。通过本实施例，能够为模拟单元赋予运动对象的运动属性，从而使模拟单元模拟运动对象的运动。具体的，所有模拟单元从上述设定的初始位置出发，在栅格地图中模拟无人驾驶车辆在地面上的运动，是指所有模拟单元从上述设定的初始位置出发，按照设定的初始运动速率和初始运动方向在栅格地图中运动，运动过程中会消耗一定的能量，当消耗的能量达到初始能量时，则该模拟单元停止进行模拟运动。爬坡能力是指模拟单元模拟运动对象在地面上爬坡时的能力。可选的，不同的模拟单元可以设置为不同的初始运动速率、初始运动方向、初始运动能量、爬坡能力和通行代价。
本实施例通过对模拟单元赋予自动驾驶车辆的固有属性和运动属性，从而使得每个模拟单元具有相应的通行能力，超出模拟单元通行能力的区域，模拟单元不会到达，因此，能够很好地模拟运动对象的运动。即通过设置模拟单元的固有属性和运动属性，来模拟车辆的行驶限制和通过能力，确保车辆不可达的区域，模拟单元不会到达，最终得到栅格地图中每个栅格的可通行性描述，对于某个区域的可通行性，不仅依赖于其区域本身的通行性，还依赖于车辆所处位置到达此处的通行代价，通行代价越高代表该区域的可通行性越低。
可选的，在模拟单元模拟运动对象的运动之前，即模拟单元的初始状态下，预先生成的栅格地图中每个栅格的访问频率为0，当然，每个栅格的访问频率也可以设置为非0的其他初始数值，本发明对此不做具体限定。
相应的，记录多个模拟单元模拟运动对象运动的过程中对预先生成的栅格地图中每个栅格的访问频率，得到访问频率信息，包括：当模拟单元 以运动对象在栅格地图中的位置为初始位置，并基于运动属性每运动预设步长时，更新模拟单元当前所处位置对应的栅格的访问频率；当更新次数达到预设的最大迭代次数，得到访问频率信息。
可选的，当模拟单元以运动对象在栅格地图中的位置为初始位置，并基于运动属性每运动预设步长时，更新模拟单元当前所处位置对应的栅格的访问频率，包括：当模拟单元以运动对象在栅格地图中的位置为初始位置，并基于运动属性每运动预设步长时，在预先生成的栅格地图上标记模拟单元当前所处位置对应的栅格，并将模拟单元当前所处位置对应的栅格的访问频率加1。
本实施例中，是对所有模拟单元预先设置一预设步长，使所有模拟单元按照该预设步长在栅格地图中进行运动，当模拟单元每运动一个预设步长时，认为该模拟单元进行了一次迭代运动，则更新一次模拟单元当前所处位置对应的栅格的访问频率。例如，在一次迭代运动之前，某一个栅格S1的访问频率为X1，则在所有模拟单元进行了一次迭代运动之后，有X2个模拟单元都运动至栅格S1，则该栅格S1的访问频率为X1+X2。其中，预设步长是指模拟单元进行一次迭代运动的位移，由于每个模拟单元具有初始运动速度和初始运动方向，因此，可以采用模拟单元进行一次迭代运动所需要的运动时间δt来表示预设步长，即预设步长＝初始运动速度*δt。如图4所示，假设模拟单元在进行一次迭代运动之前所处的位置坐标(例如粒子的圆心坐标)为A 0，进行一次迭代运动后所处的位置坐标(粒子的圆心坐标)变为了A 1，则A 0与A 1两个点之间的位移A 1A 0为预设步长。当然，本发明实施例也可以根据粒子的其他位置坐标来确定预设步长，例如，粒子的圆周上某一个点的位置坐标。
可选的，更新次数与模拟单元的迭代次数可以设置为不相等，例如更新次数大于模拟单元的最大迭代次数。
可选的，模拟单元的尺寸不小于单个栅格的尺寸；模拟单元当前所处位置对应的栅格是指模拟单元在当前所处位置处覆盖的栅格。由于模拟单元的尺寸不小于单个栅格的尺寸，因此，模拟单元运动至某一个栅格时，会将该栅格覆盖，而每个模拟单元每迭代运动一次，就会更新一次该模拟 单元的位置坐标，若更新的位置坐标位于某一个栅格内，则认为该栅格是该模拟单元当前所处位置对应的栅格。可选的，模拟单元的位置坐标可以取模拟单元的中心点位置坐标。
本发明实施例中，由于模拟单元具有预先设定的尺寸属性，模拟单元的尺寸不小于车宽，因此对于车辆无法通过的连通区域，模拟单元在运动时也会受到相同的影响而无法通过，由此可以避免连通域方法导致的由于通过尺寸产生的实际无法到达的可行驶区域。
可选的，当模拟单元以运动对象在栅格地图中的位置为初始位置，并基于运动属性每运动预设步长时，更新模拟单元当前所处位置对应的栅格的访问频率之后，本实施例的方法还包括：当运动对象的运动满足预设条件时，更新模拟单元的状态信息，其中，状态信息由通行代价表示。
在一种实施方式中，通行代价包括碰撞代价；则当运动对象的运动满足预设条件时，更新模拟单元的状态信息，包括：当模拟单元当前所处位置对应的栅格存在障碍物栅格或者模拟单元当前所处位置的位置坐标与栅格地图的边缘的位置坐标相同时，更新模拟单元的运动方向为反弹方向，并计算模拟单元的碰撞代价。具体的，若模拟单元当前所处位置对应的栅格存在障碍物或者模拟单元当前所处位置的位置坐标与栅格地图的边缘的位置坐标相同，则认为该模拟单元发生碰撞，发生碰撞后，会产生碰撞代价。可选的，每一次的碰撞所产生的碰撞代价可以设置为一常数。进一步的，还可以设置模拟单元当前所处位置对应的栅格存在障碍物的情况下，和模拟单元当前所处位置的位置坐标与栅格地图的边缘的位置坐标相同的情况下产生的单次碰撞代价为不同值。
可选的，判断模拟单元当前所处位置对应的栅格是否存在障碍物栅格时，可以通过栅格地图的局部坡度是否大于最大爬坡能力来判断，即若栅格地图的局部坡度大于或等于最大爬坡能力，则认为模拟单元当前所处位置对应的栅格存在障碍物栅格，若栅格地图的局部坡度小于最大爬坡能力，认为模拟单元当前所处位置对应的栅格不存在障碍物栅格。
具体的，对于模拟单元的运动方向更新为反弹方向，如图5所示，图中圆形代表模拟单元，V1箭头所指方向为模拟单元发生碰撞前的运动方向， OC点处的小方块代表不可通行区域，OC位置处模拟单元的切线为L，当模拟单元在不可通行区域的OC点处发生碰撞时，反弹方向为模拟单元发生碰撞前的运动方向V1关于切线L的镜像向量V2箭头所指方向。
在另一种实施方式中，通行代价包括爬坡代价；当运动对象的运动满足预设条件时，更新模拟单元的状态信息，包括：当模拟单元当前所处位置对应的栅格的局部坡度在模拟单元的运动方向上的投影超过预设的最大坡度时，则更新模拟单元的运动方向为反弹方向，并计算模拟单元的碰撞代价；当模拟单元当前所处位置对应的栅格的局部坡度在模拟单元的运动方向上的投影未超过预设的最大坡度时，则计算模拟单元的爬坡代价，其中，预设的最大坡度可以是初始状态下设定的爬坡能力。可选的，爬坡代价与局部坡度成比例。具体的，局部坡度是指栅格地图的局部法向量与地面法向量的夹角的正切值。如图6所示，假设模拟单元当前所处位置对应的栅格为S 23，则选取栅格S 23邻域内的4个栅格或者8个栅格作为其邻域栅格，本实施例以8邻域为例进行说明，假设栅格S 23的8邻域栅格分别为S 12、S 13、S 14、S 22、S 24、S 32、S 33、S 34，则栅格S 23的局部坡度是指栅格S 12、S 13、S 14、S 22、S 24、S 32、S 33、S 34与栅格S 23所形成的局部拟合面的法向量与地面法向量之间夹角 α的tan α值。本实施方式中对于碰撞代价的计算可以参见前一实施方式中关于碰撞代价的计算过程，本实施例在此不再赘述。
在前述两种实施方式中，计算模拟单元的碰撞代价，包括：在模拟单元的当前碰撞代价的基础上增加预设的单次碰撞代价。例如，模拟单元的当前碰撞代价为P1，单次碰撞代价为P0，则在发生一次碰撞后，模拟单元的碰撞代价为P1+P0。
在又一种实施方式中，通行代价包括运动代价；当模拟单元以运动对象在栅格地图中的位置为初始位置，并基于运动属性每运动预设步长时，更新模拟单元当前所处位置对应的栅格的访问频率之后，本发明实施例方法还包括：计算模拟单元的运动里程；基于运动里程计算运动代价。在本实施例中，如图7所示，假设模拟单元从位置a处运动至位置b再运动至 位置c处，则模拟单元的运动里程是指位置a与位置b之间的距离加上位置b与位置c之间的距离，模拟单元的运动里程为L1+L2，其中，L1是指位置a与位置b之间的路径长度，L2是指位置b与位置c之间的路径长度。
可选的，当运动对象的运动满足预设条件时，更新模拟单元的状态信息，包括：当模拟单元的运动里程达到预设的运动里程，且经过的栅格均不存在障碍物时，将模拟单元分裂为多个子模拟单元，并使多个子模拟单元沿随机运动方向继续运动。例如，若某一个模拟单元的运动里程累积达到了预设的运动里程，并且在此之前，该模拟单元经过了N个栅格，且这N个栅格均不存在障碍物或者该模拟单元在运动过程中一直未发生碰撞时，则认为这N个栅格对应的区域为空旷区域，通过将该模拟单元分裂为多个子模拟单元，并使多个子模拟单元沿随机运动方向继续运动，可以加快空旷区域的搜索效率，提高空旷区域的覆盖度。
可选的，当运动对象的运动满足预设条件时，更新模拟单元的状态信息，包括：计算模拟单元的累计通行代价；当模拟单元的累计通行代价超出初始能量时，模拟单元停止运动。
可选的，计算模拟单元的累计通行代价，包括：当模拟单元基于运动属性每运动预设步长，计算模拟单元的通行代价；从当前次运动向前，将所有次运动的通行代价累加，得到累计通行代价。其中，通行代价包括碰撞代价、爬坡代价和运动代价中的至少一种。
在本实施例中，若通行代价包括碰撞代价、爬坡代价和运动代价，则累计通行代价是指累计碰撞代价、累计爬坡代价和累计运动代价的累加值，累计碰撞代价、累计爬坡代价和累计运动代价分别记为P M、G M、SP M；假设模拟单元迭代运动了M次，则M次的累计通行代价为GT M＝P M+G M+SP M；若M次的累计通行代价GT M超出初始能量时，则认为该模拟单元的初始能量已消耗完，不再参与后续的迭代运动。
本发明实施例通过采用粒子随机运动来探索栅格地图中的区域，具有随机运动方向和速度的粒子从车辆所在位置出发，栅格地图中的地形特征(障碍物碰撞，坡度，粒子驶过距离)会对粒子作用一定的通行代价，每 个粒子具有一定的初始能量抵抗外部的通行代价。
对于单个粒子而言，若到达某一区域的通行代价高，就有可能在到达该区域之前耗尽设定初始能量，而不能到达，在访问频率信息上来看，可以到达该区域的粒子数量就较少，反之对于通行代价较低的区域，则有更多的粒子可以到达，最终，用粒子对某个区域的通行频率来计算该区域的可通行性。
可选的，根据访问频率信息，确定运动对象周围环境的可通行性，包括：在访问频率信息中，标记栅格的可通行性和/或可通行概率。
可选的，在访问频率信息中，标记栅格的可通行性和/或可通行概率，包括：将访问频率信息中访问频率大于第一预设访问频率的栅格，标记为能够通行，和/或标记通行概率大于第一预设概率阈值；将访问频率信息中访问频率小于第二预设访问频率的栅格，标记为不可通行，和/或通行概率小于第二预设概率阈值；将访问频率信息中访问频率大于或等于第二预设访问频率且小于或等于第一预设访问频率的栅格，标记为可通行和/或计算得到的通行概率，其中，第一预设概率阈值可以设置为小于100％，例如大于或等于50％且小于100％；第二预设阈值可以设置为大于0，例如大于0且小于10％，且第一预设概率阈值大于第二预设概率阈值。若通行概率大于第一预设概率阈值，则认为该栅格可通行，可通行的通行概率最大可达到100％；若通行概率小于第二预设概率阈值，则认为该栅格不可通行，不可通行的通行概率最低可达到0。可选的，还可以将通行概率大于或等于第二预设概率阈值且小于或等于第一预设概率阈值的栅格标记为可能能够通行。对于可能能够通行的栅格可以给出具体的通行概率。
可选的，通行概率是根据栅格的访问频率、第一预设访问频率以及第二预设访问频率计算得到。具体的，可以根据如下公式计算得到：
Figure PCTCN2019098586-appb-000001
式(1)中，P probfree为通行概率；f为更新结束后，栅格的访问频率；TH non-free为第二预设访问频率，TH free为第一预设访问频率。
当然，本发明实施例还可以设置一个概率阈值，将大于或等于概率阈值的栅格标记为可通行，小于概率阈值的栅格标记为不可通行。
图8是本发明实施例提供的另一种可通行性的识别方法的流程图。如图8所示，在记录多个模拟单元模拟运动对象运动的过程中对预先生成的栅格地图中每个栅格的访问频率，得到访问频率信息之前，本发明实施例的方法还包括如下步骤：
步骤S801、获取2.5D栅格地图。
如图1所示，可以根据车辆上搭载的探测设备采集的二维图像和三维点云融合得到描述自动驾驶车辆周围环境的2.5D栅格地图，该2.5D栅格地图的长度和宽度方向分别以等分辨率的栅格来描述水平面，栅格内的高度为栅格对应位置的最大高度值。栅格对应位置是指以自动驾驶车辆为原点，对应于现实位置的栅格，记栅格位置坐标为(P x,P y)，
Figure PCTCN2019098586-appb-000002
式(2)中，x res为2.5D栅格地图在长度方向上的分辨率；x为车辆周围环境信息在车辆坐标系下x轴上的位置坐标，单位为物理长度，可以是以米为单位；
Figure PCTCN2019098586-appb-000003
表示对
Figure PCTCN2019098586-appb-000004
取整；
Figure PCTCN2019098586-appb-000005
式(3)中，y res为2.5D栅格地图在宽度方向上的分辨率；y为车辆周围环境信息在车辆坐标系下y轴上的位置坐标，单位为物理长度，可以是以米为单位；
Figure PCTCN2019098586-appb-000006
表示对
Figure PCTCN2019098586-appb-000007
取整。
步骤S802、对2.5D栅格地图中的每个栅格，计算该栅格及对应的邻域栅格的栅格高度。
具体的，可以是对2.5D栅格地图中的每个栅格，计算该栅格及对应的4个邻域栅格的栅格高度，或者8个邻域栅格的栅格高度。栅格高度是指栅格的对地高度，即相对于地面的高度。
步骤S803、基于栅格及对应的邻域栅格的栅格高度，确定栅格的特征信息，其中，特征信息至少包括最大栅格高度、最小栅格高度、栅格高度平均值以及局部坡度。
具体的，是在4个邻域栅格的栅格高度，或者8个邻域栅格的栅格高 度中确定最大栅格高度、最小栅格高度、栅格高度平均值以及局部坡度，并将其绑定至该栅格，作为该栅格的特征信息。
步骤S804、基于特征信息判断栅格是否为可通行栅格，得到栅格地图。
具体的，可以通过判断最大栅格高度与最小栅格高度的高度差值是否小于可通行阈值，局部高度值的方差是否小于可通行阈值，局部坡度是否小于爬坡能力中至少一项，来判断该栅格是否为可通行栅格，从而得到上述实施例的栅格地图，在此栅格地图的基础上，还可以进一步得到访问频率信息。
下面通过实验数据的比对，来对本发明实施例的有益效果进行说明：
图9为现有技术中可通行性识别结果图。
图10为采用本发明实施例的方法得到的访问频率信息，其中，该访问频率信息为访问频率图的形式。
图11为采用本发明实施例的方法得到的可通行性识别结果图。
图9、图10和图11是对同一环境进行可通行性识别。如图9所示，图中灰色区域代表识别出来的可通行区域，可以看到，图中左边方框内长条形状的窄带区域R，其虽然被识别为可通行区域，然而由于自动驾驶车辆的宽度大于该窄带区域，因此，自动驾驶车辆仍然无法通过，故此部分区域R实际上属于不可通行区域。图中右边方框内较宽的区域W，虽然宽度大于车辆宽度，但是W下方的区域较窄，因此，车辆实际上也无法到达W区域，W区域实际上也属于不可通行区域。
如图10、图11所示，图中颜色较浅区域为识别出来的可通行区域，颜色越浅，代表可通行性越高，反之，颜色越深，代表可通行性越低。从图10、图11中可以看到，图9中的R区域和W区域对应的栅格，模拟单元不会访问，因此，本发明实施例的方法能够提高可通行性的识别精度。
本发明实施例提供一种可通行性的识别系统。图12为本发明实施例提供的可通行性的识别系统的结构图，如图12所示，可通行性的识别系统120包括：存储器121、处理器122和探测设备123；所述探测设备用于探测可移动平台周围环境信息得到二维图像和三维点云；所述存储器121用于存储程序代码；所述处理器122，调用所述程序代码，当程序代 码被执行时，用于执行以下操作：基于所述二维图像和所述三维点云进行融合，得到栅格地图；设置多个模拟单元，每个所述模拟单元能够模拟运动对象的运动；记录多个所述模拟单元模拟所述运动对象运动的过程中对预先生成的栅格地图中每个栅格的访问频率，得到访问频率信息；其中，所述栅格地图用于描述所述运动对象周围环境的信息；根据所述访问频率信息，确定所述运动对象周围环境的可通行性。
可选的，所述处理器122在设置多个模拟单元时，具体用于：分别设置多个所述模拟单元具有所述运动对象的属性，所述运动对象的属性包括固有属性和运动属性。
可选的，固有属性至少包括尺寸属性；所述处理器122在分别设置多个所述模拟单元具有所述运动对象的属性时，至少用于：分别设置多个所述模拟单元具有所述尺寸属性。
可选的，尺寸属性是指所述模拟单元的尺寸不小于所述运动对象的尺寸；所述处理器122在分别设置多个所述模拟单元具有所述运动对象的属性时，具体用于：分别设置多个所述模拟单元的尺寸为不小于所述运动对象的尺寸。
可选的，所述运动属性包括初始运动速率、初始运动方向、初始运动能量、爬坡能力和通行代价中的至少一项；所述处理器122在分别设置多个所述模拟单元具有所述运动对象的属性时，具体用于：分别设置所述多个所述模拟单元具有所述初始运动速率、所述初始运动方向、所述初始运动能量、所述爬坡能力和所述通行代价中的至少一项。
可选的，在所述模拟单元模拟所述运动对象的运动之前，所述预先生成的栅格地图中每个所述栅格的所述访问频率为0；所述处理器122在记录多个所述模拟单元模拟所述运动对象运动的过程中对预先生成的栅格地图中每个栅格的访问频率，得到访问频率信息时，具体用于：当所述模拟单元以所述运动对象在所述栅格地图中的位置为初始位置，并基于所述运动属性每运动预设步长时，更新所述模拟单元当前所处位置对应的栅格的访问频率；当更新次数达到预设的最大迭代次数，得到所述访问频率信息。
可选的，所述处理器122在当所述模拟单元以所述运动对象在所述栅 格地图中的位置为初始位置，并基于所述运动属性每运动预设步长时，更新所述模拟单元当前所处位置对应的栅格的访问频率时，具体用于：当所述模拟单元以所述运动对象在所述栅格地图中的位置为初始位置，并基于所述运动属性每运动预设步长时，在所述预先生成的栅格地图上标记所述模拟单元当前所处位置对应的所述栅格，并将所述模拟单元当前所处位置对应的所述栅格的所述访问频率加1。
可选的，所述模拟单元的尺寸不小于单个所述栅格的尺寸；所述模拟单元当前所处位置对应的所述栅格是指所述模拟单元在所述当前所处位置处覆盖的栅格。
可选的，所述处理器122在当所述模拟单元以所述运动对象在所述栅格地图中的位置为初始位置，并基于所述运动属性每运动预设步长时，更新所述模拟单元当前所处位置对应的栅格的访问频率之后，还用于：当所述运动对象的运动满足预设条件时，更新所述模拟单元的状态信息，所述状态信息由所述通行代价表示。
可选的，所述通行代价包括碰撞代价；所述处理器122在当所述运动对象的运动满足预设条件时，更新所述模拟单元的状态信息时，具体用于：当所述模拟单元当前所处位置对应的栅格存在障碍物栅格或者所述模拟单元当前所处位置的位置坐标与所述栅格地图的边缘的位置坐标相同时，更新所述模拟单元的运动方向为反弹方向，并计算所述模拟单元的所述碰撞代价。
可选的，所述通行代价包括爬坡代价；所述处理器122在当所述运动对象的运动满足预设条件时，更新所述模拟单元的状态信息时，具体用于：当所述模拟单元当前所处位置对应的栅格的局部坡度在所述模拟单元的运动方向上的投影超过预设的最大坡度时，则更新所述模拟单元的运动方向为反弹方向，并计算所述模拟单元的碰撞代价；当所述模拟单元当前所处位置对应的栅格的局部坡度在所述模拟单元的运动方向上的投影未超过预设的最大坡度时，则计算所述模拟单元的爬坡代价。
可选的，所述处理器122在计算所述模拟单元的碰撞代价时，具体用于：在所述模拟单元的当前碰撞代价的基础上增加预设的单次碰撞代价。
可选的，所述爬坡代价与所述坡度成比例。
可选的，所述通行代价包括运动代价；所述处理器122在当所述模拟单元以所述运动对象在所述栅格地图中的位置为初始位置，并基于所述运动属性每运动预设步长时，更新所述模拟单元当前所处位置对应的栅格的访问频率之后，还用于：计算所述模拟单元的运动里程；基于所述运动里程计算运动代价。
可选的，所述处理器122在当所述运动对象的运动满足预设条件时，更新所述模拟单元的状态信息时，具体用于：当所述模拟单元的运动里程达到预设的运动里程，且标记的栅格均不存在障碍物栅格时，将所述模拟单元分裂为多个子模拟单元，并使多个所述子模拟单元沿随机运动方向继续运动。
可选的，所述处理器122在当所述运动对象的运动满足预设条件时，更新所述模拟单元的状态信息时，具体用于：计算所述模拟单元的累计通行代价；当所述模拟单元的累计通行代价超出所述初始能量时，所述模拟单元停止运动。
可选的，所述处理器122在计算所述模拟单元的累计通行代价时，具体用于：当所述模拟单元基于所述运动属性每运动预设步长，计算所述模拟单元的所述通行代价；从当前次运动向前，将所有次运动的所述通行代价累加，得到所述累计通行代价。
可选的，所述通行代价包括碰撞代价、爬坡代价和运动代价中的至少一种。
可选的，所述处理器122在根据所述访问频率信息，确定所述运动对象周围环境的可通行性时，具体用于：在所述访问频率信息中，标记所述栅格的可通行性和/或可通行概率。
可选的，所述处理器122在所述访问频率信息中，标记所述栅格的可通行性和/或可通行概率时，具体用于：将所述访问频率信息中访问频率大于第一预设访问频率的栅格，标记为能够通行，和/或标记通行概率大于第一预设概率阈值；将所述访问频率信息中访问频率小于第二预设访问频率的栅格，标记为不可通行，和/或通行概率小于第二预设概率阈值；将所述访问频率信息中访问频率大于或等于所述第二预设访问频率且小于或等于所述第一预设访问频率的栅格，标记为可通行和/或计算得到的 通行概率，其中，所述第一预设概率阈值大于所述第二预设概率阈值。
可选的，所述通行概率是根据所述栅格的访问频率、所述第一预设访问频率以及所述第二预设访问频率计算得到。
可选的，所述运动对象为可移动机器人或车辆，具体的，车辆可以是无人驾驶车辆或搭载有ADAS系统的车辆等。
可选的，所述模拟单元为具有随机游走性能的粒子。
可选的，所述处理器122在记录多个所述模拟单元模拟所述运动对象运动的过程中对预先生成的栅格地图中每个栅格的访问频率，得到访问频率信息之前，还用于：获取2.5D栅格地图；对所述2.5D栅格地图中的每个栅格，计算所述栅格及对应的邻域栅格的栅格高度；基于所述栅格及对应的邻域栅格的栅格高度，确定所述栅格的特征信息，所述特征信息至少包括最大栅格高度、最小栅格高度、栅格高度平均值以及局部坡度；基于所述特征信息判断所述栅格是否为可通行栅格，得到所述预先生成的栅格地图。
本发明实施例提供的可通行性的识别系统的具体原理和实现方式均与上述实施例类似，此处不再赘述。
本实施例通过设置多个模拟单元，每个模拟单元能够模拟运动对象的运动；记录多个模拟单元在模拟运动对象运动的过程中对预先生成的栅格地图中每个栅格的访问频率，得到访问频率信息；其中，栅格地图用于描述运动对象周围环境的信息；根据访问频率信息，确定运动对象周围环境的可通行性。由于设置的模拟单元能够模拟运动对象在地面上的运动过程，且模拟单元的模拟运动过程采用访问频率信息来表示，因此，可以直观地了解车辆在周围环境中的可通行性，从而为自动驾驶路径规划决策提供依据。
本发明实施例提供一种可移动平台。该可移动平台包括：机身、动力系统和上述实施例所述的可通行性的识别系统；其中，动力系统安装在所述机身，用于提供移动动力。可通行性的识别系统的具体实现方式和原理与上述实施例均一致，此处不再赘述。可选的，所述可移动平台包括：可 移动机器人或车辆。当可移动平台为车辆时，所述机身可以包括车辆的车身、底盘等承载部分。当可通行性的识别系统为一个单独的设备时，其可以以前装或后装的形式集成于所述可移动平台例如车辆上；当所述可通行性识别系统为分布式的系统时，其各个部分如探测设备、处理器和存储器等可各自安装在可移动平台的相同或不同的位置。
另外，本实施例还提供一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行以实现上述实施例所述的可通行性的识别方法。
在本发明所提供的几个实施例中，应该理解到，所揭露的装置和方法，可以通过其它的方式实现。例如，以上所描述的装置实施例仅仅是示意性的，例如，所述单元的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，例如多个单元或组件可以结合或者可以集成到另一个系统，或一些特征可以忽略，或不执行。另一点，所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口，装置或单元的间接耦合或通信连接，可以是电性，机械或其它的形式。
所述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
另外，在本发明各个实施例中的各功能单元可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现，也可以采用硬件加软件功能单元的形式实现。
上述以软件功能单元的形式实现的集成的单元，可以存储在一个计算机可读取存储介质中。上述软件功能单元存储在一个存储介质中，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)或处理器(processor)执行本发明各个实施例所述方法的部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(Read-Only Memory，ROM)、随机存取存储器(Random Access Memory，RAM)、磁碟 或者光盘等各种可以存储程序代码的介质。
本领域技术人员可以清楚地了解到，为描述的方便和简洁，仅以上述各功能模块的划分进行举例说明，实际应用中，可以根据需要而将上述功能分配由不同的功能模块完成，即将装置的内部结构划分成不同的功能模块，以完成以上描述的全部或者部分功能。上述描述的装置的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。
最后应说明的是：以上各实施例仅用以说明本发明的技术方案，而非对其限制；尽管参照前述各实施例对本发明进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分或者全部技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本发明各实施例技术方案的范围。

Claims (53)

  1. 一种可通行性的识别方法，其特征在于，包括：
    设置多个模拟单元，每个所述模拟单元能够模拟运动对象的运动；
    记录多个所述模拟单元在模拟所述运动对象运动的过程中对预先生成的栅格地图中每个栅格的访问频率，得到访问频率信息；其中，所述栅格地图用于描述所述运动对象周围环境的信息；
    根据所述访问频率信息，确定所述运动对象周围环境的可通行性。
  2. 根据权利要求1所述的方法，其特征在于，所述设置多个模拟单元，包括：
    分别设置多个所述模拟单元具有所述运动对象的属性，所述运动对象的属性包括固有属性和运动属性。
  3. 根据权利要求2所述的方法，其特征在于，所述固有属性至少包括尺寸属性；
    所述分别设置多个所述模拟单元具有所述运动对象的属性，至少包括：
    分别设置多个所述模拟单元具有所述尺寸属性。
  4. 根据权利要求3所述的方法，其特征在于，所述尺寸属性是指所述模拟单元的尺寸不小于所述运动对象的尺寸；
    所述分别设置多个所述模拟单元具有所述运动对象的属性，包括：
    分别设置多个所述模拟单元的尺寸为不小于所述运动对象的尺寸。
  5. 根据权利要求2所述的方法，其特征在于，所述运动属性包括初始运动速率、初始运动方向、初始运动能量、爬坡能力和通行代价中的至少一项；
    所述分别设置多个所述模拟单元具有所述运动对象的属性，包括：
    分别设置多个所述模拟单元具有所述初始运动速率、所述初始运动方向、所述初始运动能量、所述爬坡能力和所述通行代价中的至少一项。
  6. 根据权利要求5所述的方法，其特征在于，在多个所述模拟单元模拟所述运动对象的运动之前，所述预先生成的栅格地图中每个所述栅格的所述访问频率为0；
    所述记录多个所述模拟单元模拟所述运动对象运动的过程中对预先生成的栅格地图中每个栅格的访问频率，得到访问频率信息，包括：
    当所述模拟单元以所述运动对象在所述栅格地图中的位置为初始位置，并基于所述运动属性每运动预设步长时，更新所述模拟单元当前所处位置对应的栅格的所述访问频率；
    当更新次数达到预设的最大迭代次数，得到所述访问频率信息。
  7. 根据权利要求6所述的方法，其特征在于，所述当所述模拟单元以所述运动对象在所述栅格地图中的位置为初始位置，并基于所述运动属性每运动预设步长时，更新所述模拟单元当前所处位置对应的栅格的访问频率，包括：
    当所述模拟单元以所述运动对象在所述栅格地图中的位置为初始位置，并基于所述运动属性每运动预设步长时，在所述预先生成的栅格地图上标记所述模拟单元当前所处位置对应的所述栅格，并将所述模拟单元当前所处位置对应的所述栅格的所述访问频率加1。
  8. 根据权利要求7所述的方法，其特征在于，所述模拟单元的尺寸不小于单个所述栅格的尺寸；
    所述模拟单元当前所处位置对应的所述栅格是指所述模拟单元在所述当前所处位置处覆盖的栅格。
  9. 根据权利要求7所述的方法，其特征在于，所述当所述模拟单元以所述运动对象在所述栅格地图中的位置为初始位置，并基于所述运动属性每运动预设步长时，更新所述模拟单元当前所处位置对应的栅格的访问频率之后，所述方法还包括：
    当所述运动对象的运动满足预设条件时，更新所述模拟单元的状态信息，所述状态信息由所述通行代价表示。
  10. 根据权利要求9所述的方法，其特征在于，所述通行代价包括碰撞代价；
    所述当所述运动对象的运动满足预设条件时，更新所述模拟单元的状态信息，包括：
    当所述模拟单元当前所处位置对应的栅格存在障碍物栅格或者所述模拟单元当前所处位置的位置坐标与所述栅格地图的边缘的位置坐标相同时，更新所述模拟单元的运动方向为反弹方向，并计算所述模拟单元的所述碰撞代价。
  11. 根据权利要求9所述的方法，其特征在于，所述通行代价包括爬坡代价；
    所述当所述运动对象的运动满足预设条件时，更新所述模拟单元的状态信息，包括：
    当所述模拟单元当前所处位置对应的栅格的局部坡度在所述模拟单元的运动方向上的投影超过预设的最大坡度时，则更新所述模拟单元的运动方向为反弹方向，并计算所述模拟单元的碰撞代价；
    当所述模拟单元当前所处位置对应的栅格的局部坡度在所述模拟单元的运动方向上的投影未超过预设的最大坡度时，则计算所述模拟单元的爬坡代价。
  12. 根据权利要求10或11所述的方法，其特征在于，所述计算所述模拟单元的碰撞代价，包括：
    在所述模拟单元的当前碰撞代价的基础上增加预设的单次碰撞代价。
  13. 根据权利要求11所述的方法，其特征在于，所述爬坡代价与所述局部坡度成比例。
  14. 根据权利要求9所述的方法，其特征在于，所述通行代价包括运动代价；
    所述当所述模拟单元以所述运动对象在所述栅格地图中的位置为初始位置，并基于所述运动属性每运动预设步长时，更新所述模拟单元当前所处位置对应的栅格的访问频率之后，所述方法还包括：
    计算所述模拟单元的运动里程；
    基于所述运动里程计算运动代价。
  15. 根据权利要求14所述的方法，其特征在于，所述当所述运动对象的运动满足预设条件时，更新所述模拟单元的状态信息，包括：
    当所述模拟单元的运动里程达到预设的运动里程，且经过的栅格均不存在障碍物栅格时，将所述模拟单元分裂为多个子模拟单元，并使多个所述子模拟单元沿随机运动方向继续运动。
  16. 根据权利要求9所述的方法，其特征在于，所述当所述运动对象的运动满足预设条件时，更新所述模拟单元的状态信息，包括：
    计算所述模拟单元的累计通行代价；
    当所述模拟单元的累计通行代价超出所述初始能量时，所述模拟单元停止运动。
  17. 根据权利要求16所述的方法，其特征在于，所述计算所述模拟单元的累计通行代价，包括：
    当所述模拟单元基于所述运动属性每运动预设步长，计算所述模拟单元的所述通行代价；
    从当前次运动向前，将所有次运动的所述通行代价累加，得到所述累计通行代价。
  18. 根据权利要求5或9所述的方法，其特征在于，所述通行代价包括碰撞代价、爬坡代价和运动代价中的至少一种。
  19. 根据权利要求1所述的方法，其特征在于，所述根据所述访问频率信息，确定所述运动对象周围环境的可通行性，包括：
    根据所述访问频率信息，标记所述栅格的可通行性和/或可通行概率。
  20. 根据权利要求19所述的方法，其特征在于，所述根据所述访问频率信息，标记所述栅格的可通行性和/或可通行概率，包括：
    将所述访问频率信息中访问频率大于第一预设访问频率的栅格，标记为能够通行，和/或标记通行概率大于第一预设概率阈值；
    将所述访问频率信息中访问频率小于第二预设访问频率的栅格，标记为不可通行，和/或通行概率小于第二预设概率阈值；
    将所述访问频率信息中访问频率大于或等于所述第二预设访问频率且小于或等于所述第一预设访问频率的栅格，标记为可通行和/或计算得到的通行概率，其中，所述第一预设概率阈值大于所述第二预设概率阈值。
  21. 根据权利要求20所述的方法，其特征在于，所述通行概率是根据所述栅格的访问频率、所述第一预设访问频率以及所述第二预设访问频率计算得到。
  22. 根据权利要求1-5任一项所述的方法，其特征在于，所述运动对象为可移动机器人或车辆。
  23. 根据权利要求1-5任一项所述的方法，其特征在于，所述模拟单元为具有随机游走性能的粒子。
  24. 根据权利要求1-5任一项所述的方法，其特征在于，所述记录多 个所述模拟单元模拟所述运动对象运动的过程中对预先生成的栅格地图中每个栅格的访问频率，得到访问频率信息之前，所述方法还包括：
    获取2.5D栅格地图；
    对所述2.5D栅格地图中的每个栅格，计算所述栅格及对应的邻域栅格的栅格高度；
    基于所述栅格及对应的邻域栅格的栅格高度，确定所述栅格的特征信息，所述特征信息至少包括最大栅格高度、最小栅格高度、栅格高度平均值以及局部坡度；
    基于所述特征信息判断所述栅格是否为可通行栅格，得到栅格地图。
  25. 根据权利要求1-5任一项所述的方法，其特征在于，所述访问频率信息包括访问频率图。
  26. 一种可通行性的识别系统，其特征在于，包括：探测设备、存储器和处理器；
    所述探测设备用于探测可移动平台周围环境信息得到二维图像和三维点云；
    所述存储器用于存储程序代码；
    所述处理器，调用所述程序代码，当程序代码被执行时，用于执行以下操作：
    设置多个模拟单元，每个所述模拟单元能够模拟运动对象的运动；
    记录多个所述模拟单元模拟所述运动对象运动的过程中对预先生成的栅格地图中每个栅格的访问频率，得到访问频率信息；其中，所述栅格地图用于描述所述运动对象周围环境的信息；所述预先生成的栅格地图是基于所述二维图像和所述三维点云进行融合得到；
    根据所述访问频率信息，确定所述运动对象周围环境的可通行性。
  27. 根据权利要求26所述的系统，其特征在于，所述处理器在设置多个模拟单元时，具体用于：
    分别设置多个所述模拟单元具有所述运动对象的属性，所述运动对象的属性包括固有属性和运动属性。
  28. 根据权利要求27所述的系统，其特征在于，所述固有属性至少包括尺寸属性；
    所述处理器在分别设置多个所述模拟单元具有所述运动对象的属性时，至少用于：
    分别设置多个所述模拟单元具有所述尺寸属性。
  29. 根据权利要求28所述的系统，其特征在于，所述尺寸属性是指所述模拟单元的尺寸不小于所述运动对象的尺寸；
    所述处理器在分别设置多个所述模拟单元具有所述运动对象的属性时，具体用于：
    分别设置多个所述模拟单元的尺寸为不小于所述运动对象的尺寸。
  30. 根据权利要求27所述的系统，其特征在于，所述运动属性包括初始运动速率、初始运动方向、初始运动能量、爬坡能力和通行代价中的至少一项；
    所述处理器在分别设置多个所述模拟单元具有所述运动对象的属性时，具体用于：
    分别设置多个所述模拟单元具有所述初始运动速率、所述初始运动方向、所述初始运动能量、所述爬坡能力和所述通行代价中的至少一项。
  31. 根据权利要求30所述的系统，其特征在于，在所述模拟单元模拟所述运动对象的运动之前，所述预先生成的栅格地图中每个所述栅格的所述访问频率为0；
    所述处理器在记录多个所述模拟单元模拟所述运动对象运动的过程中对预先生成的栅格地图中每个栅格的访问频率，得到访问频率信息时，具体用于：
    当所述模拟单元以所述运动对象在所述栅格地图中的位置为初始位置，并基于所述运动属性每运动预设步长时，更新所述模拟单元当前所处位置对应的栅格的访问频率；
    当更新次数达到预设的最大迭代次数，得到所述访问频率信息。
  32. 根据权利要求31所述的系统，其特征在于，所述处理器在当所述模拟单元以所述运动对象在所述栅格地图中的位置为初始位置，并基于所述运动属性每运动预设步长时，更新所述模拟单元当前所处位置对应的栅格的访问频率时，具体用于：
    当所述模拟单元以所述运动对象在所述栅格地图中的位置为初始位 置，并基于所述运动属性每运动预设步长时，在所述预先生成的栅格地图上标记所述模拟单元当前所处位置对应的所述栅格，并将所述模拟单元当前所处位置对应的所述栅格的所述访问频率加1。
  33. 根据权利要求32所述的系统，其特征在于，所述模拟单元的尺寸不小于单个所述栅格的尺寸；
    所述模拟单元当前所处位置对应的所述栅格是指所述模拟单元在所述当前所处位置处覆盖的栅格。
  34. 根据权利要求32所述的系统，其特征在于，所述处理器在当所述模拟单元以所述运动对象在所述栅格地图中的位置为初始位置，并基于所述运动属性每运动预设步长时，更新所述模拟单元当前所处位置对应的栅格的访问频率之后，还用于：
    当所述运动对象的运动满足预设条件时，更新所述模拟单元的状态信息，所述状态信息由所述通行代价表示。
  35. 根据权利要求34所述的系统，其特征在于，所述通行代价包括碰撞代价；
    所述处理器在当所述运动对象的运动满足预设条件时，更新所述模拟单元的状态信息时，具体用于：
    当所述模拟单元当前所处位置对应的栅格存在障碍物栅格或者所述模拟单元当前所处位置的位置坐标与所述栅格地图的边缘的位置坐标相同时，更新所述模拟单元的运动方向为反弹方向，并计算所述模拟单元的所述碰撞代价。
  36. 根据权利要求34所述的系统，其特征在于，所述通行代价包括爬坡代价；
    所述处理器在当所述运动对象的运动满足预设条件时，更新所述模拟单元的状态信息时，具体用于：
    当所述模拟单元当前所处位置对应的栅格的局部坡度在所述模拟单元的运动方向上的投影超过预设的最大坡度时，则更新所述模拟单元的运动方向为反弹方向，并计算所述模拟单元的碰撞代价；
    当所述模拟单元当前所处位置对应的栅格的局部坡度在所述模拟单元的运动方向上的投影未超过所述预设的最大坡度时，则计算所述模拟单 元的爬坡代价。
  37. 根据权利要求35或36所述的系统，其特征在于，所述处理器在计算所述模拟单元的碰撞代价时，具体用于：
    在所述模拟单元的当前碰撞代价的基础上增加预设的单次碰撞代价。
  38. 根据权利要求36所述的系统，其特征在于，所述爬坡代价与所述局部坡度成比例。
  39. 根据权利要求34所述的系统，其特征在于，所述通行代价包括运动代价；
    所述处理器在当所述模拟单元以所述运动对象在所述栅格地图中的位置为初始位置，并基于所述运动属性每运动预设步长时，更新所述模拟单元当前所处位置对应的栅格的访问频率之后，还用于：
    计算所述模拟单元的运动里程；
    基于所述运动里程计算运动代价。
  40. 根据权利要求39所述的系统，其特征在于，所述处理器在当所述运动对象的运动满足预设条件时，更新所述模拟单元的状态信息时，具体用于：
    当所述模拟单元的运动里程达到预设的运动里程，且标记的栅格均不存在障碍物栅格时，将所述模拟单元分裂为多个子模拟单元，并使多个所述子模拟单元沿随机运动方向继续运动。
  41. 根据权利要求34所述的系统，其特征在于，所述处理器在当所述运动对象的运动满足预设条件时，更新所述模拟单元的状态信息时，具体用于：
    计算所述模拟单元的累计通行代价；
    当所述模拟单元的累计通行代价超出所述初始能量时，所述模拟单元停止运动。
  42. 根据权利要求41所述的系统，其特征在于，所述处理器在计算所述模拟单元的累计通行代价时，具体用于：
    当所述模拟单元基于所述运动属性每运动预设步长，计算所述模拟单元的所述通行代价；
    从当前次运动向前，将所有次运动的所述通行代价累加，得到所述累 计通行代价。
  43. 根据权利要求30或34所述的系统，其特征在于，所述通行代价包括碰撞代价、爬坡代价和运动代价中的至少一种。
  44. 根据权利要求26所述的系统，其特征在于，所述处理器在根据所述访问频率信息，确定所述运动对象周围环境的可通行性时，具体用于：
    根据所述访问频率信息，标记所述栅格的可通行性和/或可通行概率。
  45. 根据权利要求44所述的系统，其特征在于，所述处理器根据所述访问频率信息，标记所述栅格的可通行性和/或可通行概率时，具体用于：
    将所述访问频率信息中访问频率大于第一预设访问频率的栅格，标记为能够通行，和/或标记通行概率大于第一预设概率阈值；
    将所述访问频率信息中访问频率小于第二预设访问频率的栅格，标记为不可通行，和/或通行概率小于第二预设概率阈值；
    将所述访问频率信息中访问频率大于或等于所述第二预设访问频率且小于或等于所述第一预设访问频率的栅格，标记为可通行和/或计算得到的通行概率，其中，所述第一预设概率阈值大于所述第二预设概率阈值。
  46. 根据权利要求45所述的系统，其特征在于，所述通行概率是根据所述栅格的访问频率、所述第一预设访问频率以及所述第二预设访问频率计算得到。
  47. 根据权利要求26-30任一项所述的系统，其特征在于，所述运动对象为可移动机器人或车辆。
  48. 根据权利要求26-30任一项所述的系统，其特征在于，所述模拟单元为具有随机游走性能的粒子。
  49. 根据权利要求26-30任一项所述的系统，其特征在于，所述处理器在记录多个所述模拟单元模拟所述运动对象运动的过程中对预先生成的栅格地图中每个栅格的访问频率，得到访问频率信息之前，还用于：
    获取2.5D栅格地图；
    对所述2.5D栅格地图中的每个栅格，计算所述栅格及对应的邻域栅格的栅格高度；
    基于所述栅格及对应的邻域栅格的栅格高度，确定所述栅格的特征信 息，所述特征信息至少包括最大栅格高度、最小栅格高度、栅格高度平均值以及局部坡度；
    基于所述特征信息判断所述栅格是否为可通行栅格，得到所述预先生成的栅格地图。
  50. 根据权利要求26-30任一项所述的系统，其特征在于，所述访问频率信息包括访问频率图。
  51. 一种可移动平台，其特征在于，包括：
    机身；
    动力系统，安装在所述机身，用于提供移动动力；
    以及权利要求26-50任一项所述的系统。
  52. 根据权利要求51所述的可移动平台，其特征在于，所述可移动平台包括：可移动机器人或车辆。
  53. 一种计算机可读存储介质，其特征在于，其上存储有计算机程序，所述计算机程序被处理器执行以实现如权利要求1-25任一项所述的方法。
PCT/CN2019/098586 2019-07-31 2019-07-31 可通行性的识别方法、系统、设备及计算机可读存储介质 WO2021016920A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
PCT/CN2019/098586 WO2021016920A1 (zh) 2019-07-31 2019-07-31 可通行性的识别方法、系统、设备及计算机可读存储介质
CN201980034105.XA CN112166446A (zh) 2019-07-31 2019-07-31 可通行性的识别方法、系统、设备及计算机可读存储介质

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2019/098586 WO2021016920A1 (zh) 2019-07-31 2019-07-31 可通行性的识别方法、系统、设备及计算机可读存储介质

Publications (1)

Publication Number Publication Date
WO2021016920A1 true WO2021016920A1 (zh) 2021-02-04

Family

ID=73859803

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/098586 WO2021016920A1 (zh) 2019-07-31 2019-07-31 可通行性的识别方法、系统、设备及计算机可读存储介质

Country Status (2)

Country Link
CN (1) CN112166446A (zh)
WO (1) WO2021016920A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113110424A (zh) * 2021-03-26 2021-07-13 大连海事大学 一种基于海图信息的无人艇避碰方法
CN113222025A (zh) * 2021-05-18 2021-08-06 浙江大学 一种基于激光雷达的可行区域标签生成方法
CN115509216A (zh) * 2021-06-21 2022-12-23 广州视源电子科技股份有限公司 路径规划方法、装置、计算机设备和存储介质
CN117762150A (zh) * 2024-02-22 2024-03-26 北京航空航天大学 多智能体分布式区域覆盖控制方法、智能体和程序产品

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050182518A1 (en) * 2004-02-13 2005-08-18 Evolution Robotics, Inc. Robust sensor fusion for mapping and localization in a simultaneous localization and mapping (SLAM) system
US20150284010A1 (en) * 2013-09-16 2015-10-08 Disney Enterprises, Inc. Shared control of semi-autonomous vehicles including collision avoidance in multi-agent scenarios
CN105511457A (zh) * 2014-09-25 2016-04-20 科沃斯机器人有限公司 机器人静态路径规划方法
US20170132334A1 (en) * 2015-11-05 2017-05-11 Zoox, Inc. Simulation system and methods for autonomous vehicles
CN107992040A (zh) * 2017-12-04 2018-05-04 重庆邮电大学 基于地图栅格与qpso算法结合的机器人路径规划方法
CN108645411A (zh) * 2018-05-15 2018-10-12 深圳大学 基于粒子群算法的机器人路径规划方法、装置及终端设备
CN109211255A (zh) * 2017-07-06 2019-01-15 罗伯特·博世有限公司 用于为具有自动车辆系统的机动车规划路线的方法

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103472828A (zh) * 2013-09-13 2013-12-25 桂林电子科技大学 基于改进蚁群粒子群算法的移动机器人路径规划方法
CN107063280B (zh) * 2017-03-24 2019-12-31 重庆邮电大学 一种基于控制采样的智能车辆路径规划系统及方法

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050182518A1 (en) * 2004-02-13 2005-08-18 Evolution Robotics, Inc. Robust sensor fusion for mapping and localization in a simultaneous localization and mapping (SLAM) system
US20150284010A1 (en) * 2013-09-16 2015-10-08 Disney Enterprises, Inc. Shared control of semi-autonomous vehicles including collision avoidance in multi-agent scenarios
CN105511457A (zh) * 2014-09-25 2016-04-20 科沃斯机器人有限公司 机器人静态路径规划方法
US20170132334A1 (en) * 2015-11-05 2017-05-11 Zoox, Inc. Simulation system and methods for autonomous vehicles
CN109211255A (zh) * 2017-07-06 2019-01-15 罗伯特·博世有限公司 用于为具有自动车辆系统的机动车规划路线的方法
CN107992040A (zh) * 2017-12-04 2018-05-04 重庆邮电大学 基于地图栅格与qpso算法结合的机器人路径规划方法
CN108645411A (zh) * 2018-05-15 2018-10-12 深圳大学 基于粒子群算法的机器人路径规划方法、装置及终端设备

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113110424A (zh) * 2021-03-26 2021-07-13 大连海事大学 一种基于海图信息的无人艇避碰方法
CN113222025A (zh) * 2021-05-18 2021-08-06 浙江大学 一种基于激光雷达的可行区域标签生成方法
CN115509216A (zh) * 2021-06-21 2022-12-23 广州视源电子科技股份有限公司 路径规划方法、装置、计算机设备和存储介质
CN117762150A (zh) * 2024-02-22 2024-03-26 北京航空航天大学 多智能体分布式区域覆盖控制方法、智能体和程序产品
CN117762150B (zh) * 2024-02-22 2024-04-26 北京航空航天大学 多智能体分布式区域覆盖控制方法、智能体和程序产品

Also Published As

Publication number Publication date
CN112166446A (zh) 2021-01-01

Similar Documents

Publication Publication Date Title
WO2021016920A1 (zh) 可通行性的识别方法、系统、设备及计算机可读存储介质
US11726477B2 (en) Methods and systems for trajectory forecasting with recurrent neural networks using inertial behavioral rollout
US20230054914A1 (en) Vehicle localization
CN111079619B (zh) 用于检测图像中的目标对象的方法和装置
US10037613B1 (en) Systems and methods to track vehicles proximate perceived by an autonomous vehicle
Sless et al. Road scene understanding by occupancy grid learning from sparse radar clusters using semantic segmentation
WO2021046716A1 (zh) 目标对象的检测方法、系统、设备及存储介质
CN111771207A (zh) 增强的车辆跟踪
US20210389133A1 (en) Systems and methods for deriving path-prior data using collected trajectories
US20200200905A1 (en) Multi-stage object heading estimation
EP3410404A1 (en) Method and system for creating and simulating a realistic 3d virtual world
CN107808524B (zh) 一种基于无人机的道路交叉口车辆检测方法
CN117440908A (zh) 用于自动驾驶系统中基于图神经网络的行人动作预测的方法和系统
US11970185B2 (en) Data structure for storing information relating to an environment of an autonomous vehicle and methods of use thereof
CN110705385B (zh) 一种障碍物角度的检测方法、装置、设备及介质
WO2022078342A1 (zh) 动态占据栅格估计方法及装置
KR20180027242A (ko) 무인 차량을 위한 주변 환경 매핑 방법 및 장치
CN114926809A (zh) 可通行区域检测方法及装置、移动工具、存储介质
US20230056589A1 (en) Systems and methods for generating multilevel occupancy and occlusion grids for controlling navigation of vehicles
Jungnickel et al. Object tracking and dynamic estimation on evidential grids
CN114092660A (zh) 高精地图生成方法、装置及用于生成地图的车辆
US20230410423A1 (en) Three-dimensional motion grid system for autonomous vehicle perception
KR102618680B1 (ko) 영상과 라이다를 이용한 실시간 3차원 물체 인식 및 추적하는 시스템
CN117911428A (zh) 地面点云提取方法和装置、无人车和存储介质
CN115438430B (zh) 一种矿区车辆行驶稳定性预测方法和装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19939102

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19939102

Country of ref document: EP

Kind code of ref document: A1