WO2020001423A1 - 确定自动泊车策略的方法和装置 - Google Patents

确定自动泊车策略的方法和装置 Download PDF

Info

Publication number
WO2020001423A1
WO2020001423A1 PCT/CN2019/092722 CN2019092722W WO2020001423A1 WO 2020001423 A1 WO2020001423 A1 WO 2020001423A1 CN 2019092722 W CN2019092722 W CN 2019092722W WO 2020001423 A1 WO2020001423 A1 WO 2020001423A1
Authority
WO
WIPO (PCT)
Prior art keywords
parking
target
action
vehicle
included angle
Prior art date
Application number
PCT/CN2019/092722
Other languages
English (en)
French (fr)
Inventor
庄雨铮
古强
刘武龙
Original Assignee
华为技术有限公司
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 华为技术有限公司 filed Critical 华为技术有限公司
Priority to EP19826527.4A priority Critical patent/EP3805062B1/en
Priority to EP22205142.7A priority patent/EP4206051A1/en
Publication of WO2020001423A1 publication Critical patent/WO2020001423A1/zh
Priority to US17/134,858 priority patent/US11897454B2/en

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W30/00—Purposes of road vehicle drive control systems not related to the control of a particular sub-unit, e.g. of systems using conjoint control of vehicle sub-units
    • B60W30/06—Automatic manoeuvring for parking
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B62—LAND VEHICLES FOR TRAVELLING OTHERWISE THAN ON RAILS
    • B62D—MOTOR VEHICLES; TRAILERS
    • B62D15/00—Steering not otherwise provided for
    • B62D15/02—Steering position indicators ; Steering position determination; Steering aids
    • B62D15/027—Parking aids, e.g. instruction means
    • B62D15/0285—Parking performed automatically
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W60/00—Drive control systems specially adapted for autonomous road vehicles
    • B60W60/001—Planning or execution of driving tasks
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W2510/00—Input parameters relating to a particular sub-units
    • B60W2510/20—Steering systems

Definitions

  • the present application relates to the field of autonomous driving, and in particular, to a method and device for determining an automatic parking strategy.
  • the automatic parking system is a system that can automatically drive vehicles into parking spaces in parallel, vertical or oblique directions.
  • the core idea is to plan the steering angle and speed of the vehicle so that it can be used in a limited driveable space.
  • the on-board processor uses sensor data from the car servo system and distance data between the vehicle and the obstacle to calculate and execute a series of control actions in real time according to the parking strategy, so that the vehicle enters the parking space.
  • One available automatic parking method is to solve different parking scenarios (for example, different aisle widths, target parking space size, and starting position of the vehicle) to obtain available control strategies, such as through arc planning or vector
  • the parking area plans the parking path and obtains the available control strategies.
  • the above method has a strong generalization ability. However, for some complex parking scenarios, the above method is difficult to obtain satisfactory results. For example, the arc planning has higher requirements for parking space. In some small parking spaces, It cannot be implemented, and the solution calculated by the vector field cannot guarantee convergence, which eventually makes it difficult to realize automatic parking in complex parking scenarios.
  • the present application provides a method and device for determining an automatic parking strategy.
  • the entire parking process is divided into several parking phases, and different methods are used to obtain control strategies for each phase, which can improve automatic parking in complex parking scenarios. Success rate.
  • a method for determining an automatic parking strategy including: determining a target parking action corresponding to a current parking stage according to the automatic parking strategy, and the current parking stage is a plurality of processes included in a parking process of a vehicle.
  • One of the parking phases executing the target parking action; obtaining feedback information, which is used to indicate whether the result of performing the target parking action has reached a predetermined target, which is the relative position of the predetermined vehicle and the target parking space , And / or, the predetermined target is a state of the vehicle during parking; and an automatic parking strategy is updated according to the feedback information.
  • the execution device of the above method is, for example, an on-board processor.
  • the on-board processor can divide the parking process of the vehicle into three stages. For example, first, the angle between the longitudinal axis of the vehicle and the longitudinal axis of the target parking space is adjusted to less than 60 degrees. Then adjust the included angle to less than 30 degrees, and then adjust the included angle to less than 5 degrees. When the included angle is less than 5 degrees, parking can be considered successful.
  • the predetermined goal of each stage in this embodiment reduces the implementation difficulty, and it can be easier to obtain an automatic parking strategy for each stage. This improves the success rate of automatic parking in complex parking scenarios.
  • the method further includes: determining the current parking phase according to an included angle between the reference direction and a preset direction of the vehicle, where the current There is a preset corresponding relationship between the parking phase and the included angle.
  • a vehicle entering the target parking space cannot be considered a successful parking. It is also necessary to determine that the angle between the preset direction of the vehicle (for example, the longitudinal axis direction) and the reference direction (for example, the longitudinal axis direction of the target parking space) is less than Only a certain angle threshold can determine that the vehicle is parked successfully. Therefore, by referring to the relationship between the angle between the reference direction and the preset direction of the vehicle and the angle threshold, it is possible to accurately determine what stage the current parking phase is.
  • the preset direction of the vehicle is the direction of the longitudinal axis of the vehicle
  • the reference direction is the direction of the longitudinal axis of the target parking space
  • the included angle is an acute angle or a right angle.
  • the longitudinal axis of the vehicle is parallel to or nearly parallel to the longitudinal axis of the target parking space. Therefore, it is possible to use the longitudinal axis of the vehicle as the preset direction of the vehicle and the longitudinal axis of the target parking space as the reference direction. Convenient to determine whether the vehicle was parked successfully.
  • determining the current parking phase according to an included angle between the reference direction and a preset direction of the vehicle includes: when the included angle is greater than or equal to a first included angle threshold, determining that the current parking phase is an initial phase, and the first The included threshold is less than 90 degrees.
  • the first included angle threshold may be a threshold set according to expert experience. When the included angle is greater than or equal to the first included angle threshold, it indicates that the posture of the vehicle at this time is far from the posture when the parking is completed. Therefore, it can be determined that the vehicle is in the initial stage at this time, and the posture of the vehicle is adjusted according to the target parking action in the initial stage, so that the vehicle completes the predetermined goal in the initial stage, so as to facilitate the next stage of action.
  • the target parking action in the initial stage is: driving toward the target parking space according to the first steering angle, the first steering angle is less than the maximum steering angle of the vehicle, and the predetermined goal in the initial stage is: the included angle is less than or equal to the first The included angle threshold, and the vehicle enters the target parking space.
  • Both the target parking action and the predetermined target in the initial stage can be set according to the experience of an expert, wherein the first steering angle can be a relatively large angle in order to quickly adjust the vehicle to achieve the predetermined target.
  • the automatic parking strategy is: determining a parking action with the highest value among a plurality of parking actions corresponding to the current parking stage as a target parking action, and updating the automatic parking strategy according to the feedback information includes: : When the feedback information is the included angle at two neighboring moments, and when the absolute value of the difference between the included angles at two neighboring moments is greater than a preset included threshold value, the target parking motion is reduced the value of.
  • the value of the parking motion is directly proportional to the probability that the parking motion becomes the target parking motion.
  • the feedback information and the preset value increase and decrease rules are used to determine the target parking motion after the execution is completed. Value, you can check whether the target parking action is suitable for the current parking phase.
  • the processor updates the automatic driving strategy and re-determines the parking action with the highest value, which makes the automatic driving strategy continuously improve.
  • determining the current parking phase according to an included angle between the reference direction and a preset direction of the vehicle includes: when the included angle is greater than a second included angle threshold and smaller than the first included threshold threshold, determining the current parking phase as In the transition phase, the first included angle threshold and the second included angle threshold are both smaller than 90 degrees, and the second included angle threshold is smaller than the first included angle threshold.
  • the first included angle threshold value and the second included angle threshold value may be thresholds set according to expert experience.
  • the first included angle threshold value is less than 90 degrees.
  • the threshold value indicates that the vehicle's posture has been adjusted to a proper posture at this time. Therefore, it can be determined that the vehicle is in the transition phase at this time, and the vehicle's posture is adjusted according to the target parking behavior of the transition phase, so that the vehicle completes the transition phase. Pre-defined goals for the next phase of action.
  • the target parking action in the transition phase is: driving toward the target parking space according to the second steering angle, the second steering angle is equal to the maximum steering angle of the vehicle, and the predetermined target in the transition phase is: the included angle is less than or equal to the second The included angle threshold, and the vehicle enters the target parking space.
  • Both the target parking behavior and the predetermined target in the transition phase can be set according to the experience of the experts. Since the vehicle's posture has been adjusted to a proper posture during the initial phase, the second steering angle can be the maximum steering angle of the vehicle, in order to facilitate Quickly adjust the vehicle to achieve the intended goal.
  • the automatic parking strategy is: determining a parking action with the highest value among a plurality of parking actions corresponding to the current parking stage as a target parking action, and updating the automatic parking strategy according to the feedback information includes: : When the feedback information is ⁇ t and ⁇ t-1 , and when
  • the value of the parking motion is directly proportional to the probability that the parking motion becomes the target parking motion.
  • the feedback information and the preset value increase and decrease rules are used to determine the target parking motion after the execution is completed. Value, you can check whether the target parking action is suitable for the current parking phase.
  • the processor updates the automatic driving strategy and re-determines the parking action with the highest value, which makes the automatic driving strategy continuously improve.
  • determining the current parking phase according to an included angle between the reference direction and a preset direction of the vehicle includes: when the included angle is less than or equal to a second included angle threshold, determining that the current parking phase is a fine-tuning phase, and the second The included threshold is less than 90 degrees.
  • the second included angle threshold may be a threshold set according to expert experience.
  • the second included angle threshold is a value less than 90 degrees.
  • the included angle is less than or equal to the second included threshold value, it indicates that the vehicle's posture and The posture of the successful parking is almost the same. Therefore, it can be determined that the vehicle is in the fine adjustment phase at this time, and the vehicle's posture is adjusted according to the target parking action of the fine adjustment phase, so that the vehicle completes the predetermined goal of the fine adjustment phase and completes parking.
  • the target parking action in the fine-tuning phase is: driving toward the target parking space according to a third steering angle, the third steering angle is less than the maximum steering angle of the vehicle, and the predetermined goal in the fine-tuning phase is: the included angle is less than Or equal to the third included angle threshold, and the vehicle enters the target parking space, wherein the third included angle threshold is smaller than the second included angle threshold.
  • the target parking motion and the predetermined target in the fine-tuning phase can be set based on the experience of the experts. Because the posture of the vehicle during the fine-tuning phase is almost the same as that of the vehicle when parking is successful, the third steering angle can be a smaller angle. In order to achieve the intended goal.
  • the automatic parking strategy is: determining a parking action with the highest value among a plurality of parking actions corresponding to the current parking stage as a target parking action, and updating the automatic parking strategy according to the feedback information includes: : When the feedback information is d t , d t-1 , ⁇ t and ⁇ t-1 , and when
  • the value of the parking motion is directly proportional to the probability that the parking motion becomes the target parking motion.
  • the feedback information and the preset value increase and decrease rules are used to determine the target parking motion after the execution is completed. Value, you can check whether the target parking action is suitable for the current parking phase.
  • the processor updates the automatic driving strategy and re-determines the parking action with the highest value, which makes the automatic driving strategy continuously improve.
  • updating the automatic parking strategy according to the feedback information further includes: reducing the value of the target parking action when the feedback information is that a collision and / or out of bounds occurred during the execution of the target parking action; and / Or, when the feedback information is to reach a predetermined target, the value of the target parking action is increased.
  • the target parking action is not applicable to the current parking phase.
  • the value of the target parking action needs to be reduced. There is merit in the target parking action, and the value of the target parking action needs to be increased.
  • a device for determining an automatic parking strategy can implement the functions corresponding to the steps in the method according to the first aspect, and the functions can be implemented by hardware or can be executed by hardware.
  • the hardware or software includes one or more units or modules corresponding to the functions described above.
  • the device includes a processor and a communication interface, and the processor is configured to support the device to perform a corresponding function in the method according to the first aspect.
  • the communication interface is used to support communication between the device and other network elements.
  • the device may also include a memory for coupling to the processor, which stores program instructions and data necessary for the device.
  • a computer-readable storage medium stores computer program code, and when the computer program code is executed by a processing unit or a processor, the apparatus for determining an automatic parking strategy executes a first The method described in one aspect.
  • a computer program product includes computer program code, when the computer program code is determined by a communication unit or a communication interface of a device for an automatic parking strategy, and a processing unit or a processor is running. , So that the apparatus for determining an automatic parking strategy executes the method of the first aspect.
  • FIG. 1 is a schematic diagram of an automatic parking scene applicable to the present application
  • FIG. 2 is a schematic diagram of an automatic parking process provided by the present application.
  • FIG. 3 is a schematic diagram of another automatic parking process provided by the present application.
  • FIG. 4 is a schematic diagram of a method for determining an automatic parking strategy provided by the present application.
  • FIG. 5 is a schematic diagram of a self-car position at an initial stage provided by the present application.
  • FIG. 6 is a schematic diagram of a self-car position in a transition stage provided by the present application.
  • FIG. 7 is a schematic diagram of a self-vehicle position during a fine-tuning phase provided by the present application.
  • FIG. 8 is a schematic diagram of a device for determining an automatic parking strategy provided by the present application.
  • FIG. 9 is a schematic diagram of another apparatus for determining an automatic parking strategy provided by the present application.
  • FIG. 1 shows a schematic diagram of an automatic parking scene applicable to the present application.
  • the parking scene shown in Figure 1 includes 4 parking spaces and 4 cars, of which 3 cars occupy 3 parking spaces, and the fourth car (referred to as “self-driving car”) needs to drive in from the position shown in Figure 1 The remaining one of the four parking spaces (ie, the target parking space), and avoids collision with other vehicles during driving.
  • the vehicle is equipped with an environmental awareness module, a planning control module, and a vehicle control module.
  • the environmental awareness module is used to measure environmental information such as the position, orientation, location of the target parking space, and the distance between the vehicle and obstacles (including other vehicles).
  • the planning control module (for example, an on-board processor) is used to determine the target parking action according to the automatic parking strategy and the environmental information measured by the environmental awareness module, and output an action command corresponding to the target parking action.
  • the vehicle control module controls The action command output by the module controls the vehicle to drive around the obstacle into the target parking space to complete automatic parking.
  • the above automatic parking process is shown in Figure 2.
  • the process for determining an automatic parking strategy provided in this application includes:
  • the planning control module obtains the vehicle status information from the environment awareness module, including the location, orientation of the vehicle relative to the target parking space, and distance information between the vehicle and the surrounding obstacles.
  • S2 Determine the parking phase where the vehicle is located according to the vehicle status information, and adaptively define the corresponding action space and environmental feedback mechanism according to the characteristics of the current parking phase;
  • the first stage of S2.1 is the initial stage.
  • the goal is to adjust the vehicle to the ideal starting position.
  • S2.2 Phase II is the transition phase.
  • the goal is to adjust the position of the vehicle to the position of the target vehicle by making full use of space.
  • the third stage of S2.3 is the fine-tuning stage.
  • the goal is to fine-tune the self-driving position to the ideal parking position.
  • the planning control module selects a parking action (for example, action A1) from the current parking stage as the target parking action of the current parking stage, and the environmental feedback function is used to determine the value of action A1, that is, the action Whether A1 can continue as the target parking action.
  • a parking action for example, action A1
  • the environmental feedback function is used to determine the value of action A1, that is, the action Whether A1 can continue as the target parking action.
  • a ⁇ -greedy strategy in a deep Q-learning neural network is used to select a target parking motion from the motion space defined by S2, where the parking motion is randomly selected from the motion space defined by S2 as the The probability of the target parking motion is ⁇ , and the probability of randomly selecting the parking motion with the largest current Q value from the motion space defined by S2 as the target parking motion is 1- ⁇ .
  • the execution device of the method 400 may be a vehicle-mounted processor, a vehicle including a processor, or a server.
  • the application does not limit the execution device of the method 400.
  • the method 400 includes:
  • S410 Determine a target parking action corresponding to the current parking stage according to the automatic parking strategy, and the current parking stage is one of a plurality of parking stages included in a parking process of the vehicle.
  • the parking process can be divided into two parking stages, or the parking process can be divided into three parking stages, and the number of multiple parking stages can also be other values. It should be noted that the above parking process refers to a complete parking process. As an optional example, a vehicle (also referred to as "self-driving") does not need to go through a complete parking process to complete parking. For example, according to the angle between the longitudinal axis of the vehicle and the longitudinal axis of the target parking space, a complete parking process is divided into an initial stage and a fine-tuning stage. The initial stage has a larger included angle, and the fine-tuning stage has a smaller included angle.
  • the vehicle enters After the garage, it is determined that the target parking space is directly in front of the vehicle, that is, the angle between the longitudinal axis of the vehicle and the longitudinal axis of the target parking space is small, and the vehicle can directly enter the fine-tuning stage.
  • the automatic parking strategy may be to select a suitable parking action from a set of parking actions corresponding to the current parking stage as a target parking action.
  • the parking action is, for example, control of a steering angle and speed.
  • the appropriate parking action is, for example, It is the parking action with the highest value in the parking action set, or the appropriate parking action is, for example, the parking action with the least time in the parking action set without collision.
  • the automatic parking strategy may also be a target parking action determined in real time according to the current parking phase, for example, a target parking action determined according to an arc plan or a vector field. This application does not limit the automatic parking strategy.
  • the processor outputs an action instruction corresponding to the target parking action, so as to execute the target parking action.
  • Acquire feedback information which is used to indicate whether the result of performing the target parking action has reached a predetermined target
  • the predetermined target is the relative position of the predetermined vehicle and the target parking space, and / or the predetermined target is that the vehicle is parked The state of the car.
  • the feedback information may be the distance between the vehicle and the obstacle, or the relative position of the vehicle and the target parking space, or it may be "the predetermined target is reached” or "the predetermined target is not reached”.
  • the predetermined goal is to complete parking.
  • the processor may determine that the vehicle has reached the target parking action based on the feedback information.
  • Predetermined goal is to complete parking.
  • the processor may determine that the vehicle has reached the predetermined target by executing the target parking action based on the feedback information.
  • the predetermined target and feedback information can also make other content, for example, the predetermined target is a discrete small-angle steering angle, and the feedback information is the steering angle at each time. If the steering angle at each time is less than or equal to the steering angle specified by the predetermined target, Then, the processor determines that the target parking action performed by the vehicle has reached a predetermined target according to the feedback information; otherwise, it determines that the target parking action performed by the vehicle does not reach the predetermined target.
  • S420 and S430 can be executed during the actual parking process, or they can be executed in a simulator or a simulation environment.
  • the feedback information indicates that the execution of the target parking action has reached the predetermined target, it indicates that the target parking action is suitable for the current parking phase, and the value of the target parking action can be increased, making the target parking action easier to match the current parking phase. Matching, that is, updating the automatic parking strategy; if the feedback information indicates that the execution of the target parking action does not reach the predetermined target, it indicates that the target parking action is not applicable to the current parking phase, and the value of the target parking action can be reduced. This makes it difficult for the target parking action to match the current parking phase, that is, to update the automatic parking strategy.
  • the method 400 reduces the difficulty of achieving the goal of each stage and obtains the completeness by dividing the parking process of the vehicle into at least two stages.
  • the automatic parking strategy corresponding to the parking process is easier, thereby improving the success rate of automatic parking in complex parking scenarios.
  • the method 400 further includes:
  • a current parking phase is determined according to an included angle between the reference direction and a preset direction of the vehicle, wherein there is a preset correspondence relationship between the current parking phase and the included angle.
  • a vehicle entering the target parking space cannot be considered a successful parking. It is also necessary to determine that the angle between the preset direction of the vehicle (for example, the longitudinal axis direction) and the reference direction (for example, the longitudinal axis direction of the target parking space) is less than A certain angle threshold can determine that the vehicle is parked successfully. Therefore, the relationship between the angle between the preset direction and the reference direction of the vehicle and the angle threshold can accurately determine what stage the current parking phase is.
  • the preset direction of the vehicle is the direction of the longitudinal axis of the vehicle
  • the reference direction is the direction of the longitudinal axis of the target parking space
  • the angle between the longitudinal axis of the vehicle and the longitudinal axis of the target parking space is an acute angle or a right angle.
  • the projection of the vehicle on the ground is usually rectangular, and the longitudinal axis of the vehicle is the direction of the long side of the rectangle of the vehicle.
  • the longitudinal axis of the vehicle can also be defined in other ways, for example, the midpoint of the connection between the two front wheels of the vehicle Line, or the direction the driver is facing, or the direction the car is driving straight.
  • the target parking space is also generally rectangular, and the longitudinal axis of the target parking space can be defined as the direction in which the long sides of the rectangle are located.
  • the longitudinal axis of the vehicle is parallel to or nearly parallel to the longitudinal axis of the target parking space. Therefore, it is possible to use the longitudinal axis of the vehicle as the preset direction of the vehicle and the longitudinal axis of the target parking space as the reference direction. Convenient to determine whether the vehicle was parked successfully.
  • determining the current parking stage according to an included angle between the reference direction and a preset direction of the vehicle includes:
  • the included angle is greater than or equal to the first included angle threshold, it is determined that the current parking phase is the initial stage, and the first included angle threshold is less than 90 degrees;
  • the first included angle threshold may be a threshold set according to expert experience. When the included angle is greater than or equal to the first included angle threshold, it indicates that the posture of the vehicle at this time is far from the posture when the parking is completed. Therefore, it can be determined that the vehicle is in the initial stage at this time, and the posture of the vehicle is adjusted according to the target parking action in the initial stage, so that the vehicle completes the predetermined goal in the initial stage, so as to facilitate the next stage of action.
  • the target parking action in the initial stage is: driving toward the target parking space according to the first steering angle, the first steering angle is less than the maximum steering angle of the vehicle, and the predetermined goal in the initial stage is: the included angle is less than or equal to the first The included angle threshold, and the vehicle enters the target parking space.
  • Both the target parking action and the predetermined target in the initial stage can be set according to the experience of an expert, wherein the first steering angle can be a relatively large angle in order to quickly adjust the vehicle to achieve the predetermined target.
  • FIG. 5 shows a schematic diagram of the position and posture of the vehicle at the initial stage provided by the present application. Taking the longitudinal axis of the own vehicle as the preset direction of the own vehicle and the longitudinal axis of the target parking space as the reference direction, if the first included angle threshold is determined according to expert experience, the vertical axis of the own vehicle and the longitudinal axis of the target parking space If the angle ⁇ is 80 degrees, it is determined that the current parking phase is the initial phase.
  • the first steering angle may be a smaller angle (for example, 5 degrees), which may make the parking trajectory at the initial stage smoother.
  • the target parking action can be determined according to the initial parking strategy.
  • the automatic parking strategy in the initial stage is: determining a parking action with the highest value among a plurality of parking actions corresponding to the current parking stage as a target parking action, and updating the automatic parking strategy according to feedback information, Including: when the absolute value of the difference between the two adjacent times is greater than a preset included angle threshold, reducing the value of the target parking action; and / or, when the feedback information is for performing the target parking action When a collision and / or out of bound occurs during the process, the value of the target parking action is reduced; and / or, when the feedback information is that the predetermined goal is completed, the value of the target parking action is increased.
  • the value of the parking motion is directly proportional to the probability that the parking motion becomes the target parking motion.
  • the feedback information and the preset value increase and decrease rules are used to determine the target parking motion after the execution is completed.
  • the value can check whether the target parking action is suitable for the current parking phase.
  • the processor updates the automatic driving strategy and re-determines the parking action with the highest value, so that the automatic driving strategy is continuously improved.
  • the above-mentioned update of the automatic parking strategy may also include other rules that increase or decrease the value of the target parking action.
  • determining the current parking stage according to an included angle between a preset direction of the vehicle and a reference direction includes:
  • the included angle is larger than the second included threshold and smaller than the first included threshold, it is determined that the current parking phase is a transition phase, and the first included threshold and the second included threshold are both less than 90 degrees, and the second included angle The threshold is smaller than the first included threshold.
  • the first included angle threshold value and the second included angle threshold value may be thresholds set according to expert experience.
  • the first included angle threshold value is less than 90 degrees.
  • the threshold value indicates that the vehicle's posture has been adjusted to a proper posture at this time. Therefore, it can be determined that the vehicle is in the transition phase at this time, and the vehicle's posture is adjusted according to the target parking behavior of the transition phase, so that the vehicle completes the transition phase Pre-defined goals for the next phase of action.
  • the target parking action in the transition phase is: driving toward the target parking space according to the second steering angle, the second steering angle is equal to the maximum steering angle of the vehicle, and the predetermined target in the transition phase is: the included angle is less than or equal to the second The included angle threshold, and the vehicle enters the target parking space.
  • Both the target parking behavior and the predetermined target in the transition phase can be set according to the experience of the experts. Since the vehicle's posture has been adjusted to a proper posture during the initial phase, the second steering angle can be the maximum steering angle of the vehicle, in order to facilitate Quickly adjust the vehicle to achieve the intended goal.
  • FIG. 6 is a schematic diagram of a self-car position in a transition stage provided by the present application.
  • the longitudinal direction of the own vehicle is If the angle ⁇ between the axis and the longitudinal axis of the target parking space is 59 degrees, it is determined that the current parking phase is a transition phase. Because the posture is usually adjusted well when the vehicle is in the transition phase, the first steering angle can be the steering angle corresponding to the positive full wheel or the negative full wheel, which can make the vehicle quickly transition from the transition phase to the next phase, improving parking efficiency .
  • the target parking action can be determined according to the automatic parking strategy in the transition phase.
  • the automatic parking strategy in the transition phase is: determining the parking value with the highest value among the multiple parking actions corresponding to the current parking phase as the target parking action, and updating the automatic parking strategy according to the feedback information, including: When the feedback information is ⁇ t and ⁇ t-1 , and when
  • the value of the target parking action is increased
  • the value of the target parking action is reduced by 10
  • the value of the target parking action is increased by 10
  • the value of the target parking action is not increased or decreased.
  • the above-mentioned automatic parking strategy is updated based on the feedback information, that is, the environmental feedback mechanism in the transition stage.
  • the feedback mechanism in the transition stage may also include other rules that increase or decrease the value of the target parking action.
  • determining the current parking stage according to an included angle between a preset direction of the vehicle and a reference direction includes:
  • the included angle is less than or equal to the second included angle threshold, it is determined that the current parking phase is a fine adjustment phase, and the second included angle threshold is less than 90 degrees.
  • the second included angle threshold may be a threshold set according to expert experience.
  • the second included angle threshold is a value less than 90 degrees.
  • the included angle is less than or equal to the second included threshold value, it indicates that the vehicle's posture and The posture of the successful parking is almost the same. Therefore, it can be determined that the vehicle is in the fine adjustment phase at this time, and the vehicle's posture is adjusted according to the target parking action of the fine adjustment phase, so that the vehicle completes the predetermined goal of the fine adjustment phase and completes parking.
  • the target parking action in the fine-tuning phase is: driving toward the target parking space according to a third steering angle, the third steering angle is less than the maximum steering angle of the vehicle, and the predetermined goal in the fine-tuning phase is: the included angle is less than Or equal to the third included angle threshold, and the vehicle enters the target parking space, wherein the third included angle threshold is smaller than the second included angle threshold.
  • the target parking motion and the predetermined target in the fine-tuning phase can be set based on the experience of the experts. Because the posture of the vehicle during the fine-tuning phase is almost the same as that of the vehicle when parking is successful, the third steering angle can be a smaller angle. In order to achieve the intended goal.
  • FIG. 7 is a schematic diagram of a self-vehicle posture during a fine-tuning phase provided in the present application. Taking the longitudinal axis of the own vehicle as the preset direction of the own vehicle and the longitudinal axis of the target parking space as the reference direction, if the second included angle threshold value is determined according to the experience of experts, the longitudinal axis of the own vehicle and the longitudinal axis of the target parking space If the angle ⁇ is 9 degrees, it is determined that the current parking phase is the initial phase.
  • the third steering angle may be a smaller angle (for example, 1 degree), which may make the parking trajectory in the fine-tuning stage smoother.
  • the target parking action can be determined according to the automatic parking strategy of the fine-tuning phase.
  • the automatic parking strategy in the fine-tuning phase is: determining the parking value with the highest value among the multiple parking actions corresponding to the current parking phase as the target parking action, and updating the automatic parking strategy according to the feedback information, including: When the feedback information is d t , d t-1 , ⁇ t and ⁇ t-1 , and when
  • the above-mentioned update of the automatic parking strategy based on the feedback information is the environmental feedback mechanism in the fine-tuning phase, and the feedback mechanism in the transition phase may also include other rules that increase or decrease the value of the target parking action.
  • the device for determining an automatic parking strategy includes a hardware structure and / or a software module corresponding to each function.
  • this application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a certain function is performed by hardware or computer software-driven hardware depends on the specific application and design constraints of the technical solution. A professional technician can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.
  • This application may divide the functional units of the device for determining an automatic parking strategy according to the above method example.
  • each functional unit may be divided corresponding to each function in the manner shown in FIG. 4, or two or more functions may be divided.
  • the above integrated unit may be implemented in the form of hardware or in the form of software functional unit. It should be noted that the division of units in this application is schematic, and is only a logical function division. There may be another division manner in actual implementation.
  • FIG. 8 shows a possible structural schematic diagram of an apparatus for determining an automatic parking strategy involved in the foregoing embodiment.
  • the apparatus 800 for determining an automatic parking strategy includes a processing unit 801 and an obtaining unit 802.
  • the processing unit 801 is configured to support the device 800 for determining an automatic parking strategy, and perform the steps of determination, update, and the like shown in FIG. 4.
  • the obtaining unit 802 obtains feedback information.
  • the processing unit 801 and the acquisition unit 802 may also be used to perform other processes of the techniques described herein.
  • the apparatus 800 may further include a storage unit for storing program code and data of the apparatus 800. E.g:
  • the processing unit 801 is configured to determine a target parking action corresponding to a current parking stage according to an automatic parking strategy, and the current parking stage is one of a plurality of parking stages included in a parking process of the vehicle; and execute the target parking action.
  • the obtaining unit 802 is configured to obtain feedback information, which is used to indicate whether the execution of the target parking action has reached a predetermined target.
  • the processing unit 801 is further configured to update the automatic parking strategy according to the feedback information.
  • the processing unit 801 may be a processor or a controller, for example, it may be a central processing unit (CPU), a general-purpose processor, a digital signal processor (DSP), and an application-specific integrated circuit (application-specific integrated circuit). (ASIC), field programmable gate array (field programmable gate array, FPGA) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logical blocks, modules, and circuits described in connection with the disclosure of this application.
  • the processor may also be a combination that realizes computing functions, for example, a combination including one or more microprocessors, a combination of a DSP and a microprocessor, and so on.
  • the obtaining unit 802 may be a transceiver or a communication interface.
  • the storage unit may be a memory.
  • the processing unit 801 is a processor
  • the obtaining unit 802 is a communication interface
  • the storage unit is a memory
  • the device for determining an automatic parking strategy involved in this application may be the device shown in FIG. 9.
  • the apparatus 900 includes: a processor 901, a communication interface 902, and a storage 903.
  • the processor 901, the communication interface 902, and the memory 903 can communicate with each other through an internal connection path, and transfer control and / or data signals.
  • the device 800 and the device 900 provided in this application reduce the difficulty of achieving the goals of each stage by dividing the parking process of the vehicle into at least two stages, and it is easier to obtain the automatic parking strategy corresponding to the complete parking process. This improves the success rate of automatic parking in complex parking scenarios.
  • the steps in the device embodiment and the method embodiment correspond exactly, and the corresponding module executes the corresponding steps, for example, the obtaining unit executes the obtaining step in the method embodiment, and steps other than the obtaining step may be performed by a processing unit or a processor.
  • the obtaining unit executes the obtaining step in the method embodiment
  • steps other than the obtaining step may be performed by a processing unit or a processor.
  • a processing unit or a processor for the function of the specific unit, reference may be made to the corresponding method embodiment, which will not be described in detail.
  • the size of the sequence number of each process does not mean the order of execution.
  • the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of this application.
  • Computer-readable media may include, but are not limited to: magnetic storage devices (eg, hard disks, floppy disks, or magnetic tapes, etc.), optical disks (eg, compact discs (CDs), digital versatile discs (DVDs) Etc.), smart cards and flash memory devices (for example, erasable programmable read-only memory (EPROM), cards, sticks or key drives, etc.).
  • magnetic storage devices eg, hard disks, floppy disks, or magnetic tapes, etc.
  • optical disks eg, compact discs (CDs), digital versatile discs (DVDs) Etc.
  • smart cards and flash memory devices for example, erasable programmable read-only memory (EPROM), cards, sticks or key drives, etc.
  • various storage media described herein may represent one or more devices and / or other machine-readable media used to store information.
  • machine-readable medium may include, but is not limited to, various other media capable of storing, containing, and / or carrying instruction (s) and / or data.
  • the disclosed apparatus and method may be implemented in other ways.
  • the device embodiments described above are only schematic.
  • the division of the unit is only a logical function division.
  • multiple units or components may be combined or 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, which may be 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, may be located in one place, or may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the objective of the solution of this embodiment.
  • each functional unit in each embodiment of the present application may be integrated into one processing unit, or each of the units may exist separately physically, or two or more units may be integrated into one unit.
  • the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium.
  • the technical solution of the present application is essentially a part that contributes to the existing technology or a part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium, including Several instructions are used to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in the embodiments of the present application.
  • the aforementioned storage media include: U disks, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks or optical disks, and other media that can store program codes .

Landscapes

  • Engineering & Computer Science (AREA)
  • Transportation (AREA)
  • Mechanical Engineering (AREA)
  • Automation & Control Theory (AREA)
  • Human Computer Interaction (AREA)
  • Chemical & Material Sciences (AREA)
  • Combustion & Propulsion (AREA)
  • Control Of Driving Devices And Active Controlling Of Vehicle (AREA)
  • Steering Control In Accordance With Driving Conditions (AREA)
  • Traffic Control Systems (AREA)

Abstract

一种确定自动泊车策略的方法和装置，以及执行该方法的计算机可读存储介质。该方法包括：根据自动泊车策略确定与当前泊车阶段对应的目标泊车动作，当前泊车阶段为车辆的泊车过程包括的多个泊车阶段中的一个；执行该目标泊车动作；获取反馈信息，该反馈信息用于指示执行目标泊车动作的结果是否达到了预定目标，该预定目标为预定的车辆与目标车位的相对位置，和/或，该预定目标为车辆在泊车过程中的状态；根据反馈信息更新自动泊车策略。上述方法将整个泊车过程划分为几个泊车阶段，针对每个阶段使用不同的方法获取控制策略，可以提高复杂泊车场景中自动泊车的成功率。

Description

确定自动泊车策略的方法和装置
本申请要求于2018年06月29日提交中国专利局、申请号为201810696037.0、申请名称为“确定自动泊车策略的方法和装置”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及自动驾驶领域，尤其涉及一种确定自动泊车策略的方法和装置。
背景技术
自动泊车系统是一种可以通过平行、垂直或斜向的方式自动将车辆驶入车位的系统，其核心思想是对车辆的转向角和速度进行规划，从而得到在有限的可行驶空间内可执行的理想泊车路径。例如，车载处理器利用来自汽车伺服系统的传感器数据和车辆与障碍物的距离数据，根据泊车策略实时计算并执行一系列控制动作，使得车辆驶入车位。
一种可用的自动泊车方法是对不同的泊车场景(例如，不同过道宽度、目标车位大小和自车起始位姿)求解，以获取可用的控制策略，例如，通过圆弧规划或矢量场对泊车路径进行规划，获取可用的控制策略。
上述方法具有较强的泛化能力，然而，对于一些复杂的泊车场景，上述方法难以获得满意的结果，例如，圆弧规划对泊车空间的要求较高，在一些狭小的泊车空间中无法实施，矢量场计算得到的解不能保证收敛，最终导致在复杂的泊车场景中难以实现自动泊车。
发明内容
本申请提供一种确定自动泊车策略的方法和装置，将整个泊车过程划分为几个泊车阶段，针对每个阶段使用不同的方法获取控制策略，可以提高复杂泊车场景中自动泊车的成功率。
第一方面，提供了一种确定自动泊车策略的方法，包括：根据自动泊车策略确定与当前泊车阶段对应的目标泊车动作，当前泊车阶段为车辆的泊车过程包括的多个泊车阶段中的一个；执行该目标泊车动作；获取反馈信息，该反馈信息用于指示执行目标泊车动作的结果是否达到了预定目标，该预定目标为预定的车辆与目标车位的相对位置，和/或，该预定目标为车辆在泊车过程中的状态；根据反馈信息更新自动泊车策略。
上述方法的执行设备例如是车载处理器，车载处理器可以将车辆的泊车过程划分为三个阶段，例如，首先将车辆的纵轴线与目标车位的纵轴线的夹角调整到小于60度，再将上述夹角调整到小于30度，再将上述夹角调整到小于5度，当上述夹角小于5度时，可以认为泊车成功。相对于在负载泊车场景中直接将上述夹角调整到小于5度的方案，本实施例的每个阶段的预定目标均降低了实现难度，获得每个阶段的自动泊车策略可以更加容易，从而提高复杂泊车场景中自动泊车的成功率。
可选地，根据自动泊车策略确定与当前泊车阶段对应的目标泊车动作之前，所述方法还包括：根据参考方向与车辆的预设方向的夹角确定当前泊车阶段，其中，当前泊车阶段与所述夹角存在预设的对应关系。
通常情况下，车辆驶入目标车位并不能被认为是泊车成功，还需要确定车辆的预设方向(例如，纵轴线方向)与参考方向(例如，目标车位的纵轴线方向)的夹角小于某个夹角阈值才能确定该车辆泊车成功，因此，通过参考方向与车辆的预设方向的夹角与夹角阈值的大小关系能够准确判断当前泊车阶段具体处于什么阶段。
可选地，车辆的预设方向为车辆的纵轴线的方向，参考方向为目标车位的纵轴线的方向，且所述夹角为锐角或直角。
当车辆泊车成功时，车辆的纵轴线与目标车位的纵轴线是平行的或者接近平行的，因此，将车辆的纵轴线作为车辆的预设方向以及将目标车位的纵轴线作为参考方向能够更加方便判断车辆是否泊车成功。
可选地，根据参考方向与车辆的预设方向的夹角确定当前泊车阶段，包括：当所述夹角大于或等于第一夹角阈值时，确定当前泊车阶段为初始阶段，第一夹角阈值小于90度。
第一夹角阈值可以是根据专家经验设定的阈值，当所述夹角大于或等于第一夹角阈值时，说明车辆此时的位姿距离泊车完成时的位姿具有较大的差距，因此，可以确定此时车辆处于初始阶段，并按照初始阶段的目标泊车动作调整车辆的位姿，使得车辆完成初始阶段的预定目标，以便于进行下一阶段的动作。
可选地，初始阶段的目标泊车动作为：按照第一转向角向目标车位行驶，第一转向角小于车辆的最大转向角，初始阶段的预定目标为：所述夹角小于或等于第一夹角阈值，且，车辆进入目标车位。
初始阶段的目标泊车动作和预定目标均可以根据专家经验设定，其中，第一转向角可以是较大的角度，以便于快速调整车辆完成预定目标。
可选地，所述自动泊车策略为：确定与当前泊车阶段对应的多个泊车动作中价值最大的泊车动作为目标泊车动作，所述根据反馈信息更新自动泊车策略，包括：当反馈信息为相邻两个时刻的所述夹角时，并且，当相邻两个时刻的所述夹角的差的绝对值大于预设的夹角阈值时，减小目标泊车动作的价值。
泊车动作的价值与该泊车动作成为目标泊车动作的概率成正比，对于已确定的目标泊车动作，通过反馈信息以及预设的价值增减规则确定目标泊车动作被执行完成后的价值，可以检验目标泊车动作是否适合当前泊车阶段，每次目标泊车动作被执行完后，处理器更新自动驾驶策略，重新确定价值最大的泊车动作，使得自动驾驶策略不断完善。
可选地，根据参考方向与车辆的预设方向的夹角确定当前泊车阶段，包括：当所述夹角大于第二夹角阈值且小于第一夹角阈值时，确定当前泊车阶段为过渡阶段，第一夹角阈值和第二夹角阈值均小于90度，且第二夹角阈值小于第一夹角阈值。
第一夹角阈值和第二夹角阈值可以是根据专家经验设定的阈值，第一夹角阈值为小于90度的值，当所述夹角大于第二夹角阈值且小于第一夹角阈值时，说明车辆此时的位姿已经调整至一个合适的位姿，因此，可以确定此时车辆处于过渡阶段，并按照过渡阶段的目标泊车动作调整车辆的位姿，使得车辆完成过渡阶段的预定目标，以便于进行下一阶段的动作。
可选地，过渡阶段的目标泊车动作为：按照第二转向角向目标车位行驶，第二转向角等于车辆的最大转向角，过渡阶段的预定目标为：所述夹角小于或等于第二夹角阈值，且，车辆进入目标车位。
过渡阶段的目标泊车动作和预定目标均可以根据专家经验设定，由于初始阶段已经将车辆的位姿调整至合适的位姿，因此，第二转向角可以是车辆的最大转向角，以便于快速调整车辆完成预定目标。
可选地，所述自动泊车策略为：确定与当前泊车阶段对应的多个泊车动作中价值最大的泊车动作为目标泊车动作，所述根据反馈信息更新自动泊车策略，包括：当反馈信息为θ t和θ t-1时，并且，当|θ t|<|θ t-1|时，根据θ t和θ t-1增加目标泊车动作的价值，其中，θ t和θ t-1为相邻两个时刻的所述夹角，|θ t|≠0，目标泊车动作的价值与
Figure PCTCN2019092722-appb-000001
成正比。
泊车动作的价值与该泊车动作成为目标泊车动作的概率成正比，对于已确定的目标泊车动作，通过反馈信息以及预设的价值增减规则确定目标泊车动作被执行完成后的价值，可以检验目标泊车动作是否适合当前泊车阶段，每次目标泊车动作被执行完后，处理器更新自动驾驶策略，重新确定价值最大的泊车动作，使得自动驾驶策略不断完善。
可选地，根据参考方向与车辆的预设方向的夹角确定当前泊车阶段，包括：当所述夹角小于或等于第二夹角阈值时，确定当前泊车阶段为微调阶段，第二夹角阈值小于90度。
第二夹角阈值可以是根据专家经验设定的阈值，第二夹角阈值为小于90度的值，当所述夹角小于或等于第二夹角阈值时，说明车辆此时的位姿与泊车成功的位姿相差无几，因此，可以确定此时车辆处于微调阶段，并按照微调阶段的目标泊车动作调整车辆的位姿，使得车辆完成微调阶段的预定目标，完成泊车。
可选地，微调阶段的目标泊车动作为：按照第三转向角向所述目标车位行驶，第三转向角小于所述车辆的最大转向角，微调阶段的预定目标为：所述夹角小于或等于第三夹角阈值，且，车辆进入目标车位，其中，第三夹角阈值小于第二夹角阈值。
微调阶段的目标泊车动作和预定目标均可以根据专家经验设定，由于微调阶段车辆的位姿与泊车成功时车辆的位姿相差无几，因此，第三转向角可以是较小的角度，以便于完成预定目标。
可选地，所述自动泊车策略为：确定与当前泊车阶段对应的多个泊车动作中价值最大的泊车动作为目标泊车动作，所述根据反馈信息更新自动泊车策略，包括：当反馈信息为d t、d t-1、θ t和θ t-1时，并且，当|d t|<|d t-1|且|θ t|<|θ t-1|时，根据d t和d t-1增加目标泊车动作的价值，其中，d t和d t-1为相邻两个时刻的车辆与目标车位的欧几里得距离，θ t和θ t-1为该相邻两个时刻的所述夹角，|d t|≠0，目标泊车动作的价值与
Figure PCTCN2019092722-appb-000002
成正比。
泊车动作的价值与该泊车动作成为目标泊车动作的概率成正比，对于已确定的目标泊车动作，通过反馈信息以及预设的价值增减规则确定目标泊车动作被执行完成后的价值，可以检验目标泊车动作是否适合当前泊车阶段，每次目标泊车动作被执行完后，处理器更新自动驾驶策略，重新确定价值最大的泊车动作，使得自动驾驶策略不断完善。
可选地，根据反馈信息更新所述自动泊车策略，还包括：当反馈信息为执行目标泊车动作的过程中发生了碰撞和/或出界时，减小目标泊车动作的价值；和/或，当反馈信息为 达到预定目标时，增加目标泊车动作的价值。
若泊车过程中发生了碰撞和/或出界，说明目标泊车动作不适用于当前的泊车阶段，需要减小目标泊车动作的价值，若执行目标泊车动作之后达到了预定目标，说明目标泊车动作有可取之处，需要增加目标泊车动作的价值。
第二方面，提供了一种确定自动泊车策略的装置，该装置可以实现上述第一方面所涉及的方法中各个步骤所对应的功能，所述功能可以通过硬件实现，也可以通过硬件执行相应的软件实现。所述硬件或软件包括一个或多个与上述功能相对应的单元或模块。
在一种可能的设计中，该装置包括处理器和通信接口，该处理器被配置为支持该装置执行上述第一方面所涉及的方法中相应的功能。该通信接口用于支持该装置与其它网元之间的通信。该装置还可以包括存储器，该存储器用于与处理器耦合，其保存该装置必要的程序指令和数据。
第三方面，提供了一种计算机可读存储介质，该计算机可读存储介质中存储了计算机程序代码，该计算机程序代码被处理单元或处理器执行时，使得确定自动泊车策略的装置执行第一方面所述的方法。
第四方面，提供了一种计算机程序产品，该计算机程序产品包括：计算机程序代码，当该计算机程序代码被确定自动泊车策略的装置的通信单元或通信接口、以及处理单元或处理器运行时，使得确定自动泊车策略的装置执行上述第一方面的方法。
附图说明
图1是一种适用于本申请的自动泊车场景示意图；
图2是本申请提供的一种自动泊车流程的示意图；
图3是本申请提供的另一种自动泊车流程的示意图；
图4是本申请提供的一种确定自动泊车策略的方法的示意图；
图5是本申请提供的一种初始阶段的自车位姿示意图；
图6是本申请提供的一种过渡阶段的自车位姿示意图；
图7是本申请提供的一种微调阶段的自车位姿示意图；
图8是本申请提供的一种确定自动泊车策略的装置的示意图；
图9是本申请提供的另一种确定自动泊车策略的装置的示意图。
具体实施方式
下面将结合附图，对本申请中的技术方案进行描述。
图1示出了一种适用于本申请的自动泊车场景示意图。图1所示的泊车场景包括4个车位和4辆汽车，其中，3辆汽车占用了3个车位，第4个汽车(简称为“自车”)需要从图1所示的位置驶入4个车位中剩余的一个车位(即，目标车位)，并且在驶入过程中避免与其它车辆发生碰撞。
自车上安装有环境感知模块、规划控制模块和车辆控制模块，环境感知模块用于测量自车的位置、朝向、目标车位的位置以及自车与障碍物(包括其它车辆)的距离等环境信息，规划控制模块(例如，车载处理器)用于根据自动泊车策略以及环境感知模块测得的环境信息确定目标泊车动作，并输出目标泊车动作对应的动作命令，车辆控制模块根据规 划控制模块输出的动作命令控制自车绕开障碍物驶入目标车位，完成自动泊车。上述自动泊车流程如图2所示。
通常情况下，自车周围的障碍物越多，泊车环境越复杂，确定自动泊车策略的难度也较高，自车周围的障碍物越少，泊车环境越简单，确定自动泊车策略的难度也较低，下面，将结合图3描述本申请提供的确定自动泊车策略的方法。
如图3所示，本申请提供的确定自动泊车策略的流程包括：
S1.规划控制模块从环境感知模块获取自车状态信息，包括自车相对于目标车位的位置、朝向和自车与周边障碍物之间的距离信息。
S2.根据自车状态信息判断自车所处的泊车阶段，根据当前泊车阶段的特征自适应地定义对应的动作空间和环境反馈机制；
S2.1阶段一为初始阶段，目标是将自车调整至理想起始位姿。
S2.2阶段二为过渡阶段，目标是通过充分利用空间将自车位姿向目标车位位姿调整。
S2.3阶段三为微调阶段，目标是通过微调将自车位姿调整至理想泊车位姿。
S2.4根据S2.1-S2.3的阶段目标，得到适用于不同阶段的动作空间，并定义不同阶段的环境反馈函数，其中，动作空间是多个泊车动作的集合(例如，A1，A2，A3)，规划控制模块根据当前泊车阶段从中选择一个泊车动作(例如，动作A1)作为当前泊车阶段的目标泊车动作，环境反馈函数用于确定动作A1的价值，即，动作A1是否能够继续作为目标泊车动作。
S3.采用基于强化学习的方法从S2中定义的动作空间中选择所应执行的动作。例如，采用深度Q学习神经网络(deep Q-network，DQN)中的ε-greedy策略从S2定义的动作空间中选择目标泊车动作，其中，从S2定义的动作空间中随机选取泊车动作作为目标泊车动作的概率为ε，从S2定义的动作空间中随机选取当前Q值最大的泊车动作作为目标泊车动作的概率为1-ε。
S4.执行S3中所选择的动作，根据从S2中定义的环境反馈机制获得来自环境的反馈，即获取目标泊车行动作的好坏的即时评价信号。例如，执行完目标泊车动作后发生了碰撞给予一个负反馈，完成阶段性目标给予一个正反馈等。
S5.将从S2、S3和S4中获得的信息进行存储，通过小批量随机梯度下降的方式对不同阶段对应的强化学习模型进行更新；三个阶段的模型通过串联的方式训练获得。
S6.判断各阶段模型是否收敛。该步骤需要基于模型在应用模式下的达成阶段目标的成功率、所执行轨迹的平滑性和其它可能的评判标准进行判断，这里不做特别的限定。
基于上述流程，下面，将对本申请提供的确定自动泊车策略的方法做详细介绍。
如图4所示，方法400的执行设备可以是车载处理器，也可以是包括处理器的车辆，还可以是服务器，本申请对方法400的执行设备不作限定。方法400包括：
S410，根据自动泊车策略确定与当前泊车阶段对应的目标泊车动作，当前泊车阶段为车辆的泊车过程包括的多个泊车阶段中的一个。
可以将泊车过程划分为两个泊车阶段，也可以将泊车过程划分为三个泊车阶段，多个泊车阶段的数量还可以是其它数值。需要说明的是，上述泊车过程指的是一个完整的泊车过程，作为一个可选的示例，车辆(也可以称为“自车”)不需要经历完整的泊车过程即可完成泊车，例如，根据车辆的纵轴线与目标车位的纵轴线的夹角将一个完整的泊车过程划 分为初始阶段和微调阶段，初始阶段的夹角较大，微调阶段的夹角较小，车辆进入车库后确定目标车位在车辆的正前方，即，车辆的纵轴线与目标车位的纵轴线的夹角较小，则车辆可以直接进入微调阶段。
自动泊车策略可以是从当前泊车阶段对应的泊车动作集合中选择合适的泊车动作作为目标泊车动作，泊车动作例如是对转向角和速度的控制，该合适的泊车动作例如是泊车动作集合中价值最高的泊车动作，或者，该合适的泊车动作例如是在不发生碰撞的前提下泊车动作集合中用时最少的泊车动作。自动泊车策略也可以是根据当前泊车阶段实时确定的目标泊车动作，例如，根据圆弧规划或矢量场确定的目标泊车动作。本申请对自动泊车策略不作限定。
S420，执行该目标泊车动作。
处理器输出目标泊车动作对应的动作指令，以便于执行目标泊车动作。
S430，获取反馈信息，该反馈信息用于指示执行目标泊车动作的结果是否达到了预定目标，该预定目标为预定的车辆与目标车位的相对位置，和/或，该预定目标为车辆在泊车过程中的状态。
反馈信息可以是自车与障碍物的距离，也可以是自车与目标车位的相对位置，还可以是“达到了预定目标”或者“未达到预定目标”。例如，预定目标是完成泊车，当反馈信息为自车与临近车位的车辆(即，障碍物)的距离小于50厘米时，处理器可以根据该反馈信息确定自车执行目标泊车动作达到了预定目标。又例如，预定目标是完成泊车，当反馈信息为自车完全进入目标车位时，处理器可以根据该反馈信息确定自车执行目标泊车动作达到了预定目标。
预定目标和反馈信息还可以使其它内容，例如，预定目标是离散化的小角度的转向角，反馈信息是各个时刻的转向角，若各个时刻的转向角小于或等于预定目标规定的转向角，则处理器根据该反馈信息确定自车执行目标泊车动作达到了预定目标，反之，则确定自车执行目标泊车动作未达到预定目标。
需要说明的是，S420和S430和可以是在实际泊车过程中执行，也可以是在模拟器或者仿真环境中执行。
S440，根据反馈信息更新自动泊车策略。
若反馈信息指示执行目标泊车动作达到了预定目标，说明该目标泊车动作适用于当前泊车阶段，可以增加该目标泊车动作的价值，使得该目标泊车动作更加容易与当前泊车阶段匹配上，即，更新自动泊车策略；若反馈信息指示执行目标泊车动作未达到预定目标，说明该目标泊车动作不适用于当前泊车阶段，可以减小该目标泊车动作的价值，使得该目标泊车动作不容易与当前泊车阶段匹配上，即，更新自动泊车策略。
综上，相对于以完整的泊车过程为目标确定自动泊车策略的方法，方法400通过将车辆的泊车过程划分为至少两个阶段，降低了每个阶段的目标的实现难度，获取完整的泊车过程对应的自动泊车策略更加容易，从而提高了复杂泊车场景中自动泊车的成功率。
可选地，在S410之前，方法400还包括：
根据参考方向与车辆的预设方向的夹角确定当前泊车阶段，其中，当前泊车阶段与所述夹角存在预设的对应关系。
通常情况下，车辆驶入目标车位并不能被认为是泊车成功，还需要确定车辆的预设方 向(例如，纵轴线方向)与参考方向(例如，目标车位的纵轴线方向)的夹角小于某个夹角阈值才能确定该车辆泊车成功，因此，通过车辆的预设方向与参考方向的夹角与夹角阈值的大小关系能够准确判断当前泊车阶段具体处于什么阶段。
还可以根据其它方式(例如，车辆与目标车位的距离)确定当前泊车阶段处于泊车过程的哪个阶段，本申请对此不作限定。
可选地，车辆的预设方向为车辆的纵轴线的方向，参考方向为目标车位的纵轴线的方向，且车辆的纵轴线方向与目标车位的纵轴线方向的夹角为锐角或直角。
车辆在地面上的投影通常为长方形，车辆的纵轴线即为车辆的长方形的长边所在方向，还可以通过其它方式定义车辆的纵轴线，例如，车辆的两个前轮的连线的中垂线，或者，驾驶员面对的方向，或者，汽车直线行驶的方向。目标车位通常也是长方形，目标车位的纵轴线可以定义为该长方形的长边所在的方向。
当车辆泊车成功时，车辆的纵轴线与目标车位的纵轴线是平行的或者接近平行的，因此，将车辆的纵轴线作为车辆的预设方向以及将目标车位的纵轴线作为参考方向能够更加方便判断车辆是否泊车成功。
可选地，所述根据参考方向与车辆的预设方向的夹角确定当前泊车阶段，包括：
当夹角大于或等于第一夹角阈值时，确定当前泊车阶段为初始阶段，第一夹角阈值小于90度；
第一夹角阈值可以是根据专家经验设定的阈值，当所述夹角大于或等于第一夹角阈值时，说明车辆此时的位姿距离泊车完成时的位姿具有较大的差距，因此，可以确定此时车辆处于初始阶段，并按照初始阶段的目标泊车动作调整车辆的位姿，使得车辆完成初始阶段的预定目标，以便于进行下一阶段的动作。
可选地，初始阶段的目标泊车动作为：按照第一转向角向目标车位行驶，第一转向角小于车辆的最大转向角，初始阶段的预定目标为：所述夹角小于或等于第一夹角阈值，且，车辆进入目标车位。
初始阶段的目标泊车动作和预定目标均可以根据专家经验设定，其中，第一转向角可以是较大的角度，以便于快速调整车辆完成预定目标。
图5示出了本申请提供的初始阶段的自车位姿示意图。以自车的纵轴线为自车的预设方向，目标车位的纵轴线为参考方向，若根据专家经验确定第一夹角阈值为60度，自车的纵轴线与目标车位的纵轴线的夹角θ为80度，则确定当前泊车阶段为初始阶段。其中，第一转向角可以为较小的角度(例如，5度)，可以使得初始阶段的泊车轨迹更加平滑。
确定了当前泊车阶段为初始阶段后，即可根据初始阶段的自动泊车策略确定目标泊车动作。
可选地，初始阶段的自动泊车策略为：确定与当前泊车阶段对应的多个泊车动作中价值最大的泊车动作为目标泊车动作，所述根据反馈信息更新自动泊车策略，包括：当相邻两个时刻的所述夹角的差的绝对值大于预设的夹角阈值时，减小目标泊车动作的价值；和/或，当反馈信息为执行目标泊车动作的过程中发生了碰撞和/或出界时，减小目标泊车动作的价值；和/或，当反馈信息为完成预定目标时，增加目标泊车动作的价值。
泊车动作的价值与该泊车动作成为目标泊车动作的概率成正比，对于已确定的目标泊车动作，通过反馈信息以及预设的价值增减规则确定目标泊车动作被执行完成后的价值， 可以检验目标泊车动作是否适合当前泊车阶段，每次目标泊车动作被执行完后，处理器更新自动驾驶策略，重新确定价值最大的泊车动作，使得自动驾驶策略不断完善。
例如，当θ t-θ t-1>5°时，将目标泊车动作的价值减去0.05*|θ t-θ t-1|，θ t和θ t-1为相邻两个时刻的所述夹角(即，车辆的纵轴线方向与目标车位的纵轴线方向的夹角)；当执行目标泊车动作的过程中发生了碰撞和/或出界时，将目标泊车动作的价值减去10；当执行目标泊车动作后完成预定目标时，将目标泊车动作的价值增加10；当执行目标泊车动作的过程中出现其它状况时，不增加也不减小目标泊车动作的价值。
上述根据反馈信息更新自动泊车策略即初始阶段的环境反馈机制，初始阶段的反馈机制还可以包括其它增加或减小目标泊车动作的价值的规则。
可选地，根据车辆的预设方向与参考方向的夹角确定当前泊车阶段，包括：
当所述夹角大于第二夹角阈值且小于第一夹角阈值时，确定当前泊车阶段为过渡阶段，第一夹角阈值和第二夹角阈值均小于90度，且第二夹角阈值小于第一夹角阈值。
第一夹角阈值和第二夹角阈值可以是根据专家经验设定的阈值，第一夹角阈值为小于90度的值，当所述夹角大于第二夹角阈值且小于第一夹角阈值时，说明车辆此时的位姿已经调整至一个合适的位姿，因此，可以确定此时车辆处于过渡阶段，并按照过渡阶段的目标泊车动作调整车辆的位姿，使得车辆完成过渡阶段的预定目标，以便于进行下一阶段的动作。
可选地，过渡阶段的目标泊车动作为：按照第二转向角向目标车位行驶，第二转向角等于车辆的最大转向角，过渡阶段的预定目标为：所述夹角小于或等于第二夹角阈值，且，车辆进入目标车位。
过渡阶段的目标泊车动作和预定目标均可以根据专家经验设定，由于初始阶段已经将车辆的位姿调整至合适的位姿，因此，第二转向角可以是车辆的最大转向角，以便于快速调整车辆完成预定目标。
图6示出了本申请提供的过渡阶段的自车位姿示意图。以自车的纵轴线为自车的预设方向，目标车位的纵轴线为参考方向，若根据专家经验确定第一夹角阈值为60度，第二夹角阈值与10度，自车的纵轴线与目标车位的纵轴线的夹角θ为59度，则确定当前泊车阶段为过渡阶段。由于车辆处于过渡阶段时位姿通常调整的比较良好，第一转向角可以为正向满轮或负向满轮对应的转向角，可以使得车辆快速由过渡阶段进入下个阶段，提高泊车效率。
确定了当前泊车阶段为过渡阶段后，即可根据过渡阶段的自动泊车策略确定目标泊车动作。
可选地，过渡阶段的自动泊车策略为：确定与当前泊车阶段对应的多个泊车动作中价值最大的泊车动作为目标泊车动作，根据反馈信息更新自动泊车策略，包括：当反馈信息为θ t和θ t-1时，并且，当|θ t|<|θ t-1|时，根据θ t和θ t-1增加目标泊车动作的价值，其中，θ t和θ t-1为相邻两个时刻的所述夹角，|θ t|≠0，目标泊车动作的价值与
Figure PCTCN2019092722-appb-000003
成正比；和/或，当反馈信息为执行目标泊车动作的过程中发生了碰撞和/或出界时，减小目标泊车动作的价值；和/或，当反馈信息为完成预定目标时，增加目标泊车动作的价值。
例如，当|θ t|<|θ t-1|时，将目标泊车动作的价值增加
Figure PCTCN2019092722-appb-000004
当执行目标泊车动作的过程中发生了碰撞和/或出界时，将目标泊车动作的价值减去10；当执行目标泊车动作后完成预定目标时，将目标泊车动作的价值增加10；当执行目标泊车动作的过程中出现其它状况时，不增加也不减小目标泊车动作的价值。
上述根据反馈信息更新自动泊车策略即过渡阶段的环境反馈机制，过渡阶段的反馈机制还可以包括其它增加或减小目标泊车动作的价值的规则。
可选地，根据车辆的预设方向与参考方向的夹角确定当前泊车阶段，包括：
当所述夹角小于或等于第二夹角阈值时，确定当前泊车阶段为微调阶段，第二夹角阈值小于90度。
第二夹角阈值可以是根据专家经验设定的阈值，第二夹角阈值为小于90度的值，当所述夹角小于或等于第二夹角阈值时，说明车辆此时的位姿与泊车成功的位姿相差无几，因此，可以确定此时车辆处于微调阶段，并按照微调阶段的目标泊车动作调整车辆的位姿，使得车辆完成微调阶段的预定目标，完成泊车。
可选地，微调阶段的目标泊车动作为：按照第三转向角向所述目标车位行驶，第三转向角小于所述车辆的最大转向角，微调阶段的预定目标为：所述夹角小于或等于第三夹角阈值，且，车辆进入目标车位，其中，第三夹角阈值小于第二夹角阈值。
微调阶段的目标泊车动作和预定目标均可以根据专家经验设定，由于微调阶段车辆的位姿与泊车成功时车辆的位姿相差无几，因此，第三转向角可以是较小的角度，以便于完成预定目标。
图7示出了本申请提供的微调阶段的自车位姿示意图。以自车的纵轴线为自车的预设方向，目标车位的纵轴线为参考方向，若根据专家经验确定第二夹角阈值为10度，自车的纵轴线与目标车位的纵轴线的夹角θ为9度，则确定当前泊车阶段为初始阶段。其中，第三转向角可以为较小的角度(例如，1度)，可以使得微调阶段的泊车轨迹更加平滑。
确定了当前泊车阶段为微调阶段后，即可根据微调阶段的自动泊车策略确定目标泊车动作。
可选地，微调阶段的自动泊车策略为：确定与当前泊车阶段对应的多个泊车动作中价值最大的泊车动作为目标泊车动作，根据反馈信息更新自动泊车策略，包括：当反馈信息为d t、d t-1、θ t和θ t-1时，并且，当|d t|<|d t-1|且|θ t|<|θ t-1|时，根据d t和d t-1增加目标泊车动作的价值，其中，d t和d t-1为相邻两个时刻的车辆与目标车位的欧几里得距离，θ t和θ t-1为该相邻两个时刻的所述夹角，|d t|≠0，目标泊车动作的价值与
Figure PCTCN2019092722-appb-000005
成正比；和/或，当反馈信息为执行目标泊车动作的过程中发生了碰撞和/或出界时，减小目标泊车动作的价值；和/或，当反馈信息为完成预定目标时，增加目标泊车动作的价值。
例如，当|d t|<|d t-1|且|θ t|<|θ t-1|时，将目标泊车动作的价值增加
Figure PCTCN2019092722-appb-000006
当执行目标泊车动作的过程中发生了碰撞和/或出界时，将目标泊车动作的价值减去10；当执行目标泊车动作后完成预定目标时，将目标泊车动作的价值增加10；当执行目标泊车动作的过程中出现其它状况时，不增加也不减小目标泊车动作的价值。
上述根据反馈信息更新自动泊车策略即微调阶段的环境反馈机制，过渡阶段的反馈机 制还可以包括其它增加或减小目标泊车动作的价值的规则。
上文详细介绍了本申请提供的确定自动泊车策略的方法的示例。可以理解的是，确定自动泊车策略的装置为了实现上述功能，其包含了执行各个功能相应的硬件结构和/或软件模块。本领域技术人员应该很容易意识到，结合本文中所公开的实施例描述的各示例的单元及算法步骤，本申请能够以硬件或硬件和计算机软件的结合形式来实现。某个功能究竟以硬件还是计算机软件驱动硬件的方式来执行，取决于技术方案的特定应用和设计约束条件。专业技术人员可以对每个特定的应用来使用不同方法来实现所描述的功能，但是这种实现不应认为超出本申请的范围。
本申请可以根据上述方法示例对确定自动泊车策略的装置进行功能单元的划分，例如，可以按照图4所示的方式对应各个功能划分各个功能单元，也可以将两个或两个以上的功能集成在一个处理单元中。上述集成的单元既可以采用硬件的形式实现，也可以采用软件功能单元的形式实现。需要说明的是，本申请中对单元的划分是示意性的，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。
在采用集成的单元的情况下，图8示出了上述实施例中所涉及的确定自动泊车策略的装置的一种可能的结构示意图。确定自动泊车策略的装置800包括：处理单元801和获取单元802。处理单元801用于支持确定自动泊车策略的装置800执行图4所示的确定、更新等步骤。获取单元802于获取反馈信息。处理单元801和获取单元802还可以用于执行本文所描述的技术的其它过程。装置800还可以包括存储单元，用于存储装置800的程序代码和数据。例如：
处理单元801用于：根据自动泊车策略确定与当前泊车阶段对应的目标泊车动作，当前泊车阶段为车辆的泊车过程包括的多个泊车阶段中的一个；执行该目标泊车动作。
获取单元802用于：获取反馈信息，该反馈信息用于指示执行目标泊车动作是否达到了预定目标。
处理单元801还用于根据反馈信息更新自动泊车策略。
处理单元801可以是处理器或控制器，例如可以是中央处理器(central processing unit，CPU)，通用处理器，数字信号处理器(digital signal processor，DSP)，专用集成电路(application-specific integrated circuit，ASIC)，现场可编程门阵列(field programmable gate array，FPGA)或者其它可编程逻辑器件、晶体管逻辑器件、硬件部件或者其任意组合。其可以实现或执行结合本申请公开内容所描述的各种示例性的逻辑方框，模块和电路。所述处理器也可以是实现计算功能的组合，例如包含一个或多个微处理器组合，DSP和微处理器的组合等等。获取单元802可以是收发器或通信接口。存储单元可以是存储器。
当处理单元801为处理器，获取单元802为通信接口，存储单元为存储器时，本申请所涉及的确定自动泊车策略的装置可以为图9所示的装置。
参阅图9所示，该装置900包括：处理器901、通信接口902和存储903。其中，处理器901、通信接口902以及存储器903可以通过内部连接通路相互通信，传递控制和/或数据信号。
本领域的技术人员可以清楚地了解到，为了描述的方便和简洁，上述描述的装置和单元的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。
本申请提供的装置800和装置900，通过将车辆的泊车过程划分为至少两个阶段，降 低了每个阶段的目标的实现难度，获取完整的泊车过程对应的自动泊车策略更加容易，从而提高了复杂泊车场景中自动泊车的成功率。
装置实施例和方法实施例中的步骤完全对应，由相应的模块执行相应的步骤，例如获取单元执行方法实施例中的获取步骤，除获取步骤以外的其它步骤可以由处理单元或处理器执行。具体单元的功能可以参考相应的方法实施例，不再详述。
在本申请各个实施例中，各过程的序号的大小并不意味着执行顺序的先后，各过程的执行顺序应以其功能和内在逻辑确定，而不应对本申请的实施过程构成任何限定。
另外，本文中术语“和/或”，仅仅是一种描述关联对象的关联关系，表示可以存在三种关系，例如，A和/或B，可以表示：单独存在A，同时存在A和B，单独存在B这三种情况。另外，本文中字符“/”，一般表示前后关联对象是一种“或”的关系。
本申请的各个方面或特征可以实现成方法、装置或使用标准编程和/或工程技术的制品。本申请中使用的术语“制品”涵盖可从任何计算机可读器件、载体或介质访问的计算机程序。例如，计算机可读介质可以包括，但不限于:磁存储器件(例如，硬盘、软盘或磁带等)，光盘(例如，压缩盘(compact disc，CD)、数字通用盘(digital versatile disc，DVD)等)，智能卡和闪存器件(例如，可擦写可编程只读存储器(erasable programmable read-only memory，EPROM)、卡、棒或钥匙驱动器等)。另外，本文描述的各种存储介质可代表用于存储信息的一个或多个设备和/或其它机器可读介质。术语“机器可读介质”可包括但不限于:能够存储、包含和/或承载指令和/或数据的各种其它介质。
本领域普通技术人员可以意识到，结合本文中所公开的实施例描述的各示例的单元及算法步骤，能够以电子硬件、或者计算机软件和电子硬件的结合来实现。这些功能究竟以硬件还是软件方式来执行，取决于技术方案的特定应用和设计约束条件。专业技术人员可以对每个特定的应用来使用不同方法来实现所描述的功能，但是这种实现不应认为超出本申请的范围。
所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，上述描述的系统、装置和单元的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。
在本申请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，在本申请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。
所述功能如果以软件功能单元的形式实现并作为独立的产品销售或使用时，可以存储在一个计算机可读取存储介质中。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分或者该技术方案的部分可以以软件产品的形式体现出来，该计算机 软件产品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)执行本申请各个实施例所述方法的全部或部分步骤。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。
以上所述，仅为本申请的具体实施方式，但本申请的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本申请揭露的技术范围内，可轻易想到变化或替换，都应涵盖在本申请的保护范围之内。因此，本申请的保护范围应以所述权利要求的保护范围为准。

Claims (28)

  1. 一种确定自动泊车策略的方法，其特征在于，包括：
    根据自动泊车策略确定与当前泊车阶段对应的目标泊车动作，所述当前泊车阶段为车辆的泊车过程包括的多个泊车阶段中的一个；
    执行所述目标泊车动作；
    获取反馈信息，所述反馈信息用于指示执行所述目标泊车动作的结果是否达到了预定目标，所述预定目标为预定的所述车辆与目标车位的相对位置，和/或，所述预定目标为所述车辆在泊车过程中的状态；
    根据所述反馈信息更新所述自动泊车策略。
  2. 根据权利要求1所述的方法，其特征在于，所述根据自动泊车策略确定与所述当前泊车阶段对应的目标泊车动作之前，所述方法还包括：
    根据参考方向与所述车辆的预设方向的夹角确定所述当前泊车阶段，其中，所述当前泊车阶段与所述夹角存在预设的对应关系。
  3. 根据权利要求2所述的方法，其特征在于，所述车辆的预设方向为所述车辆的纵轴线方向，所述参考方向为所述目标车位的纵轴线方向，且所述夹角为锐角或直角。
  4. 根据权利要求3所述的方法，其特征在于，所述根据参考方向与所述车辆的预设方向的夹角确定所述当前泊车阶段，包括：
    当所述夹角大于或等于第一夹角阈值时，确定所述当前泊车阶段为初始阶段，所述第一夹角阈值小于90度。
  5. 根据权利要求4所述的方法，其特征在于，
    所述目标泊车动作为：按照第一转向角向所述目标车位行驶，所述第一转向角小于所述车辆的最大转向角，
    所述预定目标为：所述夹角小于或等于所述第一夹角阈值，且，所述车辆进入所述目标车位。
  6. 根据权利要求4或5所述的方法，其特征在于，所述自动泊车策略为：确定与所述当前泊车阶段对应的多个泊车动作中价值最大的泊车动作为所述目标泊车动作，
    所述根据所述反馈信息更新所述自动泊车策略，包括：
    当所述反馈信息为相邻两个时刻的所述夹角时，并且，当相邻两个时刻的所述夹角的差的绝对值大于预设的夹角阈值时，减小所述目标泊车动作的价值。
  7. 根据权利要求3至6中任一项所述的方法，其特征在于，所述根据参考方向与所述车辆的预设方向的夹角确定所述当前泊车阶段，包括：
    当所述夹角大于第二夹角阈值且小于第一夹角阈值时，确定所述当前泊车阶段为过渡阶段，所述第一夹角阈值和所述第二夹角阈值均小于90度，且所述第二夹角阈值小于所述第一夹角阈值。
  8. 根据权利要求7所述的方法，其特征在于，
    所述目标泊车动作为：按照第二转向角向所述目标车位行驶，所述第二转向角等于所述车辆的最大转向角，
    所述预定目标为：所述夹角小于或等于所述第二夹角阈值，且，所述车辆进入所述目标车位。
  9. 根据权利要求7或8所述的方法，其特征在于，所述自动泊车策略为：确定与所述当前泊车阶段对应的多个泊车动作中价值最大的泊车动作为所述目标泊车动作，
    所述根据所述反馈信息更新所述自动泊车策略，包括：
    当所述反馈信息为θ t和θ t-1时，并且，当|θ t|<|θ t-1|时，根据θ t和θ t-1增加所述目标泊车动作的价值，其中，θ t和θ t-1为相邻两个时刻的所述夹角，|θ t|≠0，所述目标泊车动作的价值与
    Figure PCTCN2019092722-appb-100001
    成正比。
  10. 根据权利要求3至9中任一项所述的方法，其特征在于，所述根据参考方向与所述车辆的预设方向的夹角确定所述当前泊车阶段，包括：
    当所述夹角小于或等于第二夹角阈值时，确定所述当前泊车阶段为微调阶段，所述第二夹角阈值小于90度。
  11. 根据权利要求10所述的方法，其特征在于，
    所述目标泊车动作为：按照第三转向角向所述目标车位行驶，所述第三转向角小于所述车辆的最大转向角，
    所述预定目标为：所述夹角小于或等于第三夹角阈值，且，所述车辆进入所述目标车位，其中，所述第三夹角阈值小于所述第二夹角阈值。
  12. 根据权利要求10或11所述的方法，其特征在于，所述自动泊车策略为：确定与所述当前泊车阶段对应的多个泊车动作中价值最大的泊车动作为所述目标泊车动作，
    所述根据所述反馈信息更新所述自动泊车策略，包括：
    当所述反馈信息为d t、d t-1、θ t和θ t-1时，并且，当|d t|<|d t-1|且|θ t|<|θ t-1|时，根据d t和d t-1增加所述目标泊车动作的价值，其中，d t和d t-1为相邻两个时刻的所述车辆与所述目标车位的欧几里得距离，θ t和θ t-1为所述相邻两个时刻的所述夹角，|d t|≠0，所述目标泊车动作的价值与
    Figure PCTCN2019092722-appb-100002
    成正比。
  13. 根据权利要求6或9或12所述的方法，其特征在于，所述根据所述反馈信息更新所述自动泊车策略，还包括：
    当所述反馈信息为执行所述目标泊车动作的过程中发生了碰撞和/或出界时，减小所述目标泊车动作的价值；和/或，
    当所述反馈信息为达到所述预定目标时，增加所述目标泊车动作的价值。
  14. 一种确定自动泊车策略的装置，其特征在于，包括处理单元和获取单元，
    所述处理单元用于：根据自动泊车策略确定与当前泊车阶段对应的目标泊车动作，所述当前泊车阶段为车辆的泊车过程包括的多个泊车阶段中的一个；
    所述处理单元还用于：执行所述目标泊车动作；
    所述获取单元用于：获取反馈信息，所述反馈信息用于指示执行所述目标泊车动作的结果是否达到了预定目标，所述预定目标为预定的所述车辆与目标车位的相对位置，和/或，所述预定目标为所述车辆在泊车过程中的状态；
    所述处理单元还用于：根据所述反馈信息更新所述自动泊车策略。
  15. 根据权利要求14所述的装置，其特征在于，所述根据自动泊车策略确定与所述当前泊车阶段对应的目标泊车动作之前，所述处理单元还用于：
    根据参考方向与所述车辆的预设方向的夹角确定所述当前泊车阶段，其中，所述当前泊车阶段与所述夹角存在预设的对应关系。
  16. 根据权利要求15所述的装置，其特征在于，所述车辆的预设方向为所述车辆的纵轴线方向，所述参考方向为所述目标车位的纵轴线方向，且所述夹角为锐角或直角。
  17. 根据权利要求16所述的装置，其特征在于，所述处理单元具体用于：
    当所述夹角大于或等于第一夹角阈值时，确定所述当前泊车阶段为初始阶段，所述第一夹角阈值小于90度。
  18. 根据权利要求17所述的装置，其特征在于，
    所述目标泊车动作为：按照第一转向角向所述目标车位行驶，所述第一转向角小于所述车辆的最大转向角，
    所述预定目标为：所述夹角小于或等于所述第一夹角阈值，且，所述车辆进入所述目标车位。
  19. 根据权利要求17或18所述的装置，其特征在于，所述自动泊车策略为：确定与所述当前泊车阶段对应的多个泊车动作中价值最大的泊车动作为所述目标泊车动作，
    所述处理单元具体用于：
    当所述反馈信息为相邻两个时刻的所述夹角时，并且，当相邻两个时刻的所述夹角的差的绝对值大于预设的夹角阈值时，减小所述目标泊车动作的价值。
  20. 根据权利要求16至19中任一项所述的装置，其特征在于，所述处理单元具体用于：
    当所述夹角大于第二夹角阈值且小于第一夹角阈值时，确定所述当前泊车阶段为过渡阶段，所述第一夹角阈值和所述第二夹角阈值均小于90度，且所述第二夹角阈值小于所述第一夹角阈值。
  21. 根据权利要求20所述的装置，其特征在于，
    所述目标泊车动作为：按照第二转向角向所述目标车位行驶，所述第二转向角等于所述车辆的最大转向角，
    所述预定目标为：所述夹角小于或等于所述第二夹角阈值，且，所述车辆进入所述目标车位。
  22. 根据权利要求20或21所述的装置，其特征在于，所述自动泊车策略为：确定与所述当前泊车阶段对应的多个泊车动作中价值最大的泊车动作为所述目标泊车动作，
    所述处理单元具体用于：
    当所述反馈信息为θ t和θ t-1时，并且，当|θ t|<|θ t-1|时，根据θ t和θ t-1增加所述目标泊车动作的价值，其中，θ t和θ t-1为相邻两个时刻的所述夹角，|θ t|≠0，所述目标泊车动作的价值与
    Figure PCTCN2019092722-appb-100003
    成正比。
  23. 根据权利要求16至22中任一项所述的装置，其特征在于，所述处理单元具体用于：
    当所述夹角小于或等于第二夹角阈值时，确定所述当前泊车阶段为微调阶段，所述第二夹角阈值小于90度。
  24. 根据权利要求23所述的装置，其特征在于，
    所述目标泊车动作为：按照第三转向角向所述目标车位行驶，所述第三转向角小于所述车辆的最大转向角，
    所述预定目标为：所述夹角小于或等于第三夹角阈值，且，所述车辆进入所述目标车位，其中，所述第三夹角阈值小于所述第二夹角阈值。
  25. 根据权利要求23或24所述的装置，其特征在于，所述自动泊车策略为：确定与所述当前泊车阶段对应的多个泊车动作中价值最大的泊车动作为所述目标泊车动作，
    所述处理单元具体用于：
    当所述反馈信息为d t、d t-1、θ t和θ t-1时，并且，当|d t|<|d t-1|且|θ t|<|θ t-1|时，根据d t和d t-1增加所述目标泊车动作的价值，其中，d t和d t-1为相邻两个时刻的所述车辆与所述目标车位的欧几里得距离，θ t和θ t-1为所述相邻两个时刻的所述夹角，|d t|≠0，所述目标泊车动作的价值与
    Figure PCTCN2019092722-appb-100004
    成正比。
  26. 根据权利要求19或22或25所述的装置，其特征在于，所述处理单元具体还用于：
    当所述反馈信息为执行所述目标泊车动作的过程中发生了碰撞和/或出界时，减小所述目标泊车动作的价值；和/或，
    当所述反馈信息为达到所述预定目标时，增加所述目标泊车动作的价值。
  27. 一种确定自动泊车策略的装置，其特征在于，包括处理器和存储器，所述存储单元存储有指令，当所述指令被所述处理单元运行时，使得所述处理单元执行如权利要求1至13中任一项所述的方法。
  28. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质存储有计算机程序，所述计算机程序使得计算机执行权利要求1至13中任一项所述的方法。
PCT/CN2019/092722 2018-06-29 2019-06-25 确定自动泊车策略的方法和装置 WO2020001423A1 (zh)

Priority Applications (3)

Application Number Priority Date Filing Date Title
EP19826527.4A EP3805062B1 (en) 2018-06-29 2019-06-25 Method and device for determining automatic parking strategy
EP22205142.7A EP4206051A1 (en) 2018-06-29 2019-06-25 Method and apparatus for determining automatic parking strategy
US17/134,858 US11897454B2 (en) 2018-06-29 2020-12-28 Method and apparatus for determining automatic parking strategy

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201810696037.0 2018-06-29
CN201810696037.0A CN109895764B (zh) 2018-06-29 2018-06-29 确定自动泊车策略的方法和装置

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US17/134,858 Continuation US11897454B2 (en) 2018-06-29 2020-12-28 Method and apparatus for determining automatic parking strategy

Publications (1)

Publication Number Publication Date
WO2020001423A1 true WO2020001423A1 (zh) 2020-01-02

Family

ID=66943111

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/092722 WO2020001423A1 (zh) 2018-06-29 2019-06-25 确定自动泊车策略的方法和装置

Country Status (4)

Country Link
US (1) US11897454B2 (zh)
EP (2) EP3805062B1 (zh)
CN (1) CN109895764B (zh)
WO (1) WO2020001423A1 (zh)

Families Citing this family (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109895764B (zh) * 2018-06-29 2023-06-27 华为技术有限公司 确定自动泊车策略的方法和装置
CN110347167B (zh) * 2019-08-27 2022-10-14 广州小鹏汽车科技有限公司 一种速度规划方法及速度规划系统
CN111126598B (zh) * 2019-12-19 2023-08-01 深圳南方德尔汽车电子有限公司 自动泊车方法、装置、计算机设备和存储介质
CN112092810B (zh) * 2020-09-24 2022-06-14 上海汽车集团股份有限公司 一种车辆泊出方法、装置及电子设备
CN113706916B (zh) * 2020-10-29 2023-01-17 董笑天 一种用于停车场的智慧停车管理系统
CN112339747A (zh) * 2020-10-30 2021-02-09 上海欧菲智能车联科技有限公司 自动泊车轨迹的生成方法、装置、电子设备及存储介质
CN113252366A (zh) * 2021-06-23 2021-08-13 奇瑞新能源汽车股份有限公司 车辆的自动泊车评测方法、装置、车辆及存储介质
CN113467481B (zh) * 2021-08-11 2022-10-25 哈尔滨工程大学 一种基于改进Sarsa算法的路径规划方法
CN115140022B (zh) * 2022-06-24 2024-05-24 重庆赛力斯新能源汽车设计院有限公司 自动泊车调试方法、装置、计算机设备和存储介质

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2010098170A1 (ja) * 2009-02-25 2010-09-02 アイシン精機株式会社 駐車支援装置
CN102602391A (zh) * 2012-03-16 2012-07-25 深圳市豪恩汽车电子装备有限公司 一种辅助泊车系统及方法
CN104260722A (zh) * 2014-09-23 2015-01-07 北京理工大学 一种自动泊车系统
CN105764773A (zh) * 2013-08-29 2016-07-13 罗伯特·博世有限公司 用于控制车辆的方法
CN107792062A (zh) * 2017-10-16 2018-03-13 北方工业大学 一种自动泊车控制系统
CN108121205A (zh) * 2017-12-13 2018-06-05 深圳市航盛电子股份有限公司 一种用于多种泊车场景的路径规划方法、系统及介质
CN109895764A (zh) * 2018-06-29 2019-06-18 华为技术有限公司 确定自动泊车策略的方法和装置

Family Cites Families (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102008027779A1 (de) * 2008-06-11 2009-12-17 Valeo Schalter Und Sensoren Gmbh Verfahren zur Unterstützung eines Fahrers eines Fahrzeugs beim Einparken in eine Parklücke
DE102009028261A1 (de) * 2009-08-05 2011-02-10 Robert Bosch Gmbh Verfahren und Steuerung zur Kalibrierung einer automatisch lenkenden Einparkhilfe
JP5516992B2 (ja) * 2010-11-30 2014-06-11 アイシン精機株式会社 駐車位置調整装置
CN105197010B (zh) * 2014-06-04 2018-03-27 长春孔辉汽车科技股份有限公司 辅助泊车系统以及辅助泊车控制方法
CN104627175B (zh) * 2015-01-22 2017-03-01 北京理工大学 一种人车交互智能泊车系统
CN105109482B (zh) 2015-08-24 2017-09-12 奇瑞汽车股份有限公司 停车入库方法及装置
KR101795151B1 (ko) * 2015-10-05 2017-12-01 현대자동차주식회사 주차안내 장치 및 방법
KR20170040633A (ko) * 2015-10-05 2017-04-13 현대자동차주식회사 주차지원 시스템 및 그의 주차완료기준 설정방법
CN105857306B (zh) * 2016-04-14 2018-07-10 中国科学院合肥物质科学研究院 一种用于多种泊车场景的车辆自主泊车路径规划方法
CN106627565B (zh) * 2016-11-21 2019-10-01 深圳市元征软件开发有限公司 一种泊车方法及车载设备
CN108407805B (zh) 2018-03-30 2019-07-30 中南大学 一种基于dqn的车辆自动泊车方法

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2010098170A1 (ja) * 2009-02-25 2010-09-02 アイシン精機株式会社 駐車支援装置
CN102602391A (zh) * 2012-03-16 2012-07-25 深圳市豪恩汽车电子装备有限公司 一种辅助泊车系统及方法
CN105764773A (zh) * 2013-08-29 2016-07-13 罗伯特·博世有限公司 用于控制车辆的方法
CN104260722A (zh) * 2014-09-23 2015-01-07 北京理工大学 一种自动泊车系统
CN107792062A (zh) * 2017-10-16 2018-03-13 北方工业大学 一种自动泊车控制系统
CN108121205A (zh) * 2017-12-13 2018-06-05 深圳市航盛电子股份有限公司 一种用于多种泊车场景的路径规划方法、系统及介质
CN109895764A (zh) * 2018-06-29 2019-06-18 华为技术有限公司 确定自动泊车策略的方法和装置

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3805062A4

Also Published As

Publication number Publication date
CN109895764A (zh) 2019-06-18
EP3805062A4 (en) 2021-07-28
EP4206051A1 (en) 2023-07-05
CN109895764B (zh) 2023-06-27
US20210114587A1 (en) 2021-04-22
EP3805062A1 (en) 2021-04-14
US11897454B2 (en) 2024-02-13
EP3805062B1 (en) 2022-12-21

Similar Documents

Publication Publication Date Title
WO2020001423A1 (zh) 确定自动泊车策略的方法和装置
CN110015306B (zh) 驾驶轨迹获取方法及装置
Zhang et al. Human-like autonomous vehicle speed control by deep reinforcement learning with double Q-learning
EP3699048B1 (en) Travelling track prediction method and device for vehicle
US11480971B2 (en) Systems and methods for generating instructions for navigating intersections with autonomous vehicles
US20190332110A1 (en) Reinforcement learning on autonomous vehicles
WO2022052406A1 (zh) 一种自动驾驶训练方法、装置、设备及介质
US11465650B2 (en) Model-free reinforcement learning
EP3647140A1 (en) Vehicle control method, device, and apparatus
CN111506058A (zh) 通过信息融合来计划自动驾驶的短期路径的方法及装置
CN107985313A (zh) 用于自主车辆的基于弹簧系统的变换车道方法
CN110348278B (zh) 用于自主驾驶的基于视觉的样本高效的强化学习框架
KR102166811B1 (ko) 심층강화학습과 운전자보조시스템을 이용한 자율주행차량의 제어 방법 및 장치
CN111948938A (zh) 规划用于自动驾驶车辆的开放空间轨迹的松弛优化模型
CN113682312B (zh) 一种融合深度强化学习的自主换道方法及系统
US20220230080A1 (en) System and method for utilizing a recursive reasoning graph in multi-agent reinforcement learning
CN109839937B (zh) 确定车辆自动驾驶规划策略的方法、装置、计算机设备
CN113110526B (zh) 一种模型训练的方法、无人驾驶设备的控制方法及装置
CN115402344A (zh) 泊车场景仿真方法和装置
US12017683B2 (en) Method and device for operating a self-driving car
CN114291092B (zh) 车辆换道控制方法、装置、电子控制单元及存储介质
US11295620B2 (en) Server and controlling method of server
CN113625718B (zh) 车辆的行驶路径规划方法
CN114997048A (zh) 基于探索策略改进的td3算法的自动驾驶车辆车道保持方法
CN116540602B (zh) 一种基于路段安全级别dqn的车辆无人驾驶方法

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: 19826527

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2019826527

Country of ref document: EP

Effective date: 20210110