CN107784709A - The method and apparatus for handling automatic Pilot training data - Google Patents
The method and apparatus for handling automatic Pilot training data Download PDFInfo
- Publication number
- CN107784709A CN107784709A CN201710792053.5A CN201710792053A CN107784709A CN 107784709 A CN107784709 A CN 107784709A CN 201710792053 A CN201710792053 A CN 201710792053A CN 107784709 A CN107784709 A CN 107784709A
- Authority
- CN
- China
- Prior art keywords
- state
- data
- hmm
- driving behavior
- abnormal
- Prior art date
- Legal status (The legal status 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 status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
- G07C5/00—Registering or indicating the working of vehicles
- G07C5/08—Registering or indicating performance data other than driving, working, idle, or waiting time, with or without registering driving, working, idle or waiting time
- G07C5/0808—Diagnosing performance data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/29—Graphical models, e.g. Bayesian networks
- G06F18/295—Markov models or related models, e.g. semi-Markov models; Markov random fields; Networks embedding Markov models
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
- G07C5/00—Registering or indicating the working of vehicles
- G07C5/08—Registering or indicating performance data other than driving, working, idle, or waiting time, with or without registering driving, working, idle or waiting time
- G07C5/0841—Registering performance data
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Traffic Control Systems (AREA)
Abstract
It is an object of the invention to provide a kind of method and apparatus for handling automatic Pilot training data.The method for the treatment of in accordance with the present invention automatic Pilot training data includes:A kind of method for handling automatic Pilot training data is provided, this method includes:Obtain the data filtering relevant information and automatic Pilot training data at multiple moment;According to data filtering relevant information, judge whether the driving behavior represented by the automatic Pilot training data at each moment is abnormal;Automatic Pilot training data corresponding to Exception Filter driving behavior.The technique according to the invention, which can be filtered out in automatic Pilot training data, causes the abnormal data of automatic Pilot, to ensure the stability of automatic Pilot and security.
Description
Technical field
The present invention relates to automatic Pilot technical field, more particularly to a kind of technology for handling automatic Pilot training data.
Background technology
Autonomous driving vehicle, it is that one kind realizes unpiloted intelligent automobile by computer system.Autonomous driving vehicle according to
Manually intelligence, vision calculating, radar, supervising device and global positioning system cooperative cooperating, allow computer can be not any
Under the operation of mankind's active, operate motor vehicles automatic safe.
Driving Scene under various environment is sufficiently complex and is difficult to predict, this is the difficult point of automatic Pilot problem.Therefore need
Perception, positioning, decision-making and planning are realized by merging the data of multiple sensors.Wherein, " decision-making " and " planning " is always
It is difficulties.The deep learning training of automatic Pilot, " decision-making " and " planning " for being suitable for solving in automatic Pilot are appointed
Business.Being trained needs to collect Driving Scene first and carries out Driving Decision-making, establishes the database of training data.Automatic Pilot is instructed
Practice data and be derived from the actual running environments of driver a bit, it is " dirty data " there are some in the training data of collection, can be drawn
It is unstable to play automatic Pilot.These " dirty datas " are not abnormal data caused by collection environment or equipment dependability, but are driven
Training data caused by normal meeting in the person's of sailing driving procedure, as driver drives into rest station bait, the random lane change of driver
Deng.Normally due to qualified data accounting is very high, these " dirty datas " do not interfere with automatic Pilot effect in many cases.But due to
The presence of " dirty data ", automatic Pilot can be caused to occur some abnormal behaviours once in a while.Requirement of the automatic Pilot to security is very
Height, therefore how to filter out " dirty data " in automatic Pilot training data is a problem highly studied.
The content of the invention
According to an embodiment of the invention, it is desirable to provide a kind of method and apparatus for handling automatic Pilot training data, so as to
It can filter out in automatic Pilot training data and cause the abnormal data of automatic Pilot, to ensure the stability of automatic Pilot and peace
Quan Xing.
Embodiment according to the first aspect of the invention, there is provided a kind of method for handling automatic Pilot training data, the party
Method includes:
A. the data filtering relevant information and automatic Pilot training data at multiple moment are obtained;
B. according to data filtering relevant information, the driving behavior represented by the automatic Pilot training data at each moment is judged
It is whether abnormal;
C. the automatic Pilot training data corresponding to Exception Filter driving behavior.
Specifically, step b includes:
The observer state sequence for the driving behavior that-foundation is defined based on HMM;
- HMM of driving behavior state is established, driving behavior state includes:Normal driving state and exception
Driving condition;
- according to data filtering relevant information and HMM, judging the observer state sequence at each moment is
No corresponding abnormal driving state.
Specifically, this, which establishes the step of HMM of driving behavior state, includes:
- establish the Sample Storehouse of HMM;
- data filtering the relevant information in Sample Storehouse determines the sight of the HMM training at each moment
Survey status switch;
- data filtering relevant information in Sample Storehouse and the observer state sequence of HMM training are right
HMM is trained, and determines the parameter of HMM.
Specifically, step b also includes:
Whether the judged result of abnormal driving state is corresponded to according to the observer state sequence at each moment, updates hidden Ma Erke
The parameter of husband's model.
Specifically, step b includes:
- obtain traffic signal information;
Whether-the driving behavior for judging to correspond to the moment according to data filtering relevant information and traffic signal information violates friendship
Drift is then;
Driving behavior when-determination violates the traffic regulations is abnormal driving behavior.
Specifically, the step of obtaining traffic signal information also includes:
- wireless signal that traffic signals Warning Mark is sent is received, obtain traffic signal information.
Specifically, step b includes:
- driving behavior when determining the stationary vehicle before vehicle start is abnormal driving behavior.
Embodiment according to the second aspect of the invention, there is provided a kind of device for handling automatic Pilot training data,
The device includes:
- data capture unit, it is configured to obtain the data filtering relevant information and automatic Pilot training data at multiple moment;
- abnormal deciding means, it is configured to, according to data filtering relevant information, judge the automatic Pilot training number at each moment
It is whether abnormal according to represented driving behavior;
- data filtering units, it is configured to the automatic Pilot training data corresponding to Exception Filter driving behavior.
Specifically, abnormal deciding means includes:
- observer state establishes module, is configured to establish the observation shape of the driving behavior defined based on HMM
State sequence;
- model building module, it is configured to establish the HMM of driving behavior state, driving behavior state bag
Include:Normal driving state and abnormal driving state;
- condition judgment module, it is configured to according to data filtering relevant information and HMM, when judging each
Whether the observer state sequence at quarter corresponds to abnormal driving state.
Specifically, model building module also includes:
- Sample Storehouse setting up submodule, it is configured to establish the Sample Storehouse of HMM;
- first state setting up submodule, the data filtering relevant information being configured in Sample Storehouse determine each moment
HMM training observer state sequence;
- model training submodule, it is configured to data filtering relevant information and HMM in Sample Storehouse
The observer state sequence of training, is trained to HMM, determines the parameter of HMM.
Specifically, abnormal deciding means also includes:
- parameter update module, it is configured to whether correspond to abnormal driving state according to the observer state sequence at each moment
Judged result, update the parameter of HMM.
Specifically, abnormal deciding means includes:
- traffic signals acquisition module, it is configured to obtain traffic signal information;
- traffic rules judge module, it is configured to judge to correspond to according to data filtering relevant information and traffic signal information
Whether the driving behavior at moment violates the traffic regulations;
- the first abnormal determining module, driving behavior when being configured to determine to violate the traffic regulations is abnormal driving behavior.
Specifically, traffic signals acquisition module also includes:
- signal receiving submodule, it is configured to receive the wireless signal that traffic signals Warning Mark is sent, obtains traffic signals
Information.
Specifically, abnormal deciding means includes:
- the second abnormal determining module, the driving behavior during stationary vehicle for being configured to determine before vehicle start is abnormal drives
Sail behavior.
Embodiment according to the third aspect of the present invention, there is provided a kind of computer equipment, including:At one or more
Manage device;Memory, for storing one or more programs, when one or more programs are executed by one or more processors,
So that the method that one or more processors perform processing automatic Pilot training data as the aforementioned.
Embodiment according to the fourth aspect of the present invention, there is provided a kind of computer-readable recording medium, store thereon
There is computer program, wherein, the computer program realizes foregoing processing automatic Pilot training data when being executed by processor
Method.
Embodiment according to the fifth aspect of the present invention, there is provided a kind of computer program product, when the computer
The method that foregoing processing automatic Pilot training data is realized when program product is performed by computer equipment.
It can be exported by the automatic Pilot training data as above filtered by any suitable output equipment, including it is but unlimited
In the display, projecting apparatus, printer of computer, for the control to automatic Pilot process or other suitable operations.
Compared with prior art, embodiments of the invention have advantages below:By filtering out in automatic Pilot training data
The safety and stability for causing the unstable data of automatic Pilot, improving automatic Pilot.Embodiments of the invention make
The method that the unstable data in automatic Pilot training data are filtered out with HMM, creatively by hidden Ma Erke
Husband's model is applied in the technical scheme filtered out of automatic Pilot training data, can not only make full use of existing training number
According to, obtain more accurately filtering out result, and can be updated according to the newest continuous iteration of training data, it is more multiple to adapt to
Miscellaneous automatic Pilot scene and environment.
Brief description of the drawings
By reading the detailed description made to non-limiting example made referring to the drawings, of the invention is other
Feature, objects and advantages will become more apparent upon:
Fig. 1 shows to be suitable to the block diagram for being used for realizing the exemplary computer system/server of embodiment of the present invention;
Fig. 2 is the schematic flow sheet of the processing automatic Pilot training data method based on embodiments of the invention;
Fig. 3 is to judge the whether abnormal flow of driving behavior by HMM based on embodiments of the invention
Schematic diagram;
Fig. 4 is the flow signal of the HMM for establishing driving behavior state based on embodiments of the invention
Figure;
Fig. 5 is to judge whether abnormal flow is shown for driving behavior by traffic signal information based on embodiments of the invention
It is intended to.
Fig. 6 is the schematic device of the processing automatic Pilot training data based on the preferred embodiments of the present invention;
Fig. 7 is to judge the whether abnormal exception of driving behavior by HMM based on embodiments of the invention
The schematic diagram of judging unit;
Fig. 8 is the schematic diagram of the model building module based on embodiments of the invention;
Fig. 9 is to judge whether abnormal exception is sentenced for driving behavior by traffic signal information based on embodiments of the invention
The schematic diagram of disconnected unit.
Same or analogous reference represents same or analogous part in accompanying drawing.
Embodiment
It should be mentioned that some exemplary embodiments are described as before exemplary embodiment is discussed in greater detail
The processing described as flow chart or method.Although operations are described as the processing of order by flow chart, therein to be permitted
Multioperation can be implemented concurrently, concomitantly or simultaneously.In addition, the order of operations can be rearranged.When it
The processing can be terminated when operation is completed, it is also possible to the additional step being not included in accompanying drawing.The processing
It can correspond to method, function, code, subroutine, subprogram etc..
Alleged within a context " computer equipment ", also referred to as " computer ", referring to can be by running preset program or referring to
Order performs the intelligent electronic device of the predetermined process process such as numerical computations and/or logical calculated, its can include processor with
Memory, the survival that is prestored in memory by computing device are instructed to perform predetermined process process, or by ASIC,
The hardware such as FPGA, DSP perform predetermined process process, or are realized by said two devices combination.Computer equipment includes but unlimited
In server, PC, notebook computer, tablet personal computer, smart mobile phone etc..
The computer equipment includes user equipment and the network equipment.Wherein, the user equipment includes but is not limited to electricity
Brain, smart mobile phone, PDA etc.;The network equipment includes but is not limited to single network server, multiple webservers form
Server group or the cloud being made up of a large amount of computers or the webserver based on cloud computing (Cloud Computing), wherein,
Cloud computing is one kind of Distributed Calculation, a super virtual computer being made up of the computer collection of a group loose couplings.Its
In, the computer equipment can isolated operation realize the present invention, also can access network and by with other calculating in network
The present invention is realized in the interactive operation of machine equipment.Wherein, the network residing for the computer equipment include but is not limited to internet,
Wide area network, Metropolitan Area Network (MAN), LAN, VPN etc..
It should be noted that the user equipment, the network equipment and network etc. are only for example, other are existing or from now on may be used
The computer equipment or network that can occur such as are applicable to the present invention, should also be included within the scope of the present invention, and to draw
It is incorporated herein with mode.
Method (some of them are illustrated by flow) discussed hereafter can be by hardware, software, firmware, centre
Part, microcode, hardware description language or its any combination are implemented.Implement when with software, firmware, middleware or microcode
When, to implement the program code of necessary task or code segment can be stored in machine or computer-readable medium and (for example deposit
Storage media) in.(one or more) processor can implement necessary task.
Concrete structure and function detail disclosed herein are only representational, and are for describing showing for the present invention
The purpose of example property embodiment.But the present invention can be implemented by many alternative forms, and it is not interpreted as
It is limited only by the embodiments set forth herein.
It should be appreciated that when a unit is referred to as " connecting " or during " coupled " to another unit, it can directly connect
Connect or be coupled to another unit, or there may be temporary location.On the other hand, when a unit is referred to as " directly connecting
Connect " or " direct-coupling " when arriving another unit, then in the absence of temporary location.It should in a comparable manner explain and be used to retouch
State the relation between unit other words (such as " between being in ... " compared to " between being directly in ... ", " and with ... it is adjacent
Closely " compared to " with ... be directly adjacent to " etc.).
Although it should be appreciated that may have been used term " first ", " second " etc. herein to describe unit,
But these units should not be limited by these terms.It is used for the purpose of using these terms by a unit and another unit
Make a distinction.For example, in the case of the scope without departing substantially from exemplary embodiment, it is single that first module can be referred to as second
Member, and similarly second unit can be referred to as first module.Term "and/or" used herein above include one of them or
Any combination of more listed associated items.
Term used herein above is not intended to limit exemplary embodiment just for the sake of description specific embodiment.Unless
Context clearly refers else, otherwise singulative used herein above "one", " one " also attempt to include plural number.Should also
When understanding, term " comprising " and/or "comprising" used herein above provide stated feature, integer, step, operation,
The presence of unit and/or component, and do not preclude the presence or addition of other one or more features, integer, step, operation, unit,
Component and/or its combination.
It should further be mentioned that in some replaces realization modes, the function/action being previously mentioned can be according to different from attached
The order indicated in figure occurs.For example, depending on involved function/action, the two width figures shown in succession actually may be used
Substantially simultaneously to perform or can perform in a reverse order sometimes.
The present invention is described in further detail below in conjunction with the accompanying drawings.
Fig. 1 shows the block diagram suitable for being used for the exemplary computer system/server for realizing embodiment of the present invention.Figure
The computer system/server 12 of 1 display is only an example, should not be to the function and use range band of the embodiment of the present invention
Carry out any restrictions.
As shown in figure 1, computer system/server 12 is showed in the form of universal computing device.Computer system/service
The component of device 12 can include but is not limited to:One or more processor or processing unit 16, system storage 28, connection
The bus 18 of different system component (including system storage 28 and processing unit 16).
Bus 18 represents the one or more in a few class bus structures, including memory bus or Memory Controller,
Peripheral bus, graphics acceleration port, processor or the local bus using any bus structures in a variety of bus structures.Lift
For example, these architectures include but is not limited to industry standard architecture (ISA) bus, MCA (MAC)
Bus, enhanced isa bus, VESA's (VESA) local bus and periphery component interconnection (PCI) bus.
Computer system/server 12 typically comprises various computing systems computer-readable recording medium.These media can be appointed
What usable medium that can be accessed by computer system/server 12, including volatibility and non-volatile media, it is moveable and
Immovable medium.
Memory 28 can include the computer system readable media of form of volatile memory, such as random access memory
Device (RAM) 30 and/or cache memory 32.Computer system/server 12 may further include it is other it is removable/no
Movably, volatile/non-volatile computer system storage medium.Only as an example, storage system 34 can be used for reading and writing
Immovable, non-volatile magnetic media (Fig. 1 is not shown, commonly referred to as " hard disk drive ").Although not shown in Fig. 1, can
To provide the disc driver being used for may move non-volatile magnetic disk (such as " floppy disk ") read-write, and to removable non-volatile
Property CD (such as CD-ROM, DVD-ROM or other optical mediums) read-write CD drive.In these cases, it is each to drive
Dynamic device can be connected by one or more data media interfaces with bus 18.Memory 28 can include at least one program
Product, the program product have one group of (for example, at least one) program module, and these program modules are configured to perform the present invention
The function of each embodiment.
Program/utility 40 with one group of (at least one) program module 42, such as memory 28 can be stored in
In, such program module 42 includes --- but being not limited to --- operating system, one or more application program, other programs
Module and routine data, the realization of network environment may be included in each or certain combination in these examples.Program mould
Block 42 generally performs function and/or method in embodiment described in the invention.
Computer system/server 12 can also be (such as keyboard, sensing equipment, aobvious with one or more external equipments 14
Show device 24 etc.) communication, it can also enable a user to lead to the equipment that the computer system/server 12 interacts with one or more
Letter, and/or any set with make it that the computer system/server 12 communicated with one or more of the other computing device
Standby (such as network interface card, modem etc.) communicates.This communication can be carried out by input/output (I/O) interface 22.And
And computer system/server 12 can also pass through network adapter 20 and one or more network (such as LAN
(LAN), wide area network (WAN) and/or public network, such as internet) communication.As illustrated, network adapter 20 passes through bus
18 communicate with other modules of computer system/server 12.It should be understood that although not shown in Fig. 1, computer can be combined
Systems/servers 12 use other hardware and/or software module, include but is not limited to:Microcode, device driver, at redundancy
Manage unit, external disk drive array, RAID system, tape drive and data backup storage system etc..
Processing unit 16 is stored in the program in memory 28 by operation, so as to perform various function application and data
Processing.
For example, the various functions for performing the present invention and the computer program of processing, processing are stored with memory 28
When unit 16 performs corresponding computer program, the present invention is implemented in the identification that network-side is intended to incoming call.
It is described in detail below present invention determine that concrete function/step for handling automatic Pilot training data.
Fig. 2 is the schematic flow sheet of the method for the processing automatic Pilot training data of embodiments of the invention.The present invention's
The method of embodiment is used to handle automatic Pilot training data, can be realized by electronic equipment.Electronic equipment is included but not
It is limited to computer equipment and vehicle electronics.Computer equipment refers to that number can be performed by running preset program or instruction
The intelligent electronic device of the predetermined process process such as value calculating and/or logical calculated, it can include processor and memory, by
Reason device performs the survival to prestore in memory and instructed to perform predetermined process process, or by hardware such as ASIC, FPGA, DSP
Predetermined process process is performed, or is realized by said two devices combination.Computer equipment includes but is not limited to server, personal electricity
Brain, notebook computer, tablet personal computer, smart mobile phone etc..Server include but is not limited to individual server, multiple server groups into
Or the cloud being made up of a large amount of computers or server based on cloud computing, wherein, cloud computing is one kind of Distributed Calculation, by one
One super virtual computer of the computer collection composition of group's loose couplings.Vehicle electronics are to be used on automobile or and vapour
The related electronic equipment of car.
As shown in Fig. 2 according to the present embodiment processing automatic Pilot training data method include step S1, step S2 and
Step S3.
First, in step sl, the data filtering relevant information and automatic Pilot training data at multiple moment are obtained.Automatically
Driver training data refer to the data trained or learnt for automatic Pilot.These data can be that driver drives vehicle processes
The data that navigation map such as each inductor, monitor or radar on middle vehicle etc. gathers together, can be each sense
Answer device, monitor or the radar caused data in caused data or software or database in drive simulating environment.
Data filtering relevant information refers to that for filtering out the unstable training data of automatic Pilot can be caused in automatic Pilot training data
Required information, including following information:Gps data, cartographic information, vehicle control parameters information.Gps data refers to and satellite
Data message acquired in the satellite communication of Position Fixing Navigation System, including the positional information of time and vehicle.In the present invention,
Gps data is not limited to the data message obtained from GPS (Global Position System, global positioning system), other
The similar navigation positioning system from Beidou navigation satellite system, Galilean satellite alignment system or existing at present or following appearance
The information such as the time of middle acquisition and position is such as applicable to the present invention, should also be included within the scope of the present invention, and to draw
It is incorporated herein with mode.Cartographic information refers to the data message obtained from navigation map, including at least the road for having vehicle drive
The position on road, road shape, road attribute, road mounting situation, road junction, the information such as direction of passing through.Wagon control
Parameter information refers to when driver drives vehicle to caused by the control operations such as the at the uniform velocity traveling of vehicle, acceleration, braking, steering
The data of control parameter, such as the steering wheel angle of the speed of vehicle, vehicle, vehicle turn signal etc..Data filter out relevant information
In gps data and cartographic information can be obtained from the digital map navigation equipment with GPS receiver device, vehicle control parameters letter
Breath can be obtained by the inductor installed on car, or pass through the vehicle driving trace at each moment and cartographic information etc. one
Rise, obtained after computing.Such as, it is known that the position of the vehicle at each moment, information is it is known that at two according to the map
Vehicle between quarter is all straight-line travelling and knows the distance between two moment points of vehicle, it is hereby achieved that the speed of vehicle
Degree.
Data filter out relevant information and automatic Pilot training data carries multiple timing point informations, and each moment point is corresponding
One group of data filters out relevant information and automatic Pilot training data.These timing point informations can be gps time or during from GPS
Between at the time of point information or other times source at the time of point information.These moment points can be that time interval is fixed,
As 1/40s or time interval it is unfixed, it is scattered, such as:In one section of driving locus, starting or stop segment process
At the time of dot frequency be 25 times per second, the middle frequency at the uniform velocity travelling section is 5 times per second.
Automatic Pilot training data and data, which filter out relevant information, can include the driving data of one section of driving locus, also may be used
To include the driving data of multistage driving locus.The data information content that automatic Pilot training data includes is a lot.It is if automatic
Include the data that data filter out relevant information in driver training data, that data filters out relevant information and can be from automatic
Driver training data.
In step s 2, according to data filtering relevant information, judge represented by the automatic Pilot training data at each moment
Driving behavior it is whether abnormal.The unstable training data of automatic Pilot training does not refer to data mutation or irrational exception
Data, such as occur a speed very little or very big value suddenly in the training data that speed is stablized always, because speed is not
There may be so violent change, therefore this kind of data are the wrong data or irrational exception occurred in data acquisition
Data, can be by eliminating to modes such as data filterings.But this data are not to wish what is filtered out in embodiments of the invention
Training data.What is filtered out in embodiments of the invention is the automatic Pilot training data for making automatic Pilot unstable.It is for example, automatic
Driver training data be the vehicle directly gathered it is actual drive data when, during driver is refueled or thought in driving procedure
Way parking rest, therefore the off-line training data of vehicle occurs in training data;If driver makes a phone call with other people
Or chat, it may be noted that power is not concentrated so that the training data of emergency brake occurs in driving data.If these data are used for automatic Pilot
Deep learning or training if, will cause when Function for Automatic Pilot activates may occur during automatic Pilot anxious
Brake or off-line situation.Therefore need to try every possible means to filter out, to improve the stability of automatic Pilot.
Specifically, step S2 is to filter out relevant information according to the data at each moment, to driving corresponding to the moment
Member driving behavior judged, judge driving behavior whether be automatic Pilot training abnormal driving behavior.Automatic Pilot is instructed
Experienced abnormal driving behavior refers to influential on automatic Pilot, makes the driving behavior that automatic Pilot is unstable.Automatic Pilot is instructed
Experienced abnormal driving behavior includes:Abnormal acceleration, abnormal brake, exception are turned to, drifted off the course.Here " abnormal accelerate ", " different
Exception in the abnormal driving behavior such as often brake ", " abnormal to turn to " refers to the abnormal driving for not being suitable for automatic Pilot training
Behavior.Judge that the mode of driving behavior includes:Data can be filtered out relevant information be input to data graphic simulation it is soft
In part, by the amplitude of each data of figure shows vehicle control parameters information, when former and later two moment amplitudes of each data
Change exceedes threshold value, if now showing the driver without exception, corresponding of environment where vehicle in automatic Pilot training data
Behavior is defined as abnormal driving behavior.It can in addition contain the content by manually combining automatic Pilot training data, to vehicle control
The amplitude change of each data in parameter information processed is manually marked more than the abnormity point of threshold value, marks out the corresponding moment
Driving behavior is abnormal driving behavior.
Fig. 3 is according to an embodiment of the invention to judge the whether abnormal flow of driving behavior by HMM
Schematic diagram.As shown in figure 3, in a preferred embodiment, step S2 includes step S21, step S22 and step S23.
In the step s 21, the observer state sequence of the driving behavior defined based on HMM is established.Drive row
For observer state sequence refer to the state that can be observed in HMM, in the present invention by driving behavior as
The Observable state of HMM.The observer state sequence of driving behavior includes following Observable state:Normally travel
State, anxious acceleration mode, state of bringing to a halt, state of taking a sudden turn, state of drifting off the course.To data filter relevant information by entering constantly
Row sampling, the vehicle control information and cartographic information at corresponding each moment can be obtained.Believed according to vehicle control information and map
Breath can obtain the Observable state of the driving behavior at each moment.Such as:According to vehicle control information it is known that moment t1
Vehicle speed v1, steering wheel for vehicle corner k1, it is then increased according to the speed v0 of previous moment vehicle, the speed of vehicle
Acceleration a1=(v1-v0)/(t1-t0), if a1 value has exceeded default acceleration threshold value, the observer state at current time is
Anxious acceleration mode;Acceleration b1=(v0-v1)/(t1-t0) that the speed of vehicle reduces, if b1 value has exceeded default deceleration
Threshold value, then the observer state at current time is state of bringing to a halt;According to the steering wheel angle k0 of previous moment vehicle, the side of vehicle
Amplitude of variation c1=(k1-k0)/(t1-t0) of the angle rotated to disk, if c1 value has exceeded default corner threshold value, when
The observer state at preceding moment is zig zag state;Information is known that the position of vehicle driving trace according to the map, if when current
The vehicle driving trace at quarter differed with the position of the vehicle driving trace of detachment on cartographic information apart from setover threshold value, then
The observer state at current time is state of drifting off the course;If vehicle is not at anxious acceleration mode, state of bringing to a halt, zig zag state
With drift off the course state when, it is believed that the observer state at current time is normal vehicle operation state.Can be by each moment above
Each vehicle running state composition Observable state chain of point.For example the observer state of the driving behavior at moment 1 is the normal row of vehicle
Sail state, the observer state at moment 2 is vehicle zig zag state, and the observer state at moment 3 is that vehicle drifts off the course state, then car
Normally travel state->Vehicle zig zag state->The vehicle state of drifting off the course constitutes moment 1->The Observable shape at moment 3
State chain.Above Observable state can be encoded, in order to realize and understand.It should be noted that above with respect to acquisition
The content of the Observable state of driving behavior is only for example, other it is existing or the acquisition driving behavior that is likely to occur from now on can
The content of observer state is such as applicable to the present invention, should also be included within the scope of the present invention, and include by reference
In this.
In step S22, the HMM of driving behavior state is established, driving behavior state includes:Normally drive
Sail state and abnormal driving state.Normal driving state and abnormal driving state are two implicit shapes in HMM
State.The process for establishing the HMM of driving behavior state is mainly to determine the parameter of HMM, i.e., just
Beginning probability vector, state transition probability matrix and observation probability matrix.In step S22, artificial labeled data can be used
Mode determines HMM parameter.The mode of this artificial labeled data needs substantial amounts of artificial mark training number
According to.
Fig. 4 is the flow signal of the HMM for establishing driving behavior state of the preferred embodiments of the present invention
Figure.As shown in figure 4, in the preferred embodiment, HMM is established by way of model learning or model training.
Wherein, the step of establishing the HMM of driving behavior state includes step S31, step S32 and step S33.
In step S31, the Sample Storehouse of HMM is established.The Sample Storehouse of HMM is hidden Ma Er
Sample database that can be required for model learning, including the data with time information of one or more driving locus sections filter out
Relevant information.
In step s 32, the data filtering relevant information in Sample Storehouse determines the Hidden Markov mould at each moment
The observer state sequence of type training.Because the Sample Storehouse of model, which includes multiple segment data, filters out relevant information, therefore to each
The data of individual driving locus section filter out relevant information and are handled to obtain observer state sequence corresponding to multiple moment points.By one
The observer state sequence of multiple moment points of driving locus section is connected as an observer state sequence chain.The length of observer state chain
Degree can be the number of the whole moment points included in one section of driving locus or be determined according to equipment or system capability
One value.The length of Observable state chain is unrestricted in an embodiment of the present invention.
In step S33, the observation of data filtering relevant information and HMM training in Sample Storehouse
Status switch, HMM is trained, determines the parameter of HMM.Determine HMM
Parameter namely determine probability vector, state transition probability matrix and the observation probability matrix of model.Probability to
Amount includes that initial time is in normal driving state and initial time is in abnormal driving shape probability of state.State transition probability square
Battle array contains the probability mutually shifted between normal driving state and abnormal driving state.Observation probability matrix, which contains, normally to be driven
Sail the probability that each observer state is generated under state and abnormal driving state.The data of one section of driving locus in Sample Storehouse are filtered
HMM is trained except relevant information:In the driving locus section, when first obtaining each according to step S32
The value of the observer state sequence at quarter, for example, determining that the moment 1 is normally travel state according to vehicle control parameters, the moment 2 is to accelerate
State, moment 3 are normally travel states, and the moment 4 is braking state ...;Then hidden horse is estimated according to data filtering relevant information
The parameter (probability vector, state transition probability matrix and observation probability matrix) of Er Kefu models so that join in the model
The maximum probability of the observation sequence of several lower acquisitions.Maximum likelihood can be used by estimating the algorithm of the parameter of HMM
Method, Baum-Welch algorithms or other algorithms can also be used to be trained study.Estimate the parameter of HMM
Specific algorithm is unrestricted in the present invention.By the data of multiple driving locus in Sample Storehouse filter out relevant information all respectively to
Upper method is trained to the HMM of driving behavior state.The HMM that each training is obtained
Parameter is handled, it is determined that the parameter value of final HMM.For example, obtained according to multistage model training data
Transition probability P of the normal driving state of every segment data to normal driving state1, P2, P3... ..., PI(I represents model training number
According to total hop count), then the transition probability P of the normal driving state in end-state transition probability matrix to normal driving state is
P1, P2, P3... ..., PIAverage value.
In step S23, according to data filtering relevant information and HMM, the observation at each moment is judged
Whether status switch corresponds to abnormal driving state.Specifically, according to step S21 and S22, Hidden Markov mould is had been obtained for
The parameter of type, and the value of the observer state chain corresponding to the automatic Pilot training data of current desired processing, therefore can root
According to the value and HMM of acquired observer state chain, driving behavior corresponding to the observer state at each moment is predicted
State is normal driving state or abnormal driving state.The algorithm of prediction can use approximate data or viterbi algorithm.Make
It is unrestricted in an embodiment of the present invention with which kind of prediction algorithm.
In a further advantageous embodiment, judge whether abnormal step S2 also includes step S24 (in figure not for driving behavior
Mark):Whether the judged result of abnormal driving state is corresponded to according to the observer state sequence at each moment, updates Hidden Markov
The parameter of model.Driven corresponding to obtaining the observer state sequence at each moment according to step S21, step S22 and step S23
After the result for sailing behavior state, by the value of the result and the observer state sequence at each moment together, driving row is applied to
In the HMM of state, to obtain the value of this HMM parameter, and by new Hidden Markov mould
Shape parameter and former HMM parameter are handled together, such as are averaged, and replace original with the value after processing
HMM parameter.
The electronic equipment of embodiments of the invention, HMM has been used to filter out in automatic Pilot training data
Unstable data.This method is the innovation that abnormal data of the HMM in automatic Pilot training data filters out
The application of property.HMM is creatively applied the technical scheme filtered out in automatic Pilot training data by this method
On, existing training data can be not only made full use of, obtains more accurately filtering out result, and can be according to newest instruction
Practice the continuous iteration renewal of data, to adapt to the automatic Pilot scene and environment of more complexity.
Fig. 5 is that the preferred embodiments of the present invention by traffic signal information judge that the whether abnormal flow of driving behavior is shown
It is intended to.As shown in figure 5, in the preferred embodiment, step S2 includes step S41, step S42 and step S43.In step S41
In, obtain traffic signal information.The method for obtaining traffic signal information can be a variety of.Can by image recognition technology,
Traffic signal information is determined according to image recognition technology from the image information of acquisition.Traffic can be obtained with traffic scheduling center
The data of signal message.In a preferred embodiment, step S41 includes receiving the wireless communication that traffic signals Warning Mark is sent
Number, obtain traffic signal information.Specifically, wireless base station apparatus is installed on traffic signals Warning Mark, by traffic signals
Result or the information such as changing rule and traffic signals at the time of point and position sent out by wireless signal.Here
Wireless base station apparatus is not limited to which kind of wireless transmit-receive technology and wireless communication protocol or message format used.The embodiment of the present invention
Electronic equipment the radio receiver to match with transmitting portion can be installed, directly receive traffic signals Warning Mark send
Traffic signal information.Or the information that other electronic equipment elder generation wireless receiving traffic signals Warning Marks are sent, then this hair
The electronic equipment of bright embodiment is communicated with other electronic equipments by wirelessly or non-wirelessly mode, finally obtains traffic signal information.
In step S42, the driving row at corresponding moment is judged according to data filtering relevant information and traffic signal information
Whether to violate the traffic regulations.Specifically, the cartographic information in data filtering relevant information, GPS information, and traffic
Signal message, it is known that crossing where the traffic sign near vehicle position and vehicle region, such as vehicle
The traffic speed limit at crossing where the traffic lights situation or vehicle of position or traffic are passed through the requirement in direction, therefore can be according to being obtained
The road signs information that takes, data filtering relevant information together, judge whether vehicle has and violate traffic lights mark, hypervelocity or traveling
In the behavior of the first-class illegal traffic rules of incorrect road.
In step S43, it is determined that driving behavior when violating the traffic regulations is abnormal driving behavior.According to step S42's
Judged result, if driving behavior does not have illegal traffic rules, automatic Pilot training data need not be handled.If
The illegal traffic rules of driving behavior, then be defined as driving behavior now the abnormal driving behavior of automatic Pilot training.
In another preferred embodiment of the invention, step S2 includes:When determining the stationary vehicle before vehicle start
Driving behavior is abnormal driving behavior.When vehicle waits red light, or vehicle stops before the traffic sign of " stopping ", or because of rule
Keep away after pedestrian or barrier when stopping vehicle, vehicle remains static;After the condition that stationary vehicle waits disappears, vehicle
There is one from static to the process of startup.Because driver's reaction time in itself is different, therefore vehicle can be with from conditions permit
The time for starting to start between the real startup of vehicle is different.So as to reflect in automatic Pilot training data, during stationary vehicle
The length of corresponding training data is different.Data during this section of stationary vehicle are useless numbers for automatic Pilot training
According to, because autopilot facility has the processing reaction time of oneself in itself, the vehicle before this section of vehicle start in training data
Corresponding automatic Pilot training data is the abnormal data for needing to filter out when static, it is thus determined that the stationary vehicle before vehicle start
When driving behavior be abnormal driving behavior.
In step s3, the automatic Pilot training data corresponding to Exception Filter driving behavior.Specifically, in step S2
In have determined that automatic Pilot training abnormal driving behavior, when also specify that corresponding to these abnormal driving behaviors
Carve, therefore the namely unwanted automatic Pilot training data of institute of automatic Pilot training data corresponding to these moment.Filtering
Mode is that the training data that these abnormal driving behaviors correspond to the moment is directly deleted from original training data, or with not influenceing
The data of training replace.
Fig. 6 is the schematic flow sheet of the device of the processing automatic Pilot training data of embodiments of the invention.The present invention's
The device of embodiment is used to handle automatic Pilot training data, can apply on an electronic device.Electronic equipment includes but unlimited
In computer equipment and vehicle electronics.Computer equipment refers to that numerical value can be performed by running preset program or instruction
The intelligent electronic device of the predetermined process process such as calculating and/or logical calculated, it can include processor and memory, by handling
Device performs the survival to prestore in memory and instructed to perform predetermined process process, or is held by hardware such as ASIC, FPGA, DSP
Row predetermined process process, or realized by said two devices combination.Computer equipment includes but is not limited to server, personal electricity
Brain, notebook computer, tablet personal computer, smart mobile phone etc..Server include but is not limited to individual server, multiple server groups into
Or the cloud being made up of a large amount of computers or server based on cloud computing, wherein, cloud computing is one kind of Distributed Calculation, by one
One super virtual computer of the computer collection composition of group's loose couplings.Vehicle electronics are to be used on automobile or and vapour
The related electronic equipment of car.
As shown in fig. 6, included according to the device of the processing automatic Pilot training data of the present embodiment:Data capture unit
51st, abnormal deciding means 52 and data filtering units 53.
First, data capture unit 51 is configured to obtain data filtering relevant information and the automatic Pilot training at multiple moment
Data.Automatic Pilot training data refers to the data trained or learnt for automatic Pilot.These data can be that driver drives
The data that navigation maps such as each inductor, monitor or the radar in vehicle processes on vehicle etc. gather together are sailed, can
To be each inductor, monitor or radar in drive simulating environment in caused data or software or database
Caused data.Data filtering relevant information refers to be used to filter out to cause automatic Pilot unstable in automatic Pilot training data
Training data required for information, including following information:Gps data, cartographic information, vehicle control parameters information.Gps data
Refer to and the data message acquired in the satellite communication of NAVSTAR, including the positional information of time and vehicle.
In the present invention, gps data is not limited to the data obtained from GPS (Global Position System, global positioning system)
Information, other it is similar from Beidou navigation satellite system, Galilean satellite alignment system or it is existing at present or it is following occur lead
Time for obtaining and position etc., information was such as applicable to the present invention in boat alignment system, should also be included in the scope of the present invention with
It is interior, and be incorporated herein by reference.Cartographic information refers to the data message obtained from navigation map, including at least there is vehicle
The position of the road of driving, road shape, road attribute, road mounting situation, road junction, the information such as direction of passing through.
Vehicle control parameters information refers to when driver drives vehicle to control operations such as the at the uniform velocity traveling of vehicle, acceleration, braking, steerings
The data of caused control parameter, such as the steering wheel angle of the speed of vehicle, vehicle, vehicle turn signal etc..Data filter out
Gps data and cartographic information in relevant information can obtain from the digital map navigation equipment with GPS receiver device, vehicle control
Parameter information processed can be obtained by the inductor installed on car, or be believed by the vehicle driving trace and map at each moment
Breath etc. together, obtains after computing.Such as, it is known that the position of the vehicle at each moment, according to the map information it is known that
Vehicle between two moment is all straight-line travelling and knows the distance between two moment points of vehicle, it is hereby achieved that car
Speed.
Data filter out relevant information and automatic Pilot training data carries multiple timing point informations, and each moment point is corresponding
One group of data filters out relevant information and automatic Pilot training data.These timing point informations can be gps time or during from GPS
Between at the time of point information or other times source at the time of point information.These moment points can be that time interval is fixed,
As 1/40s or time interval it is unfixed, it is scattered, such as:In one section of driving locus, starting or stop segment process
At the time of dot frequency be 25 times per second, the middle frequency at the uniform velocity travelling section is 5 times per second.
Automatic Pilot training data and data, which filter out relevant information, can include the driving data of one section of driving locus, also may be used
To include the driving data of multistage driving locus.The data information content that automatic Pilot training data includes is a lot.It is if automatic
Include the data that data filter out relevant information in driver training data, that data filters out relevant information and can be from automatic
Driver training data.
Abnormal deciding means 52 is configured to, according to data filtering relevant information, judge the automatic Pilot training number at each moment
It is whether abnormal according to represented driving behavior.The unstable training data of automatic Pilot training does not refer to data mutation or not conformed to
The abnormal data of reason, such as occur a speed very little or very big value suddenly in the training data that speed is stablized always, because
There can not possibly be so violent change for speed, therefore this kind of data are the wrong data occurred in data acquisition or do not conformed to
The abnormal data of reason, can be by eliminating to modes such as data filterings.But this data are not to be wished in embodiments of the invention
Hope the training data filtered out.What is filtered out in embodiments of the invention is the automatic Pilot training data for making automatic Pilot unstable.
For example, automatic Pilot training data be the vehicle directly gathered it is actual drive data when, driver is carried out in driving procedure
Refuel or think that stop off is rested, therefore the off-line training data of vehicle occurs in training data;If driver and its
Other people make a phone call or chatted, it may be noted that power is not concentrated so that the training data of emergency brake occurs in driving data.If these data with
If the deep learning of automatic Pilot or training, it will cause when Function for Automatic Pilot activates, may be in automatic Pilot mistake
Occur emergency brake or off-line situation in journey.Therefore need to try every possible means to filter out, to improve the stability of automatic Pilot.
Specifically, abnormal deciding means 52 is to filter out relevant information according to the data at each moment, to the moment pair
The driving behavior of the driver answered judged, judge driving behavior whether be automatic Pilot training abnormal driving behavior.From
The abnormal driving behavior of dynamic driver training refers to influential on automatic Pilot, makes the driving behavior that automatic Pilot is unstable.From
The abnormal driving behavior of dynamic driver training includes:Abnormal acceleration, abnormal brake, exception are turned to, drifted off the course.Here it is " abnormal
Exception in the abnormal driving behavior such as acceleration ", " abnormal brake ", " abnormal to turn to " refers to not be suitable for what automatic Pilot was trained
Abnormal driving behavior.Judge that the mode of driving behavior includes:Data can be filtered out the figure that relevant information is input to data
In shape simulation softward, by the amplitude of each data of figure shows vehicle control parameters information, when each data former and later two
The change of moment amplitude exceedes threshold value, if now being shown in automatic Pilot training data, environment where vehicle is without exception, corresponding
Driving behavior be defined as abnormal driving behavior.It can in addition contain by manually combine automatic Pilot training data content,
The change of the amplitudes of each data in vehicle control parameters information is manually marked more than the abnormity point of threshold value, marked out pair
The driving behavior for answering the moment is abnormal driving behavior.
Fig. 7 is according to an embodiment of the invention to judge the whether abnormal exception of driving behavior by HMM
The schematic diagram of judging unit.In a preferred embodiment, abnormal deciding means 52 establishes module 61, model including observer state
Module 62 and condition judgment module 63 are established, as shown in Figure 7.
Observer state establishes the observation shape that module 61 is configured to establish the driving behavior defined based on HMM
State sequence.The observer state sequence of driving behavior refers to the state that can be observed in HMM, in the present invention
Observable state by driving behavior as HMM.The observer state sequence of driving behavior includes following Observable
State:Normally travel state, anxious acceleration mode, state of bringing to a halt, state of taking a sudden turn, state of drifting off the course.To data filtering phase
Close information to be sampled by the moment, the vehicle control information and cartographic information at corresponding each moment can be obtained.According to vehicle control
Information processed and cartographic information can obtain the Observable state of the driving behavior at each moment.Such as:According to vehicle control information
It is known that the speed v1 of moment t1 vehicle, steering wheel for vehicle corner k1, then according to the speed v0 of previous moment vehicle, car
Increased acceleration a1=(v1-v0)/(t1-t0) of speed, if a1 value has exceeded default acceleration threshold value, when current
The observer state at quarter is anxious acceleration mode;Acceleration b1=(v0-v1)/(t1-t0) that the speed of vehicle reduces, if b1 value surpasses
Default deceleration threshold value is crossed, then the observer state at current time is state of bringing to a halt;According to the steering wheel of previous moment vehicle
Corner k0, amplitude of variation c1=(k1-k0)/(t1-t0) of the angle of the steering wheel rotation of vehicle, presets if c1 value has exceeded
Corner threshold value, then the observer state at current time is zig zag state;Information is known that vehicle driving trace according to the map
Position, if the distance that the vehicle driving trace at current time differs with the position of the vehicle driving trace of detachment on cartographic information
Setover threshold value, then the observer state at current time is state of drifting off the course;If vehicle is not at anxious acceleration mode, brought to a halt
State, zig zag state and drift off the course state when, it is believed that the observer state at current time is normal vehicle operation state.Can
Each vehicle running state of each moment point forms Observable state chain by more than.Such as the observation shape of the driving behavior at moment 1
State is normal vehicle operation state, and the observer state at moment 2 is vehicle zig zag state, and the observer state at moment 3 is that vehicle is inclined
From course line state, then normal vehicle operation state->Vehicle zig zag state->The vehicle state of drifting off the course constitutes moment 1->
The Observable state chain at moment 3.Above Observable state can be encoded, in order to realize and understand.Need what is illustrated
It is to establish module above with respect to the observer state for the Observable state for obtaining driving behavior and be only for example, other are existing or modern
The observer state of the Observable state for the acquisition driving behavior being likely to occur afterwards establishes module and is such as applicable to the present invention, should also wrap
It is contained within the scope of the present invention, and is incorporated herein by reference.
Model building module 62 is configured to establish the HMM of driving behavior state, driving behavior state bag
Include:Normal driving state and abnormal driving state.Normal driving state and abnormal driving state are in HMM
Two hidden states.The process for establishing the HMM of driving behavior state is mainly to determine HMM
Parameter, i.e. probability vector, state transition probability matrix and observation probability matrix.Model building module 62 can use artificial
The mode of labeled data determines HMM parameter.The mode of this artificial labeled data needs substantial amounts of artificial mark
Note training data.
Fig. 8 is that the model of the HMM for establishing driving behavior state of the preferred embodiments of the present invention establishes mould
The schematic diagram of block.In the preferred embodiment, HMM is established by way of model learning or model training.Mould
Type, which establishes module 62, includes Sample Storehouse setting up submodule 71, first state setting up submodule 72 and model training submodule 73, such as
Shown in Fig. 8.
Sample Storehouse setting up submodule 71 is configured to establish the Sample Storehouse of HMM.The sample of HMM
This storehouse is the sample database required for hidden Ma Erke model learnings, including one or more driving locus sections were believed with the moment
The data of breath filter out relevant information.
The data filtering relevant information that first state setting up submodule 72 is configured in Sample Storehouse determines each moment
HMM training observer state sequence.Because the Sample Storehouse of model, which includes multiple segment data, filters out related letter
Breath, therefore relevant information is filtered out to the data of each driving locus section handled to obtain corresponding to multiple moment points and observe shape
State sequence.The observer state sequence of multiple moment points of one driving locus section is connected as an observer state sequence chain.
The length of observer state chain can be the number of the whole moment points included in one section of driving locus or according to equipment or
The value that system capability determines.The length of Observable state chain is unrestricted in an embodiment of the present invention.
Model training submodule configuration 73 is the data filtering relevant information and HMM in Sample Storehouse
The observer state sequence of training, is trained to HMM, determines the parameter of HMM.Determine hidden horse
The parameter of Er Kefu models namely determines probability vector, state transition probability matrix and the observation probability matrix of model.
Probability vector is in normal driving state including initial time and initial time is in abnormal driving shape probability of state.State
Transition probability matrix contains the probability mutually shifted between normal driving state and abnormal driving state.Observation probability matrix bag
The probability that each observer state is generated under normal driving state and abnormal driving state is contained.Model training submodule 73 is by sample
The data of one section of driving locus in storehouse filter out relevant information to be trained to HMM:In the driving locus section
In, first state setting up submodule 72 first obtains the value of the observer state sequence at each moment, for example, according to vehicle control parameters
It is normally travel state to determine the moment 1, and the moment 2 is acceleration mode, and the moment 3 is normally travel state, and the moment 4 is brake shape
State ...;Then model training submodule 73 estimates that the parameter of HMM is (initial according to data filtering relevant information
Probability vector, state transition probability matrix and observation probability matrix) so that the observation sequence obtained under the model parameter it is general
Rate is maximum.Maximum-likelihood method can be used by estimating the algorithm of the parameter of HMM, can also use Baum-Welch
Algorithm or other algorithms are trained study.Estimate HMM parameter specific algorithm in the present invention not by
Limit.The data of multiple driving locus in Sample Storehouse are filtered out relevant information and all use above method respectively by model training submodule 73
The HMM of driving behavior state is trained.The parameter for the HMM that each training is obtained is entered
Row processing, it is determined that the parameter value of final HMM.For example, every hop count is obtained according to multistage model training data
According to normal driving state to normal driving state transition probability P1, P2, P3... ..., PI(I represents the total of model training data
Hop count), then the transition probability P of the normal driving state in end-state transition probability matrix to normal driving state is P1, P2,
P3... ..., PIAverage value.
Condition judgment module 63 is configured to according to data filtering relevant information and HMM, when judging each
Whether the observer state sequence at quarter corresponds to abnormal driving state.Specifically, module 61 is established according to observer state and model is built
Formwork erection block 62 has been obtained for the parameter of HMM, and the automatic Pilot training data institute of current desired processing is right
The value for the observer state chain answered, therefore condition judgment module 63 is according to the value and Hidden Markov mould of acquired observer state chain
Type, driving behavior state corresponding to the observer state at each moment of predicting are normal driving state or abnormal driving state.In advance
The algorithm of survey can use approximate data or viterbi algorithm.Using which kind of prediction algorithm in an embodiment of the present invention not by
Limit.
In a further advantageous embodiment, abnormal deciding means 52 also includes parameter update module 64 (not marked in figure),
It is configured to whether correspond to the judged result of abnormal driving state according to the observer state sequence at each moment, updates Hidden Markov
The parameter of model.Module 61, model building module 62 and condition judgment module 63, which are established, in observer state obtains each moment
Observer state sequence corresponding to driving behavior state result after, parameter update module 64 is by the result and each moment
Observer state sequence value together, be applied in the HMM of driving behavior state, obtain this hidden Ma Erke
The value of husband's model parameter, and new HMM parameter and former HMM parameter are handled together,
Such as average, and replace former HMM parameter with the value after processing.
The device of embodiments of the invention, HMM has been used to filter out in automatic Pilot training data not
Stable data.It is innovative that the device is that abnormal data of the HMM in automatic Pilot training data filters out
Using.The device creatively applies HMM in the technical scheme filtered out of automatic Pilot training data,
Existing training data can be not only made full use of, obtains more accurately filtering out result, and can be according to newest training
The continuous iteration renewal of data, to adapt to the automatic Pilot scene and environment of more complexity.
Fig. 9 is that the preferred embodiments of the present invention by traffic signal information judge that the whether abnormal exception of driving behavior is sentenced
The schematic diagram of disconnected unit.As shown in figure 9, in the preferred embodiment, abnormal deciding means 52 includes traffic signals acquisition module
81st, 82 and first abnormal judge module 8 of traffic rules judge module.Traffic signals acquisition module 81 is configured to obtain traffic signals
Information.The method for obtaining traffic signal information can be a variety of.Traffic signals acquisition module 81 can pass through image recognition skill
Art, traffic signal information is determined according to image recognition technology from the image information of acquisition;It can also be obtained with traffic scheduling center
Take the data of traffic signal information.In a preferred embodiment, traffic signals acquisition module 81 also includes signal reception submodule
The (not shown) of block 84, signal receiving submodule 84 are configured to receive the wireless signal that traffic signals Warning Mark is sent, obtained
Take traffic signal information.Specifically, wireless base station apparatus is installed on traffic signals Warning Mark, by the result of traffic signals
Or the information such as changing rule and traffic signals at the time of point and position is sent out by wireless signal.Here wireless hair
Send device is not limited to which kind of wireless transmit-receive technology and wireless communication protocol or message format used.Traffic signals acquisition module 81
The radio receiver to match with transmitting portion can be installed, directly receive the traffic signals that traffic signals Warning Mark is sent
Information.Or the information that other electronic equipment elder generation wireless receiving traffic signals Warning Marks are sent, then traffic signals obtain mould
Block 81 is communicated with other electronic equipments by wirelessly or non-wirelessly mode, finally obtains traffic signal information.
Traffic rules judge module 82 is configured to judge to correspond to according to data filtering relevant information and traffic signal information
Whether the driving behavior at moment violates the traffic regulations.Specifically, traffic rules judge module 82 is believed according to data filtering is related
Cartographic information, GPS information in breath, and traffic signal information, it is known that vehicle position and vehicle region
The traffic speed limit at crossing or traffic where the traffic lights situation of crossing position where neighbouring traffic sign, such as vehicle or vehicle
The requirement in current direction, therefore vehicle can be judged according to acquired road signs information, data filtering relevant information together
Whether the behavior of violating traffic lights mark, hypervelocity or traveling in incorrect road first-class illegal traffic rules is had.
Driving behavior when first abnormal judge module 83 is configured to determine to violate the traffic regulations is abnormal driving behavior.Root
, need not be to driving automatically if driving behavior does not have illegal traffic rules according to the judged result of traffic rules judge module 82
Training data is sailed to be handled.If the illegal traffic rules of driving behavior, the first abnormal judge module 83 is driving now
The behavior of sailing is defined as the abnormal driving behavior of automatic Pilot training.
In another preferred embodiment of the invention, abnormal deciding means 52 is configured to determine the vehicle before vehicle start
Driving behavior when static is abnormal driving behavior.When vehicle waits red light, or vehicle stops before the traffic sign of " stopping "
When only, or because evading stopping after pedestrian or barrier vehicle, vehicle remains static;Disappear in the condition that stationary vehicle waits
After mistake, vehicle has one from static to the process of startup.Because driver's reaction time in itself is different, therefore vehicle is from condition
Allow that the time difference started between the real startup of vehicle can be started.So as to reflect in automatic Pilot training data, car
The length of corresponding training data is different when static.Data during this section of stationary vehicle are nothings for automatic Pilot training
Data, because autopilot facility has the processing reaction time of oneself in itself, before this section of vehicle start in training data
Stationary vehicle when corresponding automatic Pilot training data be to need the abnormal data that filters out, therefore abnormal deciding means 52 determines
Driving behavior during stationary vehicle before vehicle start is abnormal driving behavior.
Data filtering units 53 are configured to the automatic Pilot training data corresponding to Exception Filter driving behavior.Specifically
Say, abnormal deciding means 52 has determined that the abnormal driving behavior of automatic Pilot training, also specify that these exceptions are driven
At the time of the behavior of sailing corresponds to, therefore the unwanted automatic Pilot instruction of the namely institute of automatic Pilot training data corresponding to these moment
Practice data.The mode that data filtering units 53 cross filter data is that these abnormal driving behaviors are directly deleted from original training data
The training data at corresponding moment, or replaced with the data for not influenceing training.
It should be noted that the present invention can be carried out in the assembly of software and/or software and hardware, for example, this hair
Bright each device can using application specific integrated circuit (ASIC) or any other realized similar to hardware device.In one embodiment
In, software program of the invention can realize steps described above or function by computing device.Similarly, it is of the invention
Software program (including related data structure) can be stored in computer-readable medium.Computer-readable medium can be
Computer-readable signal media or computer-readable recording medium.Computer-readable recording medium for example can be --- but not
Be limited to --- electricity, magnetic, optical, electromagnetic, system, device or the device of infrared ray or semiconductor, or it is any more than combination.Meter
The more specifically example (non exhaustive list) of calculation machine readable storage medium storing program for executing includes:Electrical connection with one or more wires,
Portable computer diskette, hard disk, random access memory (RAM), read-only storage (ROM), erasable type may be programmed read-only deposit
Reservoir (EPROM or flash memory), optical fiber, portable compact disc read-only storage (CD-ROM), light storage device, magnetic memory
Part or above-mentioned any appropriate combination.In this document, computer-readable recording medium can any be included or store
The tangible medium of program, the program can be commanded the either device use or in connection of execution system, device.
Computer-readable signal media can include in a base band or as carrier wave a part propagation data-signal,
Wherein carry computer-readable program code.The data-signal of this propagation can take various forms, including --- but
It is not limited to --- electromagnetic signal, optical signal or above-mentioned any appropriate combination.Computer-readable signal media can also be
Any computer-readable medium beyond computer-readable recording medium, the computer-readable medium can send, propagate or
Transmit for by instruction execution system, device either device use or program in connection.
The program code included on computer-readable medium can be transmitted with any appropriate medium, including --- but it is unlimited
In --- wireless, electric wire, optical cable, RF etc., or above-mentioned any appropriate combination.
It can be write with one or more programming languages or its combination for performing the computer that operates of the present invention
Program code, described program design language include object oriented program language-such as Java, Smalltalk, C++,
Also include conventional procedural programming language-such as " C " language or similar programming language.Program code can be with
Fully perform, partly perform on the user computer on the user computer, the software kit independent as one performs, portion
Divide and partly perform or performed completely on remote computer or server on the remote computer on the user computer.
Be related in the situation of remote computer, remote computer can pass through the network of any kind --- including LAN (LAN) or
Wide area network (WAN)-be connected to subscriber computer, or, it may be connected to outer computer (such as carried using Internet service
Pass through Internet connection for business).
In addition, the present invention some steps or function can employ hardware to realize, for example, as with processor coordinate so as to
Perform the circuit of each step or function.
It is obvious to a person skilled in the art that the invention is not restricted to the details of above-mentioned one exemplary embodiment, Er Qie
In the case of without departing substantially from spirit or essential attributes of the invention, the present invention can be realized in other specific forms.Therefore, no matter
From the point of view of which point, embodiment all should be regarded as exemplary, and be nonrestrictive, the scope of the present invention is by appended power
Profit requires rather than described above limits, it is intended that all in the implication and scope of the equivalency of claim by falling
Change is included in the present invention.Any reference in claim should not be considered as to the involved claim of limitation.This
Outside, it is clear that the word of " comprising " one is not excluded for other units or step, and odd number is not excluded for plural number.That is stated in system claims is multiple
Unit or device can also be realized by a unit or device by software or hardware.The first, the second grade word is used for table
Show title, and be not offered as any specific order.
Claims (19)
1. a kind of method for handling automatic Pilot training data, including:
A. the data filtering relevant information at multiple moment and the automatic Pilot training data are obtained;
B. according to the data filtering relevant information, the driving behavior represented by the automatic Pilot training data at each moment is judged
It is whether abnormal;
C. the automatic Pilot training data corresponding to Exception Filter driving behavior.
2. according to the method for claim 1, wherein, the step b includes:
The observer state sequence for the driving behavior that-foundation is defined based on HMM;
- HMM of driving behavior state is established, the driving behavior state includes:Normal driving state and exception
Driving condition;
- according to the data filtering relevant information and the HMM, judge the observation shape at each moment
Whether state sequence corresponds to the abnormal driving state.
3. according to the method for claim 2, wherein, it is described establish driving behavior state HMM the step of
Including:
- establish the Sample Storehouse of the HMM;
- data filtering the relevant information in the Sample Storehouse determines the HMM training at each moment
Observer state sequence;
- data filtering the relevant information in the Sample Storehouse and the observer state of HMM training
Sequence, the HMM is trained, determines the parameter of HMM.
4. according to the method for claim 3, wherein, the step b also includes:
- judged result of the abnormal driving state whether is corresponded to according to the observer state sequence at each moment, update institute
State the parameter of HMM.
5. the method according to any one of claim 2 to 4, wherein, the observer state sequence includes:Normally travel shape
State, anxious acceleration mode, state of bringing to a halt, state of taking a sudden turn, state of drifting off the course.
6. according to the method for claim 1, wherein, the step b includes:
- obtain traffic signal information;
Whether-the driving behavior for judging to correspond to the moment according to the data filtering relevant information and the traffic signal information is disobeyed
Anti- traffic rules;
Driving behavior when-determination violates the traffic regulations is abnormal driving behavior.
7. according to the method for claim 6, wherein, described the step of obtaining traffic signal information, also includes:
- wireless signal that traffic signals Warning Mark is sent is received, obtain the traffic signal information.
8. according to the method for claim 1, wherein, the step b includes:
- driving behavior when determining the stationary vehicle before vehicle start is abnormal driving behavior.
9. a kind of device for handling automatic Pilot training data, wherein, described device includes:
- data capture unit, it is configured to obtain the data filtering relevant information at multiple moment and the automatic Pilot training data;
- abnormal deciding means, it is configured to, according to the data filtering relevant information, judge the automatic Pilot training number at each moment
It is whether abnormal according to represented driving behavior;
- data filtering units, it is configured to the automatic Pilot training data corresponding to Exception Filter driving behavior.
10. device according to claim 9, wherein, the abnormal deciding means includes:
- observer state establishes module, is configured to establish the observer state sequence of the driving behavior defined based on HMM
Row;
- model building module, it is configured to establish the HMM of driving behavior state, the driving behavior state bag
Include:Normal driving state and abnormal driving state;
- condition judgment module, it is configured to according to the data filtering relevant information and the HMM, judges every
Whether the observer state sequence at individual moment corresponds to the abnormal driving state.
11. device according to claim 10, wherein, the model building module also includes:
- Sample Storehouse setting up submodule, it is configured to establish the Sample Storehouse of the HMM;
- first state setting up submodule, the data filtering relevant information being configured in the Sample Storehouse determine each
The observer state sequence of the HMM training at moment;
- model training submodule, the data filtering relevant information being configured in the Sample Storehouse and the hidden Ma Er
The observer state sequence of section's husband's model training, is trained to the HMM, determines HMM
Parameter.
12. device according to claim 11, wherein, the abnormal deciding means also includes:
- parameter update module, it is configured to whether correspond to the abnormal driving shape according to the observer state sequence at each moment
The judged result of state, update the parameter of the HMM.
13. according to the device described in any one of claim 10 to 12, wherein, the observer state sequence includes:Normally travel
State, anxious acceleration mode, state of bringing to a halt, state of taking a sudden turn, state of drifting off the course.
14. device according to claim 9, wherein, the abnormal deciding means includes:
- traffic signals acquisition module, it is configured to obtain traffic signal information;
- traffic rules judge module, it is configured to according to the data filtering relevant information and traffic signal information judgement
Whether the driving behavior at corresponding moment violates the traffic regulations;
- the first abnormal determining module, driving behavior when being configured to determine to violate the traffic regulations is abnormal driving behavior.
15. device according to claim 14, wherein, the traffic signals acquisition module also includes:
- signal receiving submodule, it is configured to receive the wireless signal that traffic signals Warning Mark is sent, obtains the traffic signals
Information.
16. device according to claim 9, wherein, the abnormal deciding means includes:
- the second abnormal determining module, be configured to determine vehicle start before stationary vehicle when driving behavior be abnormal driving row
For.
17. a kind of computer equipment, the computer equipment includes:
- one or more processors;
- memory, for storing one or more programs,
- when one or more of programs are by one or more of computing devices so that one or more of processing
Device performs the method as described in any in claim 1-8.
18. a kind of computer-readable recording medium, the computer-readable recording medium storage has computer code, when the meter
When calculation machine code is performed, the method as any one of claim 1 to 8 is performed.
19. a kind of computer program product, when the computer program product is performed by computer equipment, such as claim 1
It is performed to the method any one of 8.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710792053.5A CN107784709A (en) | 2017-09-05 | 2017-09-05 | The method and apparatus for handling automatic Pilot training data |
PCT/CN2018/093353 WO2019047593A1 (en) | 2017-09-05 | 2018-06-28 | Method and device for processing automatic driving training data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710792053.5A CN107784709A (en) | 2017-09-05 | 2017-09-05 | The method and apparatus for handling automatic Pilot training data |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107784709A true CN107784709A (en) | 2018-03-09 |
Family
ID=61437564
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710792053.5A Pending CN107784709A (en) | 2017-09-05 | 2017-09-05 | The method and apparatus for handling automatic Pilot training data |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN107784709A (en) |
WO (1) | WO2019047593A1 (en) |
Cited By (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108446727A (en) * | 2018-03-09 | 2018-08-24 | 上海安亭地平线智能交通技术有限公司 | Driving behavior decision-making technique, system and electronic equipment |
WO2019047593A1 (en) * | 2017-09-05 | 2019-03-14 | 百度在线网络技术(北京)有限公司 | Method and device for processing automatic driving training data |
CN109979193A (en) * | 2019-02-19 | 2019-07-05 | 中电海康集团有限公司 | A kind of data exception diagnostic method based on Markov model |
TWI668141B (en) * | 2018-11-07 | 2019-08-11 | 國家中山科學研究院 | Virtual thermal image driving data generation system |
CN110276952A (en) * | 2019-06-27 | 2019-09-24 | 武汉中海庭数据技术有限公司 | A kind of traffic information analog acquisition method and device |
CN110558934A (en) * | 2019-08-02 | 2019-12-13 | 杭州连帆科技有限公司 | sleep quality evaluation method and device, electronic equipment and medium |
CN110660217A (en) * | 2019-09-27 | 2020-01-07 | 百度在线网络技术(北京)有限公司 | Method and device for detecting information security |
CN110806744A (en) * | 2018-07-19 | 2020-02-18 | 通用汽车环球科技运作有限责任公司 | Intersection autonomous driving decision using hierarchical option Markov decision process |
CN111047047A (en) * | 2019-11-15 | 2020-04-21 | 奇点汽车研发中心有限公司 | Driving model training method and device, electronic equipment and computer storage medium |
CN111252061A (en) * | 2018-11-30 | 2020-06-09 | 百度(美国)有限责任公司 | Real-time decision making for autonomous vehicles |
CN111401564A (en) * | 2019-01-02 | 2020-07-10 | 北京地平线信息技术有限公司 | Model updating method and device for machine learning, electronic equipment and storage medium |
CN111540079A (en) * | 2020-04-02 | 2020-08-14 | 东软睿驰汽车技术(沈阳)有限公司 | Method and device for evaluating driving state, electronic device, and storage medium |
CN111767933A (en) * | 2019-05-17 | 2020-10-13 | 北京京东尚科信息技术有限公司 | Method and device for identifying vehicle driving state |
CN112061139A (en) * | 2020-09-03 | 2020-12-11 | 三一专用汽车有限责任公司 | Automatic driving control method, automatic driving device and computer storage medium |
CN113168176A (en) * | 2018-10-17 | 2021-07-23 | 柯尼亚塔有限公司 | System and method for generating realistic simulation data for training automated driving |
CN113903102A (en) * | 2021-10-29 | 2022-01-07 | 广汽埃安新能源汽车有限公司 | Adjustment information acquisition method, adjustment device, electronic device, and medium |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111767837B (en) * | 2020-06-28 | 2023-12-05 | 中国计量大学 | Visual assessment method for valve operation normalization during loading and unloading of tank truck |
CN113232666A (en) * | 2021-03-10 | 2021-08-10 | 北华航天工业学院 | Vehicle driving training system based on driving assistance system |
CN113159410B (en) * | 2021-04-14 | 2024-02-27 | 北京百度网讯科技有限公司 | Training method of automatic control model and fluid supply system control method |
CN113177077B (en) * | 2021-04-29 | 2023-09-26 | 北京百度网讯科技有限公司 | Abnormal event determination method and device for automatic driving and electronic equipment |
CN115195757B (en) * | 2022-09-07 | 2023-08-04 | 郑州轻工业大学 | Electric bus starting driving behavior modeling and recognition training method |
CN116092056B (en) * | 2023-03-06 | 2023-07-07 | 安徽蔚来智驾科技有限公司 | Target recognition method, vehicle control method, device, medium and vehicle |
CN117609737B (en) * | 2024-01-18 | 2024-03-19 | 中国人民解放军火箭军工程大学 | Method, system, equipment and medium for predicting health state of inertial navigation system |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101633359A (en) * | 2008-07-24 | 2010-01-27 | 通用汽车环球科技运作公司 | Adaptive vehicle control system with driving style recognition |
US20140277832A1 (en) * | 2013-03-14 | 2014-09-18 | National University Corporation Nara Institute | System for detecting abnormal driving behavior |
WO2017062106A1 (en) * | 2015-10-05 | 2017-04-13 | Delphi Technologies, Inc. | Humanized steering model for automated vehicles |
CN106740864A (en) * | 2017-01-12 | 2017-05-31 | 北京交通大学 | A kind of driving behavior is intended to judge and Forecasting Methodology |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107784709A (en) * | 2017-09-05 | 2018-03-09 | 百度在线网络技术(北京)有限公司 | The method and apparatus for handling automatic Pilot training data |
-
2017
- 2017-09-05 CN CN201710792053.5A patent/CN107784709A/en active Pending
-
2018
- 2018-06-28 WO PCT/CN2018/093353 patent/WO2019047593A1/en active Application Filing
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101633359A (en) * | 2008-07-24 | 2010-01-27 | 通用汽车环球科技运作公司 | Adaptive vehicle control system with driving style recognition |
US20140277832A1 (en) * | 2013-03-14 | 2014-09-18 | National University Corporation Nara Institute | System for detecting abnormal driving behavior |
WO2017062106A1 (en) * | 2015-10-05 | 2017-04-13 | Delphi Technologies, Inc. | Humanized steering model for automated vehicles |
CN106740864A (en) * | 2017-01-12 | 2017-05-31 | 北京交通大学 | A kind of driving behavior is intended to judge and Forecasting Methodology |
Non-Patent Citations (1)
Title |
---|
卢锡凤: "基于隐马尔可夫模型的城市道路路段不良驾驶行为鉴别", 《中国优秀硕士学位论文全文数据库 工程科技II辑》 * |
Cited By (21)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2019047593A1 (en) * | 2017-09-05 | 2019-03-14 | 百度在线网络技术(北京)有限公司 | Method and device for processing automatic driving training data |
CN108446727A (en) * | 2018-03-09 | 2018-08-24 | 上海安亭地平线智能交通技术有限公司 | Driving behavior decision-making technique, system and electronic equipment |
CN110806744A (en) * | 2018-07-19 | 2020-02-18 | 通用汽车环球科技运作有限责任公司 | Intersection autonomous driving decision using hierarchical option Markov decision process |
CN113168176A (en) * | 2018-10-17 | 2021-07-23 | 柯尼亚塔有限公司 | System and method for generating realistic simulation data for training automated driving |
TWI668141B (en) * | 2018-11-07 | 2019-08-11 | 國家中山科學研究院 | Virtual thermal image driving data generation system |
CN111252061B (en) * | 2018-11-30 | 2023-08-15 | 百度(美国)有限责任公司 | Real-time decision-making for autonomous vehicles |
CN111252061A (en) * | 2018-11-30 | 2020-06-09 | 百度(美国)有限责任公司 | Real-time decision making for autonomous vehicles |
CN111401564A (en) * | 2019-01-02 | 2020-07-10 | 北京地平线信息技术有限公司 | Model updating method and device for machine learning, electronic equipment and storage medium |
CN109979193A (en) * | 2019-02-19 | 2019-07-05 | 中电海康集团有限公司 | A kind of data exception diagnostic method based on Markov model |
CN111767933A (en) * | 2019-05-17 | 2020-10-13 | 北京京东尚科信息技术有限公司 | Method and device for identifying vehicle driving state |
CN110276952A (en) * | 2019-06-27 | 2019-09-24 | 武汉中海庭数据技术有限公司 | A kind of traffic information analog acquisition method and device |
CN110276952B (en) * | 2019-06-27 | 2020-11-27 | 武汉中海庭数据技术有限公司 | Traffic information simulation acquisition method and device |
CN110558934A (en) * | 2019-08-02 | 2019-12-13 | 杭州连帆科技有限公司 | sleep quality evaluation method and device, electronic equipment and medium |
CN110660217B (en) * | 2019-09-27 | 2021-03-16 | 百度在线网络技术(北京)有限公司 | Method and device for detecting information security |
CN110660217A (en) * | 2019-09-27 | 2020-01-07 | 百度在线网络技术(北京)有限公司 | Method and device for detecting information security |
CN111047047A (en) * | 2019-11-15 | 2020-04-21 | 奇点汽车研发中心有限公司 | Driving model training method and device, electronic equipment and computer storage medium |
CN111540079A (en) * | 2020-04-02 | 2020-08-14 | 东软睿驰汽车技术(沈阳)有限公司 | Method and device for evaluating driving state, electronic device, and storage medium |
CN112061139A (en) * | 2020-09-03 | 2020-12-11 | 三一专用汽车有限责任公司 | Automatic driving control method, automatic driving device and computer storage medium |
CN112061139B (en) * | 2020-09-03 | 2022-02-18 | 三一专用汽车有限责任公司 | Automatic driving control method, automatic driving device and computer storage medium |
CN113903102A (en) * | 2021-10-29 | 2022-01-07 | 广汽埃安新能源汽车有限公司 | Adjustment information acquisition method, adjustment device, electronic device, and medium |
CN113903102B (en) * | 2021-10-29 | 2023-11-17 | 广汽埃安新能源汽车有限公司 | Adjustment information acquisition method, adjustment device, electronic equipment and medium |
Also Published As
Publication number | Publication date |
---|---|
WO2019047593A1 (en) | 2019-03-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107784709A (en) | The method and apparatus for handling automatic Pilot training data | |
US10310505B1 (en) | Seamless vehicle entry | |
US11922708B2 (en) | Multiple stage image based object detection and recognition | |
CN109496188B (en) | System and method for switching vehicle driving modes | |
US11127303B2 (en) | Vehicle scheduling method and apparatus, device and storage medium | |
US8781650B2 (en) | Aircraft navigation system | |
CN110418743B (en) | Autonomous vehicle operation management obstruction monitoring | |
CN105513389B (en) | The method and device of augmented reality | |
US20190113351A1 (en) | Turn Based Autonomous Vehicle Guidance | |
CN107810133A (en) | The rollback requests of autonomous vehicle | |
CN114255606A (en) | Auxiliary driving reminding method and device, map auxiliary driving reminding method and device and map | |
US11148668B2 (en) | Autonomous vehicle control for reverse motion | |
US11507090B2 (en) | Systems and methods for vehicle motion control with interactive object annotation | |
US20220017112A1 (en) | Systems and Methods for Generating Vehicle Corridors to Improve Path Planning Efficiency | |
CN113479195A (en) | Method for automatic valet parking and system for carrying out said method | |
US10836405B2 (en) | Continual planning and metareasoning for controlling an autonomous vehicle | |
US11604908B2 (en) | Hardware in loop testing and generation of latency profiles for use in simulation | |
CN109284801A (en) | State identification method, device, electronic equipment and the storage medium of traffic light | |
EP4116945A1 (en) | Vehicle management system | |
CN112363511A (en) | Vehicle path planning method and device, vehicle-mounted device and storage medium | |
Wishart et al. | Fundamentals of Connected and Automated Vehicles | |
CN112677763B (en) | Automatic driving vehicle journey planning processing method and device and vehicle-mounted terminal | |
CN116767182A (en) | Perception-based parking assistance for autonomous machine systems and applications | |
Aulinas et al. | AI for Cars | |
US11975726B1 (en) | Systems and methods for interaction-based trajectory prediction |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180309 |
|
RJ01 | Rejection of invention patent application after publication |