CN110481556A - Vehicle is fled from - Google Patents
Vehicle is fled from Download PDFInfo
- Publication number
- CN110481556A CN110481556A CN201910392564.7A CN201910392564A CN110481556A CN 110481556 A CN110481556 A CN 110481556A CN 201910392564 A CN201910392564 A CN 201910392564A CN 110481556 A CN110481556 A CN 110481556A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- map
- wheel
- computing device
- skidding
- 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
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W40/00—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models
- B60W40/02—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models related to ambient conditions
- B60W40/06—Road conditions
- B60W40/068—Road friction coefficient
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W50/00—Details of control systems for road vehicle drive control not related to the control of a particular sub-unit, e.g. process diagnostic or vehicle driver interfaces
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W50/00—Details of control systems for road vehicle drive control not related to the control of a particular sub-unit, e.g. process diagnostic or vehicle driver interfaces
- B60W2050/0001—Details of the control system
- B60W2050/0019—Control system elements or transfer functions
- B60W2050/0026—Lookup tables or parameter maps
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W50/00—Details of control systems for road vehicle drive control not related to the control of a particular sub-unit, e.g. process diagnostic or vehicle driver interfaces
- B60W2050/0062—Adapting control system settings
- B60W2050/0075—Automatic parameter input, automatic initialising or calibrating means
Landscapes
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Transportation (AREA)
- Mechanical Engineering (AREA)
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Human Computer Interaction (AREA)
- Traffic Control Systems (AREA)
- Control Of Driving Devices And Active Controlling Of Vehicle (AREA)
Abstract
Present disclose provides " vehicle are fled from ".Map including potential barrier can be generated by the computing device in vehicle.The coefficient of friction of the estimation between wheel and the road of stranded vehicle periphery can be generated based on sensing data.Can determine path based on the map, and can be based on the path and based on the wheel and the skidding rate of determination operates the stranded vehicle so that the stranded vehicle is got rid of poverty.
Description
Cross reference to related applications
Present patent application is that Shen is continued in the part for the U.S. Patent Application No. 15/784,432 submitted on October 16th, 2017
Its priority and all advantages please and be required, the mode which quotes in its entirety is incorporated herein.
Technical field
The present invention relates to the maps including potential barrier that can be generated by the computing device in vehicle.It can be based on
Sensing data and the coefficient of friction for generating the estimation between wheel and the road of stranded vehicle periphery.It can be based on the map
And determine path, and can be based on the path and based on the wheel and the skidding rate of determination operates the stranded vehicle
So that the stranded vehicle is got rid of poverty.
Background technique
The autonomous vehicle operation of multiple ranks has been defined in Society of automotive engineers (SAE).At 0 to 2 grade, the mankind are driven
Member usually monitors in the case where the help not from vehicle or controls most of driving task.For example, at 0 grade (" without automatic
Change "), human driver is responsible for all vehicle operatings.At 1 grade (" driver assistance "), vehicle sometimes assisted diversion, accelerate
Or braking, but driver is still responsible for most of vehicle control.2 grades (" partial automations "), vehicle can be certain
In the case of control turn to, accelerate and braking, without human interaction.At 3 to 5 grades, vehicle undertake more to relevant of driving
Business.3 grades (" conditional automations "), vehicle can handle steering, acceleration and braking in some cases, and monitoring is driven
Sail environment.However, 3 grades require driver to intervene once in a while.4 grades (" increasingly automated "), vehicle can handle identical as 3 grades
Task, but independent of the certain driving modes of driver's intervention.At 5 grades (" full automations "), vehicle be can handle almost
All tasks are without any driver's intervention.
Summary of the invention
A kind of method includes generating map including barrier and wheel and stranded vehicle week based on sensing data
The coefficient of friction for the estimation between road enclosed.The method includes determining path based on the map, and based on described
Path and the stranded vehicle is operated based on the skidding control process of the wheel.
The coefficient of friction of the estimation between the wheel and the road can be lower than with empirically determined for allowance institute
State the value of the operation of stranded vehicle.
The method may include based on be parallel to the road wheel movement and wheel rotation come determine skidding rate with
Determine the skidding control process.
The method may include the determination skidding control process, with permit the stranded vehicle be based on the path and
Determining skidding rate is operated, regardless of the coefficient of friction of the estimation.
The method may include the skidding rates of matching wheel to be got rid of poverty with the determination stranded vehicle.
The method may include the generation maps to generate the first map including being based on vehicle sensor data, including
Barrier.
The method may include the generation maps to generate the second map including being based on vehicle sensor data, including
For the stranded vehicle periphery position come estimated friction coefficient.
The method may include be combined first map and second map and be based on the combination
The first map and the second map and determine the path.
It include passive sensor data and active sensor data the method may include vehicle sensor data.
A kind of system includes processor and memory, and the memory includes instruction, and described instruction is waited for by the processor
Execute with: estimating between the road of the map and wheel and stranded vehicle periphery including barrier is generated based on sensing data
The coefficient of friction of meter;Path is determined based on the map;And the skidding control based on the path and based on the wheel
Process operates the stranded vehicle.
The coefficient of friction of the estimation between the wheel and the road can be lower than with empirically determined for allowance institute
State the value of the operation of stranded vehicle.
The system may include based on be parallel to the road wheel movement and wheel rotation come determine skidding rate with
Determine the skidding control process.
The system can include determining that skidding control process, to permit the stranded vehicle based on the path and determination
Skidding rate operated, regardless of the coefficient of friction of the estimation.
The system may include that the skidding rate of matching wheel is got rid of poverty with the determination stranded vehicle.
The system may include generate the map include based on vehicle sensor data and generate the first map, including
Barrier.
The system may include generate the map include based on vehicle sensor data and generate the second map, including
For the stranded vehicle periphery position come estimated friction coefficient.
The system may include being combined first map and second map and based on the combination
The first map and the second map and determine the path.
Vehicle sensor data may include passive sensor data and active sensor data.
A kind of system includes component for obtaining sensing data and for controlling stranded Vehicular turn, braking and power
The component of transmission system.The system comprises the computer components for performing the following operation: based on by for obtaining sensor
The sensing data of the component extraction of data and generate map including barrier and wheel and the stranded vehicle periphery
Road between estimation coefficient of friction;Path is determined based on the map;And based on the path and based on described
Wheel and the skidding rate of determination, by for controlling the stranded Vehicular turn, braking and the component of power drive system
To operate the stranded vehicle so that the stranded vehicle is got rid of poverty.
The coefficient of friction of the estimation between the wheel and the road can be lower than with empirically determined for allowance institute
State the value of the operation of stranded vehicle.
Detailed description of the invention
Fig. 1 shows the example autonomous vehicle with autonomous vehicle controller, as a part of skidding control process, institute
State the track of vehicle that autonomous vehicle controller can be planned for high friction surface.
Fig. 2 is the block diagram for showing the exemplary components of vehicle.
Fig. 3 is the control figure for showing the various operations of autonomous vehicle controller during skidding control process.
Fig. 4 A to Fig. 4 C shows the map generated by autonomous vehicle controller, for developing the vehicle for arriving high friction surface
Track simultaneously avoid the barrier detected.
Fig. 5 is the example mistake for the track of vehicle for being planned for high friction surface that can be executed by autonomous vehicle controller
The flow chart of journey.
Fig. 6 A to Fig. 6 D shows example autonomous vehicle and executes example skidding control process on low-friction surface.
Fig. 7 shows the example vehicle with sensor.
Specific embodiment
Vehicle can be assembled into be operated under both autonomous mode and occupant's driving mode.In addition to discussed above autonomous
Except control hierarchy, semi-autonomous operation mode or entirely autonomous operation mode also can be defined as wherein can be by as having
The computing device of a part of the Vehicle Information System of sensor and controller drives the operation mode of vehicle.Vehicle can be by
It occupies or unoccupied, however, which kind of situation, can drive vehicle in the case where no occupant assists.For the disclosure
Purpose, autonomous mode is defined as wherein vehicle propulsion (for example, via including the dynamic of internal combustion engine and/or electric motor
Power transmission system), braking and each of turn to the mode that is controlled by one or more vehicle computers;In semi-autonomous mode
Under, vehicle computer controls one or both of vehicle propulsion, braking and steering.In non-autonomous vehicle, these not by
Computer control.
A kind of method described herein, the method includes generated based on sensing data the map including barrier with
And the coefficient of friction of the estimation between wheel and the road of stranded vehicle periphery.Path can be determined based on the map, and
And the stranded vehicle can be operated based on the path and based on the skidding control process of the wheel.The wheel and institute
The coefficient of friction for stating the estimation between road can be lower than with the empirically determined value to permit the operation of the stranded vehicle.
Skidding rate can be determined with the determination skidding control process based on wheel movement and the wheel rotation for being parallel to the road.
Determine that the skidding control process can permit the stranded vehicle and be operated based on the path and the skidding rate determined, no
Manage the estimation coefficient of friction how.
The skidding rate of matching wheel can determine that the stranded vehicle is got rid of poverty.Generating the map may include based on vehicle
Sensing data and generate the first map, including barrier.Generating the map includes being generated based on vehicle sensor data
Second map, including the position for the stranded vehicle periphery come estimated friction coefficient.By first map and described the
Two maps be combined can based on the combination the first map and the second map and determine the path.Vehicle sensors number
According to may include passive sensor data and active sensor data.The passive sensor data may include color video number
According to.The active sensor data may include laser radar data or radar data.Active sensor data and described
Passive sensor data can be combined to form the map by rectangular projection.Path based on the map can be with
The position of operation is able to carry out while with certain buffer avoiding obstacles including the determination vehicle.
A kind of computer-readable medium is also disclosed, the computer-readable medium storage is for executing above method step
Some or all of program instruction.A kind of computer is also disclosed, the computer is programmed to for executing above-mentioned side
Some or all of method step, the computer include computer equipment, and the computer equipment is programmed to based on sensing
Device data and the coefficient of friction for generating the estimation between map and wheel and the road of stranded vehicle periphery including barrier.
Can determine path based on the map, and can based on the path and based on the skidding control process of the wheel come
Operate the stranded vehicle.The coefficient of friction of the estimation between the wheel and the road can be lower than with empirically determined
For the value of the operation of the allowance stranded vehicle.It can be determined based on wheel movement and the wheel rotation for being parallel to the road
Skidding rate is with the determination skidding control process.Determine that the skidding control process can permit the stranded vehicle based on described
Path and the skidding rate determined are operated, regardless of the coefficient of friction of the estimation.
The skidding rate that the computer is also programmed to matching wheel is got rid of poverty with the determination stranded vehicle.Generate institute
Stating map may include generating the first map based on vehicle sensor data, including barrier.Generating the map includes base
The second map is generated in vehicle sensor data, including the position for the stranded vehicle periphery come estimated friction coefficient.
By first map and second map be combined can the first map based on the combination and the second map and it is true
The fixed path.Vehicle sensor data may include passive sensor data and active sensor data.The passive sensing
Device data may include color video data.The active sensor data may include laser radar data or radar data.
The active sensor data and the passive sensor data can be combined to be formed by rectangular projection describedly
Figure.Path based on the map can include determining that the vehicle can be into while with certain buffer avoiding obstacles
The position of row operation.
Vehicle can obtain the information in relation to vehicle environmental equipped with computing device, network, sensor and controller simultaneously
And vehicle is operated based on the information.The safety and comfortable operation of vehicle can depend on obtaining about the accurate of vehicle environmental
And timely information.Vehicle sensors can provide the data about the object to be avoided in route and vehicle environmental.Vehicle
Safe and efficient operation may include whether the coefficient of friction between estimated wheel and road is quasi- with the operation for determining vehicle
Perhaps.In the lower example of coefficient of friction between wheel and road, wheel may skid, simultaneously so as to cause vehicle " stranded "
And it therefore can not operate to arrive at the destination.In this illustration, wheel can be based upon including computing device in the car
Determine skidding rate and estimated friction coefficient is lower than with empirically determined value.Vehicle is disapproved according to " logical when coefficient of friction is too low
Often " or non-skidding control means operation to arrive at the destination when, vehicle can be used skidding control process as described herein into
Row is operated to permit the operation of vehicle, although the coefficient of friction of estimation is lower than with empirically determined value.
Computing device in vehicle can be programmed to obtain described in data and the use of the external environment about vehicle
Data come determine for by vehicle from current location operate to the track of destination locations, for example, wherein track be description vehicle
Movement include position, direction and direction and speed change rate vector.For example, the data may include radar,
Laser radar and video sensor, including visible and infrared (IR) sensor.Term " track " herein will be with term " road
Diameter " is used interchangeably.Onboard sensor can be used to generate the real-time map of the position including vehicle periphery and be based in vehicle
The real-time map and planned trajectory, wherein real-time map be defined to include for before map the nearest time (for example,
Several seconds or less) in determine data map.Track may include the vehicle for describing the path that vehicle can operate on it
Position, speed, direction and horizontal and vertical acceleration.
In the example that vehicle is for example trapped in mud or deep snow etc., ambient enviroment is can be used in real time in computing device
Figure come plan will have release vehicle maximum probability and permit vehicle operating from there through vehicle " getting rid of poverty " is made to reach
The track of intended destination, wherein make vehicle get rid of poverty may include execute skidding control process, wherein vehicle steering, braking and
Power drive system is controlled to overcome wheel and make the low friction between the trapped road of vehicle by processor.In other examples
In, the data downloaded via internet from cloud can be used in vehicle, for example, the surface appearance of the position to determine vehicle periphery.
Vehicle can be used about not driving by driving wheel and the skidding rate of wheel " getting rid of poverty " and can weigh when to determine vehicle
Newly start common non-skidding control operation.This can the generation when vehicle reaches the endpoint of its expectation path, wherein by driving
The linear speed of wheel matches its rotation speed, it means that is not skidded by driving wheel.Another describing mode is example
Such as in skidding rate of the skidding rate matching for not driving wheel by driving wheel.
In this example, computing device is programmed to generate the synthetic map of the position including barrier.Computing device can be with
It is programmed to via determining track automatically by automobile navigation to selected high friction surface, while avoiding obstacles.Substitution
Ground or additionally, computing device can be programmed to by generating include that the first map of position of barrier generates synthetically
Figure.In this implementation, computing device can be programmed to by generate the first map with include the footprint of vehicle come
Generate synthetic map.Alternately or in addition, computing device can be programmed to by generating include multiple high friction surfaces
Second map of position generates synthetic map.In the possible method, processor can be programmed to by by the first
It combines to generate synthetic map the part of figure and the second map.By the part of the first map and the second map combination may include by
The position of multiple high friction surfaces from the second map and the barrier from the first map is incorporated into synthetic map.It calculates
Device can be programmed to determine whether vehicle arrived selected high friction surface.It that case, computing device can be with
Be programmed to as determine vehicle have arrived at the result of selected high friction surface and stop execute skidding control process.
As shown in Figure 1, autonomous vehicle 100 includes the computing device 105 for being programmed to control various autonomous vehicle operations.Example
Such as, as follows to be explained in greater detail, computing device 105 is programmed to receiving sensor signal and outputs a signal to throughout vehicle
The various actuators of 100 positioning.By controlling actuator, computing device 105 can automatically provide to the longitudinal direction of vehicle 100 and
Crosswise joint.That is, computing device 105 can control the propulsion, braking and steering of vehicle 100.
In addition, as follows be explained in greater detail, computing device 105 is programmed to the object near detection vehicle 100.Object
It may include other vehicles, pedestrian, road markings, lane markings etc..Computing device 105, which is programmed to detection, has low friction
Surface (referred to as " low-friction surface ", " low-mu surface " or " the low surface mu ") and with height friction (referred to as " height rubs on surface
Wiping surface ", " the high surface μ " or " the high surface mu ").In some cases, it is attached to be programmed to prediction vehicle 100 for computing device 105
The table of near field (region, the subsequent region of vehicle 100 including region, adjacent vehicle 100 in front of vehicle 100 or combinations thereof)
Face friction.In the case where giving the barrier between vehicle 100 and high friction surface, computing device 105 can be programmed to
Exploitation is from low-friction surface to the track of high friction surface.
Although illustrated as car, but vehicle 100 may include any riding or commercial vehicle, such as car, truck,
Sport vehicle, transboundary vehicle, cargo, jubilee wagen, taxi, bus etc..It is discussed in more detail below,
Vehicle 100 is oneself that can be operated under autonomous (for example, unmanned) mode, part autonomous mode and/or non-autonomous mode
Main vehicle.Part autonomous mode can refer to 2 grades of operation modes of SAE, wherein vehicle 100 can control in some cases steering,
Accelerate and brake, without human interaction.Part autonomous mode can also refer to 3 grades of operation modes of SAE, and wherein vehicle 100 can be with
Processing is turned to, accelerates and is braked in some cases, and monitoring driving environment, but sometimes for some human interactions.
Fig. 2 is the block diagram for showing the exemplary components of vehicle 100.Component shown in Fig. 2 includes actuator 110, autonomous driving
Sensor 115, memory 120 and computing device 105.The control signal control that each actuator 110 is exported by computing device 105
System.The electric control signal exported by computing device 105 can be converted into mechanical movement by actuator 110.The example of actuator 110
It may include linear actuators, servo motor, electric motor etc..Each actuator 110 can be with specific vertical or horizontal vehicle
Control is associated.For example, propulsion actuator can control the acceleration of vehicle 100.That is, actuator is promoted to can control
Air throttle, the air-flow of the throttle control to engine.In the case where electric vehicle or hybrid vehicle, actuating is promoted
Device can be electric motor or otherwise control the speed of electric motor.Brake actuator can control vehicle brake.
That is, brake actuator can activate Brake pad so that wheel slows down.Steering actuator can control the rotation of steering wheel
Turn or otherwise control the transverse shifting of vehicle 100, including helps to turn.Each actuator 110 can be based on by example
As computing device 105 export signal and control its corresponding vehicle subsystem.
Autonomous driving sensor 115 (or referred to herein simply as sensor 115) via be programmed to detection vehicle 100 outside
Circuit, chip or other electronic components of the object in portion are implemented.For example, sensor 115 may include radar sensor, scanning
Laser range finder, light detection and ranging (laser radar) device, ultrasonic sensor and such as video camera image acquisition sensor.
Each autonomous driving sensor, which can be programmed to output, indicates the signal of the object detected by sensor.For example, sensor
115, which can be programmed to output, indicates the letter of object such as other vehicles, pedestrian, road markings, lane markings and other objects
Number.Some sensors 115 can be via the circuit, chip or other ministrys of electronics industry for the certain internal states that can detecte vehicle 100
Part is implemented.The example of internal state may include wheel velocity, wheel orientation and engine and transmission variable.In addition,
Following item can be used to detect position or the orientation of vehicle in sensor 115: for example, global positioning system (GPS) sensor;Add
Speedometer, such as piezoelectricity or MEMS (MEMS) sensor;Gyroscope, such as rate, ring laser or optical fibre gyro
Instrument;Inertia measurement unit (IMU);And magnetometer.Therefore, sensor 115 can export the interior vehicle shape for indicating vehicle 100
The signal of state, position or orientation or both.Sensor 115 can be programmed to output a signal to computing device 105, therefore count
Vehicle 100 can automatically be controlled by calculating device 105, including when detection vehicle 100 travels on low-friction surface, estimates that height rubs
The position on surface is wiped, and the track of one of high friction surface is arrived in exploitation in the case where giving any barrier nearby.
Memory 120 is implemented via circuit, chip or other electronic components, and may include following one or more
Person: read-only memory (ROM), random access memory (RAM), flash memory, electrically-programmable memory (EPROM), electricity can
Program erasable memory (EEPROM), embedded multi-media card (eMMC), hard disk drive or any volatibility or non-easy
The property lost medium etc..Memory 120 can store can be by instruction and other data that processor 105 executes.It is stored in memory
Instruction and data in 120 can be computing device 105 and the other component of possibly vehicle 100 is addressable.
Processor 125 is implemented via circuit, chip or other electronic components, and may include one or more micro-controls
It is device processed, one or more field programmable gate arrays (FPGA), one or more specific integrated circuits (ASIC), one or more
Digital signal processor (DSP), one or more customer specifle integrated circuits etc..Computing device 105, which can receive and handle, to be come
Determine vehicle 100 whether on low-friction surface, estimation from the data of sensor 115, and according to the sensing data
The positioning of high friction surface where, one of the barrier near positioning vehicle 100, the high friction surface of selection, exploitation is from vehicle
100 current location automatically navigates to high friction table to the track of one of high friction surface, and by vehicle 100
Face, while avoiding the barrier detected.
In some cases, computing device 105 can be programmed to determine that vehicle 100 is trapped on low-friction surface, all
It is such as trapped in snow, in mud or on ice.Computing device 105 can determine vehicle based on the signal exported by skidding controller
100 are trapped on low-friction surface.When 100 wheel of vehicle and vehicle 100 determined based on the signal exported by skidding controller
Coefficient of friction between the surface operated on it or road lower than previously with empirically determined value when, computing device 105 can be with
Determine that vehicle 100 is stranded.Alternatively, computing device 105 can be programmed to operate as skidding controller.Therefore, dress is calculated
Setting 105 can be programmed to determine that vehicle 100 is trapped on low-friction surface relative to target skidding based on skidding to calculate, institute
It states skidding and calculates and can be calculated according to such as wheel torque, wheel velocity or other interior vehicle characteristics.Skid calculating and mesh
Difference between mark skidding can be referred to as " skidding error ".Computing device 105 can be programmed to make a reservation for when skidding error is higher than
It is inferred to vehicle 100 when threshold value to be on low-friction surface.In addition, computing device 105 can be compiled in skidding control process
Journey is to determine whether vehicle 100 is continuously attempting to flee from low-friction surface using skidding error.That is, being controlled skidding
During process, computing device 105 can attempt for wheel torque and speed to be maintained at a certain skidding target to keep momentum, and
Therefore it is moved on low-friction surface.
Computing device 105 can be programmed to after determining that vehicle 100 is on low-friction surface but attempt to move
One or more maps are generated before and during to high friction surface.Each map may include the position of high friction surface, inspection
The position of the object measured, the position of low-friction surface, the estimated location of high friction surface, the estimated location of low-friction surface,
The path or their combination that vehicle 100 can advance.In some cases, computing device 105 can be programmed to generate
Synthetic map, the synthetic map include the estimated location of for example high friction surface and the barrier that detects.Computing device 105
It can be programmed to generate synthetic map after selecting one of high friction surface.
Computing device 105, which can be programmed to that signal will be controlled, is output to actuator 110, with according to synthetic map by vehicle
100 navigate to selected high friction surface.That is, computing device 105 can be programmed to according to synthetic map develop from
Track of the current location of vehicle 100 to the position of selected high friction surface.Computing device 105 can be programmed to vehicle
100 modes for avoiding the barrier detected develop track.Exploitation track may include computing device 105 certain times by certain
A little control signals are output to one or more of actuator 110.The control signal exported by computing device 105 leads to actuator
Vehicle 100 to be navigate to one in high friction surface by 110 manipulation such as air throttles, brake and steering wheel from its current location
Person's (that is, selected high friction surface), while avoiding the barrier detected.In some cases, it is exported by computing device 105
Control signal implement skidding control process with flee from low-friction surface and by vehicle 100 towards high friction surface guide.
Fig. 3 is that show can be by example skidding control process that computing device 105 is executed when serving as skidding controller
Control figure 300.At frame 305, computing device 105 is executed to skid and be calculated.It skids to calculate and can be car speed (Vref) and vehicle
Wheel speed (Vwhl) function.Specifically, skidding to calculate can be defined as:
At frame 310, computing device 105 can calculate skidding error.Skidding error can be target value of slip relative to
The difference of the skidding calculated at equation (1).Frame 315 indicates PID skidding controller.The output of PID skidding controller be included in
Be scheduled at frame 310 determine skidding error and other input (such as mantle friction variation estimation (frame 325) and with target road
The deviation (frame 330) of diameter) in the case where be directed to driveline torque and braking torque control signal.PID sliding control
Another output of device includes the variation (frame 320) of target value of slip.Computing device 105 determines driveline torque and system
How the variation (respectively frame 335 and frame 340) of dynamic torque influences wheel velocity (frame 345).The variation feedback of wheel velocity arrives
Frame 305, therefore new skidding can be calculated, can determine new skidding error, and new output signal can be used to control
Brake force transmission system and brake.Therefore, vehicle 100 can be controlled to current skidding target and be made by computing device 105
Vehicle gear is shifted gears to obtain momentum in subsequent iteration.In addition, when vehicle 100 reaches the surface with higher friction, meter
More multi-steering, acceleration and control for brake can be applied by calculating device 105, and control vehicle 100 according to lower skidding target.
When vehicle 100 reaches the endpoint of its expectation path, computing device 105 can stop executing control figure 300, wherein
Non- driven wheel speed is matched with by driven wheel speed.In some possible methods, if meeting certain exit criterias,
Computing device 105 can be programmed to deactivate skid control (executing control figure 300 for example, terminating).Exit criteria can be based on
Driver's input, impending collision, the collision of generation, low battery condition, low fuel condition, computing device 105 are predetermined
Fail to discharge vehicle 100 etc. after the trial of number.
Fig. 7 be include sensor 702 forward, the sensor 704 towards after, towards left sensor 706 and towards
The figure of the example vehicle 100 of right sensor 708.Sensor 702 forward, the sensor 704 towards after, the biography towards a left side
Sensor 706 and it will also be collectively known as sensor 115 or referred to as sensor 115 herein towards right sensor 708.
Sensor 115 respectively has corresponding visual field 710,712,714,716, and the visual field is the region around vehicle 100, in the area
The available sensing data about the environment around vehicle 100 of each of domain inner sensor 115.Sensor 115 can
To include color video sensor 115, infrared (IR) video sensor 115, laser radar sensor 115, radar sensor 115
Deng, wherein sensor 115 can from respectively by the visual field 710,712,714,716 indicate vehicle 100 around region obtain quilt
Dynamic sensing data or active sensor data.
Passive sensor data can be based on the sensing for obtaining natural or artificial radioactivity's (such as sunlight, street lamp etc.)
Device data, the radiation energy by the surface reflection in the environment around vehicle 100 or be refracted on sensor 115 and thus by
Sensor 115 obtains and is transmitted to computing device 105.Color video data from color video sensor 116 is passively to pass
The example of sensor data.Although natural and artificial radioactivity can be increased by 100 headlight of vehicle such as night or light conditions
By force, but color video data is considered passive sensor data.Active sensor data can be based on be not acquired as by
Form (for example, IR light, laser radar IR pulse, the radar microwave pulse etc.) acquisition of dynamic sensing data is emitted by vehicle 100
Radiation energy sensing data.The radiation energy emitted can be by the surface reflection and refraction in the environment around vehicle 100
It returns on sensor 116 and is acquired as active sensor data.For example, IR video camera can filter out visible wavelength simultaneously
And only obtain IR optical wavelength.IR light can be by including that the IR lamp in vehicle 100 is supplied.IR lamp may include the hair for emitting IR light
Optical diode (LED), the IR light is reflected and refraction returns to vehicle 100 and IR video sensor 115, the IR video sensing
Device 115 obtain reflect and the IR light that reflects as IR video data to be transmitted to computing device 105.
Additionally, some sensors 115 may be mounted at 100 inside of vehicle or vehicle body (such as engine in vehicle 100
Cabin, engineer room etc.) in measure the attribute inside vehicle 100.For example, sensors with auxiliary electrode 115 may include accelerometer, mileage
Meter, tachometer, pitching and yaw sensor, vehicle-wheel speed sensor, microphone, tire pressure sensor, bio-identification sensing
Device, suspension vibration sensor etc..The generation of these sensors 115 can be analyzed to determine the surface class that vehicle 100 is currently located
The signal of type.For example, proximity sensor 115 can be generated by issuing ultrasonic wave close to data, the ultrasonic reflections and
It is refracted off surface and is approached sensor 115 and obtain with the determining distance to the surface around vehicle 100.Acquired connects
The aspect that nearly data can be analyzed further by computing device 105 to determine the surface texture of reflecting surface.
Fig. 4 A to Fig. 4 C respectively illustrates the example map 400A to 400C that can be generated by computing device 105.For example
After computing device 105 determines that vehicle 100 is trapped on low-friction surface, map 400A to 400C can be used to develop height
The track of friction surface.That is, computing device 105 can be used skidding control process, shown in such as control figure 300
Skidding control process, by the control of vehicle 100 to the selected height identified in one or more of map 400A to 400C
Friction surface.
Fig. 4 A shows the example map with the barrier 405 and path domain 410 detected by sensor 115
400A.Map 400A further includes the description to vehicle 100.Path domain 410 be include description and vehicle 100 to vehicle 100
The area of the map 400A for the track (also referred herein as path) that can safely advance without encountering barrier 405
Domain.Path domain 410 can be calculated by computing device 105, and can position and vehicle 100 based on barrier 405 behaviour
Constrain (for example, size and turning radius).In other words, path domain 410 can be based on vehicle 100 with certain buffer
The map area that can be advanced to from its current location while avoiding obstacles 405.It can be by by passive sensor data
And the position around active sensor data and vehicle 100 is combined to create map 400A.Sensing data may include
Actively and passively colored and IR video data, laser radar data, radar data, from proximity sensor 115 close to data,
Carry out the position data of acceleration sensor 115 or gyro sensor 115.Barrier 405 can be used by computing device 105
Machine vision technique handle actively and passively sensor data with will indicate the data point of barrier 405 and background segment open and
Estimation is determined from vehicle 100 to the distance of each barrier 405 and direction.Determine the segmentation size of barrier 405, distance and
Permit position of the computing device based on vehicle 100 and the sensing data for corresponding to barrier 405 is orthogonally projected to ground in direction
On Figure 30 0.
Machine vision technique for dividing actively and passively sensor data includes convolutional neural networks (CNN).CNN can
To use the example image for being annotated with ground truth to be trained.Ground truth is about the correct of the object in sensing data
The information for determining and identifying.Segmentation is the object for determining and identifying image data (including color video data and IR video data)
Machine vision technique.Object may include vehicle, pedestrian, fence, curb, building, mark, bar or landform etc..CNN can be with
Be trained to for input image data to be divided into object and based on their appearance and they relative to obtaining image data
The position of sensor 115 identifies the object.For example, computing device 105 can be passed through based on the output object from CNN by
Compared by the identity of the CNN object determined with the predetermined object list for including curb, bar, fence, building, mark and landform
Relatively determine barrier 405.
It can be by determining that segmentation object will be corresponding to the segmentation object of barrier 405 relative to the position of vehicle 100
It orthogonally projects on map 400A.It can will divide object number by the visual field 210,212,214,216 based on sensor 115
It is combined according to 115 data of range sensor (such as laser radar data, radar data, close to data or gyro data)
To determine the position of segmentation barrier 405.CNN can be configured and be trained for input segmentation object data and laser radar number
It combines according to, radar data, close to data or gyro data, and by the data with the range of cognitive disorders object 405 and side
To.It can be trained by input object data and corresponding range data and about the ground truth information of object area
CNN。
Once it is determined that being identified as position of the object data of barrier 405 relative to vehicle 100, so that it may pass through base
The position of object data in 3d space relative to vehicle 110 position 302 project barrier 405 data point by with barrier
Hinder the associated data point of object 405 to project on map 400A, by barrier 405 from based on original viewing field 210,212,214,
216 view is converted into the top-down view in map 400A.Corresponding to object data data point can along perpendicular to
The parallel lines of the planar arrangement of map 400A are projected, and thus orthogonally project to object data on map 400A.
Fig. 4 B shows the example map 400B of the position of the high friction surface 415 of estimation.Computing device 105 can be estimated
Where is the high positioning of friction surface 415, and result can be the map of similar map 400B.The high friction surface of estimation can be with
Be by computing device 105 according to the data that sensor 115 is collected determine not by the region of the coverings such as snow, mud, ice.Computing device
105 can determine the position of high friction surface 415 by analyzing from colored or IR video sensor 115 vision data,
And the reflectivity of image, color and smoothness are compared with the image data previously obtained for different surfaces.Really
Letter vehicle is advanced in particular surface matches the previous picture number obtained and identify on the surface with acquired video data
According to amount it is related.For example, the image data on white reflective surface can match the previous image obtained and identify on accumulated snow surface.
This method facilitates the aspect that surface is determined according to the different size of feature.For example, the method can by little particle with
Bulky grain is differentiated to distinguish sand ground or rubble surface and ceramic tile or rock surface.This processing can pass through computing device 105
It is executed using machine vision technique, the machine vision technique has the size and distribution in the region of similar appearance by determining
To determine the texture in image data.Additionally, with this method, vehicle can construct luminosity map to determine surface
Property.Different luminance values indicates different surfaces.For example, dry pitch can have a characteristic data value, and wet pitch
It can have another data value.For example, the data value measured road surface type different from instruction is provided input data value
The table of confidence level is compared.
In another example, active sensor 115 can be used (for example, laser radar sensor, thunder in computing device 105
Up to sensor, proximity sensor) emit the signal on surface that can be reflected and be refracted off around vehicle 100.Computing device
105 can handle the mode of the reflection on the surface around vehicle 100 and refraction, to determine the type on surface and estimate to rub
Coefficient.For example, the mode of the reflection or refraction data obtained by computing device 105 from active sensor 115 can determine vehicle
Surface near 100 is smooth (for example, instruction pitch, ice etc.) or coarse (for example, rubble or snow).Computing device 105 is also
It can handle environmental data (for example, weather data, ambient temperature data, humidity data, precipitation data etc.).Environmental data can be with
By including that sensor 115 in vehicle 100 obtains or determines via internet or other wan protocols from far from vehicle 100
The sensor of position obtains.Environmental data can increase or decrease and utilize the coefficient of friction for being passively or actively sensor 115 and determining
The associated confidence level of estimation.For example, environmental data can indicate that temperature is lower than the freezing point of water, and therefore may be snowy
Or ice.Data from actively and passively sensor 115 can be combined with environmental data to generate and estimate by computing device 105
Confidence level of the coefficient of friction and generation of meter about the estimation.
Fig. 4 C shows example synthetic map 400C.Synthetic map 400C may include the member of map 400A and map 400B
Element.That is, synthetic map 400C can be generated by the part of combinatorial map 400A and map 400B.For example, synthesis
Map 400C shows the position of barrier 405 and high friction surface 415.In some cases, map 400C can also be shown
Path domain 410.Using map 400C, computing device 105 can be planned from the current location of vehicle 100 to high friction surface
One of 415 one or more tracks.That is, computing device 105 can be based on for example in avoiding obstacles 405
Vehicle 100 is easiest to one of navigate to which high friction surface 415 and select high friction surface 415 simultaneously.At least partly
The ground any high friction surface 415 Chong Die with path domain 410 (coming from Fig. 4 A) can be selected high friction surface 415
It is candidate.In other words, computing device 105 can be programmed to can to navigate to via such as path domain 410 from vehicle 100
High friction surface 415 is selected in those high friction surfaces.Even if when path domain 410 is not included in synthetic map 400C
It is also such.Computing device 105 can develop the current location from vehicle 100 to the track of selected high friction surface 415, and
And output with the consistent various control signals of such as control figure 300 so that vehicle 100 discharges from low-friction surface and makes vehicle
100 reach high friction surface 415.Once skidding control process can terminate, and calculate in selected high friction surface
Device 105 may return to normal (that is, more conventional) autonomous control of vehicle 100.
In addition, in some cases, computing device 105 can be programmed to continuously updated map 400A to 400C.
That is computing device 105 can be programmed to when computing device 105, which attempts, discharges vehicle 100 from low-friction surface more
Any one or more of new map 400A to 400C, therefore for example computing device 105 is considered that new barrier
405, the high friction surface 415 of new estimation, the low-friction surface newly detected etc..In some cases, computing device 105 can be with
It selects to select the new high friction surface 415 being found after initial high friction surface in computing device 105.
Fig. 5 is can be by the flow chart for the instantiation procedure 500 that computing device 105 executes.Process 500 can be in vehicle 100
Any time when automatically operating executes.As long as the continuation of vehicle 100 is operated with autonomous mode, process 500 can continue to hold
Row.
At decision box 505, whether computing device 105 determines vehicle 100 on low-friction surface.Computing device 105 can
To be programmed to determine based on the signal exported vehicle 100 on low-friction surface by sensor 115.For example, computing device
105 can be programmed to internal state (such as, wheel velocity, wheel orientation and engine and speed changer based on vehicle 100
Value) and determine vehicle 100 on low-friction surface.If computing device 105 determine vehicle 100 on low-friction surface,
Process 500 may proceed to frame 510.Otherwise, frame 505 can repeat, until computing device 105 determines vehicle 100 in low friction
On surface or until process 500 terminates.
At frame 510, computing device 105 generates at least one map.Computing device 105 can be programmed to generate
By the one or more maps for the barrier that sensor 115 detects.That is, computing device 105 can be programmed to by
Barrier is identified as by any object that sensor 115 detects and generates map to show the position of barrier.Calculate dress
Setting 105 can be programmed to generate multiple maps.First map may include barrier and path domain.Second map may include
The estimated location of high friction surface.Third map, which can be, shows the position of such as barrier and high friction surface synthetically
Figure.
At frame 515, computing device 105 selects one of the high friction surface in synthetic map.Computing device 105 can
It is reached with being programmed to selection 100 most probable of vehicle in the case wheres current location, the position of barrier etc. of given vehicle 100
High friction surface.Computing device 105 is also programmed to consider whether high friction surface nearby has other low-friction surfaces.
That is, computing device 105 can based on for example high friction surface whether at least partly by low-friction surface, barrier or
The combination of the two surrounds the priority to determine high friction surface.Computing device 105 can also make the height in 100 front of vehicle
Friction surface prior to being trapped in identical low friction table to reduce vehicle 100 in the subsequent high friction surface of vehicle 100 again
A possibility that at face.After selecting high friction surface, process 500 may proceed to frame 520.
At frame 520, computing device 105 executes skidding control process to flee from low-friction surface and rub towards selected height
It is mobile to wipe surface.Computing device 105 can be programmed to for example execute skidding control process, all as discussed above and in Fig. 3
The skidding control process 300 shown.The additional detail about skidding control process is discussed below.
At decision box 525, computing device 105 determines whether to stop the skidding control process at frame 520.Computing device
105 can be programmed to determine whether vehicle 100 arrived selected high friction surface or otherwise flee from low friction table
Face.Computing device 105 can be programmed to use based on the interior vehicle state detected when executing skidding control process 300
Position data (GPS data) etc. makes such judgement.In addition, after reaching selected high friction surface, computing device
105 can be programmed to determine whether selected high friction surface provides enough frictional force in no skidding control process
Vehicle 100 is operated when 300.If computing device 105 determines that skidding control process 300, process 500 may proceed to frame
530.Otherwise, process 500 can continue to execute frame 525.Determine that vehicle 100 has arrived at selected height and rubs in computing device 105
In the case where wiping surface but still enough tractive force cannot be obtained to control vehicle 100 in no skidding control process 300,
Process 500 may return to frame 520 or frame 510, thus can be generated new map, new high friction surface can be assessed, can
To select new high friction surface, and skidding control process 300 can be executed again.
At frame 530, computing device 105 continues the normal autonomous operation of vehicle 100.That is, computing device 105 can
To stop executing skidding control process and be controlled based on being exported come to actuator 110 since the received external signal of sensor 115
Signal processed, and vehicle 100 is controlled independent of skidding control process 300.Process 500 can terminate after frame 530.
Above-mentioned skidding control process 300 allows vehicle 100 to flee from low-friction surface.Skidding control process is (sometimes referred to as
" fleeing from mode ") it is related to controlling braking, driveline torque, shift and turns to input so that vehicle 100 is from stranded situation
Release, such as when vehicle 100 is trapped in deep snow.Vehicle 100 can be with when detecting that vehicle 100 is on low-friction surface
It automatically enables and flees from mode.Alternately or in addition, vehicle 100 can be in response to pressing lower button by occupant or with its other party
The user that formula is fled from mode from the selection of the interior passenger compartment of vehicle 100 and provided inputs to enable the mode of fleeing from.It can be via example
User as described in receiving such as information entertainment inputs.In addition, in some cases, if vehicle 100 detects that it is stranded,
It can prompt occupant's activation to flee from mode via information entertainment.
It is that the user for activating mode of fleeing from inputs as a result, as the judgement vehicle 100 of computing device 105 as receiving
It is trapped as a result, above-mentioned skidding control process 300 can be started.That is, the computing device 105 started between two parties with steering wheel
Driveline torque can be applied to by driving wheel, to measure wheel velocity and the sliding skidding of wheel so that wheel is fast
Optimum target is arrived in degree control, as discussed previously.Computing device 105 makes gear shift gears towards desired direction and continuously monitor
Wheel is not driven to determine whether they start to rotate.If wheel is not driven to rotate, computing device 105 can determine vapour
Vehicle is removing stranded position.If wheel is not driven not rotate, stop rotating or slow to lower than predetermined threshold, dress is calculated
Setting 105 makes gear shift gears and drives vehicle 100 in opposite direction so that vehicle 100 vibrates (for example, swing) to and fro, until quilt
Until driving wheel stops skidding and wheel not being driven normally to roll.Computing device 105 can also test various steering angles
It spends to attempt to find to leave the path of stranded situation, especially swings vehicle 100 in trial and fail the case where obtaining significant progress
Under.Autonomous driving sensor can be used in this skidding control process avoiding colliding with neighbouring object, as discussed previously.
If the computing device 105 operated under the mode of fleeing from fails to discharge vehicle 100 after the number of attempt of setting,
Or if not driving wheel that cannot obtain peak velocity, computing device 105 can report " can not flee from ", this may include
It returns control to driver (assuming that having occupant if field), and can also propose how best to flee to driver
It is recommended that.Such as brake pedal is stepped on, accelerator is stepped on, some drivers of steering wheel angle input may cause to computing device
105 exit the mode of fleeing from and give driver and fully control.It is electricity that vehicle may be caused, which to exit the other conditions for the mode of fleeing from,
Cell voltage is lower than given threshold lower than given threshold, fuel level, in the case where being provided with certain diagnostic trouble code, or
In the case where that will occur or collide.
By this method, computing device 105 tests different steering wheel and angle while vehicle 100 swings back and forth
Degree, while monitoring by driving wheel and not driving the wheel velocity of both wheels.Accelerated with only monitoring by driving wheel and vehicle
Degree is on the contrary, the wheel velocity convergent whithin a period of time of wheel is found by driving wheel and do not driven to computing device 105.
The example skidding control process executed during the mode of fleeing from can be as follows.Initially, computing device 105 can be determined that
Whether vehicle 100 needs to be moved forward or rearward from its current location.In addition, computing device 105 can initialize counter,
The number (that is, iteration) for attempting to discharge stranded vehicle 100 is counted computing device 105 by the counter.
After choice direction and initializing counter, computing device 105 can use the rolling of non-driven wheel speed
Dynamic average value estimates the speed of vehicle 100.Relative to by the skidding of driving wheel wheel velocity, non-driven wheel speed can have
There are certain small skiddings.Rolling average value can be used for removing the noise in data, such as come road surface caused by freely for example avenging
Variation.Computing device 105 can be programmed to detection and roll average speed in reduction and will roll average value and predetermined threshold
It is compared.As determine roll average value reduce and have decreased below predetermined threshold as a result, computing device 105
It can determine that vehicle 100 no longer obtains the momentum for fleeing from stranded situation.Computing device 105 can be by the number of iterations in counter
It is compared with the limit is attempted.If the trial limit has not yet been reached, computing device 105 can be such that counter increases and continue
It determines and rolls average value and be compared it with predetermined threshold, as described above.If the number of iterations meets or is more than to attempt pole
Limit, then computing device 105 can by vehicle shift to opposite gear (for example, if vehicle 100 previously attempts to move forward,
" reversing ";If vehicle 100 previously attempts to move backward, " driving ") and again attempt to discharge stranded vehicle 100.
After vehicle has shifted to opposite gear, computing device 105 can start that vehicle 100 is caused to accelerate.It calculates
Device 105 can continue to calculate and monitor to roll average speed and determine to roll whether average speed is reducing and declining
To lower than predetermined threshold.If it does, then computing device 105 can be inferred to again vehicle 100 there is no it is enough dynamic
Amount is to flee from.Then, computing device 105 can increase iteration and determine during each Move Mode at steering angle
Whether the peak velocity of vehicle 100 has increased.If it does, then computing device 105 may return to trial to cause vehicle
100 target (for example, previously) gears moved on former direction (for example, target direction) discharge stranded vehicle 100.It is no
Then, computing device 105 can determine that current steering angle does not help vehicle 100 to flee from.It that case, computing device
105 can order vehicle 100 attempt using different steering angles to discharge own, be discussed in greater detail below.Moreover, can
It will not be fled from from this direction (for example, direction of opposite gear) with being expected vehicle 100.If rolling average speed given
It is more than a certain value in period, or if the shift length initial position calculated is too far, in order to avoid being moved to environment danger
In danger, computing device 105 can order vehicle 100 by gear change back to target direction and execute and move in the target direction
A part of associated process is moved, as described above.
When the steering angle that please be looked for novelty, current angular can be changed calibrator quantity by computing device 105.Continuously issue for several times
This request can prevent vehicle 100 from attempting to discharge own using steering angle identical with the trial of failure.Each
At new steering angle, vehicle 100 can enter normal oscillating movement (for example, swinging back and forth) and attempt release own.
If steering angle fails, new angle can be tested, exits until reaching desired result or fail condition and flees from mould
Until formula.
Fig. 6 A to Fig. 6 D shows example vehicle 100 and flees under mode operation discussed above to flee from stranded situation.
In the example that vehicle 100 is for example trapped in deep snow or mud, actively and passively sensor data are can be used in computing device 105
To make the real-time map that may be used to vehicle 100 and get rid of poverty.Vehicle 100 in Fig. 6 A to Fig. 6 D is in uneven low friction table
On face (for example, muddy road, snow road, eisbahn etc.).
In fig. 6, computing device 105 determines that vehicle 100 cannot normally advance (for example, mobile in a manner of straight).Cause
This, mode and selection target direction are fled from the starting of computing device 105.Then, computing device 105 so that vehicle 100 is swung back and forth with
Trial is advanced in the target direction.Fig. 6 B shows the example that vehicle 100 cannot advance in the target direction.That is, vehicle
100 cannot pass through uneven road surface.In this case, computing device 105 is switched to opposite gear (shown in Fig. 6 B
In view, reversing).With reference to Fig. 6 C, 105 order vehicle 100 of computing device accelerates backward, so that vehicle 100 can have additionally
Energy come through uneven road surface.With reference to Fig. 6 D, vehicle 100 Fig. 6 A to Fig. 6 C successive ignition (for example, by making vehicle
100 successive ignitions for swinging back and forth accumulation energy) after be released.That is, in figure 6d, vehicle 100 has obtained
Enough to momentum overcome the stranded situation (for example, although uneven road surface is low friction, to pass through the uneven road
Face).Once passing through, vehicle 100 can be moved on freely in the target direction.
This method with generate synthetic map combine Shi Huixiang computing device 105 give from stranded situation discharge vehicle
100 bigger chance.Stranded situation can such as be existed above by determining being determined by the skidding rate of driving wheel for vehicle 100
Definition in equation (1).So that vehicle 110 is got rid of poverty in order to permit computing device 115, can determine skidding target, wherein target of skidding
It is defined as slip wheel and is controlled the expectation skidding rate reached by computing device 115.It is operating for vehicle 100 by driving wheel
Ground is connected to the wheel of the power drive system of vehicle 100, and thus torque can be applied to by driving wheel to cause vehicle 100
It is mobile.The linear movement of wheel is confirmed as the movement of wheel, and wherein the axis of wheel is parallel with the plane immediately below wheel
It is moved in plane.This includes the linear movement that road is parallel in road unevenness or the snow that wheel accumulates on road
Or it is moved on mud.When linear movement per second is equal to rotary motion (indicating with wheel circumference per second) per second, skidding rate
It can be 100%, wherein linear movement is converted in whole rotary motions of wheel, when vehicle 110 is by the possible quilt of driving wheel
When being trapped on ice or in snow or mud, the skidding rate can be 0%, for example, wherein any rotary motion of wheel does not produce
The linear movement of raw wheel.For example, skidding rate can be used to the coefficient of friction (μ) between estimated wheel and road.
All those computing devices as discussed herein include respectively usually order, and the order can be by being such as identified above
Those of one or more computing devices execute and for executing the frame or step of the above process.For example, discussed above
Process frame can be presented as computer executable command.
Computer executable command can be compiled from the computer program for using various programming languages and/or technology to create
It translates or explains, the programming language and/or technology include, but are not limited to Java alone or in combinationTM、C、C++、Visual
Basic, Java Script, Perl, HTML etc..In general, processor (for example, microprocessor) for example can from memory, computer
It reads the receptions orders such as medium and executes these orders, thus execute one or more processes, including one as described herein or more
A process.Such order and other data can store hereof and transmitted using a variety of computer-readable mediums.Meter
The file calculated in device is typically stored in the data on the computer-readable mediums such as storage medium, random access memory
Set.
Computer-readable medium includes that participation offer can be by any Jie for the data (for example, order) that computer is read
Matter.Such medium can take many forms, including but not limited to, non-volatile media, Volatile media etc..Non-volatile Jie
Matter includes such as CD or disk and other permanent memories.Volatile media includes typically comprising the dynamic of main memory
Random access memory (DRAM).The common form of computer-readable medium includes such as floppy disk, floppy disc, hard disk, tape, appoints
What his magnetic medium, CD-ROM, DVD, any other optical medium, punched card, paper tape, any other object with sectional hole patterns
Manage medium, RAM, PROM, EPROM, FLASH-EEPROM, any other storage chip or cassette tape or computer can be with
Any other medium therefrom read.
Unless make herein it is opposite be explicitly indicated, otherwise all terms used in the claims intention provide as this
Their common meanings that field technical staff is understood.Specifically, the clearly limitation opposite except non-claimed narration, otherwise
Using "one", "the", the singular articles such as " described " should be read as describing one or more of indicated element.
Term " exemplary " used herein is to indicate exemplary meaning, such as referring to for " exemplary widget "
It should be read as simply referring to the example of widget.
The adverbial word " about " of modification value or result means that shape, structure, measurement, value, determination, calculating etc. may be due to materials
Material, processing, manufacture, sensor measurement, calculating, the processing time, call duration time etc. defect and deviate the geometric form definitely described
Shape, distance, measurement, value, determination, calculating etc..
In the accompanying drawings, identical reference label indicates identical element.It is furthermore possible to vary some in these elements or
All.About medium as described herein, process, system, method etc., it should be understood that although the step of this class process etc. is retouched
It states to occur according to a certain orderly sequence, but this class process can be by being executed with the sequence other than sequence as described herein
The step practice.It should also be understood that may be performed simultaneously certain steps, other steps can be added or can be omitted this
Certain steps described in text.It in other words, is herein to provide for the purpose of illustrating certain embodiments to the description of process,
And it should in no way be construed so as to limit the claimed invention.
According to the present invention, it provides a method, the method is included based on sensing data and generated including barrier
Map and wheel and the road of stranded vehicle periphery between estimation coefficient of friction;Road is determined based on the map
Diameter;And the stranded vehicle is operated based on the path and based on the skidding control process of the wheel.
According to embodiment, the coefficient of friction of the estimation between the wheel and the road, which is lower than with empirically determined, is
Permit the value of the operation of the stranded vehicle.
According to embodiment, foregoing invention is further characterized in that, is revolved based on the wheel movement for being parallel to the road and wheel
Determining skidding rate is transferred with the determination skidding control process.
According to embodiment, foregoing invention is further characterized in that, determines the skidding control process, to permit the stranded vehicle
Based on the path and determine skidding rate operated, regardless of the coefficient of friction of the estimation.
According to embodiment, foregoing invention is further characterized in that, it is de- with the determination stranded vehicle to match the skidding rate of wheel
It is tired.
According to embodiment, foregoing invention is further characterized in that, generate the map include based on vehicle sensor data and
Generate the first map, including barrier.
According to embodiment, foregoing invention is further characterized in that, generate the map include based on vehicle sensor data and
The second map is generated, including the position for the stranded vehicle periphery come estimated friction coefficient.
According to embodiment, foregoing invention is further characterized in that, first map and second map are combined
And the first map and the second map and the determining path based on the combination.
According to embodiment, vehicle sensor data includes passive sensor data and active sensor data.
According to the present invention, a kind of system is provided, the system includes processor;And memory, the memory packet
Include instruction, described instruction wait for being executed by the processor with: map and vehicle including barrier are generated based on sensing data
The coefficient of friction of estimation between wheel and the road of stranded vehicle periphery;Path is determined based on the map;And based on institute
It states path and operates the stranded vehicle based on the skidding control process of the wheel.
According to embodiment, the coefficient of friction of the estimation between the wheel and the road, which is lower than with empirically determined, is
Permit the value of the operation of the stranded vehicle.
According to embodiment, foregoing invention is further characterized in that, is revolved based on the wheel movement for being parallel to the road and wheel
Determining skidding rate is transferred with the determination skidding control process.
According to embodiment, foregoing invention is further characterized in that, determines skidding control process, to permit the stranded vehicle base
It is operated in the path and the skidding rate determined, regardless of the coefficient of friction of the estimation.
According to embodiment, foregoing invention is further characterized in that, it is de- with the determination stranded vehicle to match the skidding rate of wheel
It is tired.
According to embodiment, foregoing invention is further characterized in that, generate the map include based on vehicle sensor data and
Generate the first map, including barrier.
According to embodiment, foregoing invention is further characterized in that, generate the map include based on vehicle sensor data and
The second map is generated, including the position for the stranded vehicle periphery come estimated friction coefficient.
According to embodiment, foregoing invention is further characterized in that, first map and second map are combined
And the first map and the second map and the determining path based on the combination.
According to embodiment, vehicle sensor data includes passive sensor data and active sensor data.
According to the present invention, a kind of system is provided, the system includes the component for obtaining sensing data;For
Control the component of stranded Vehicular turn, braking and power drive system;Computer component for performing the following operation: based on by
Map including barrier and wheel and institute are generated for obtaining the sensing data of the component extraction of sensing data
State the coefficient of friction of the estimation between the road of stranded vehicle periphery;Path is determined based on the map;And based on described
Path and determination skidding rate based on the wheel, by for controlling the stranded Vehicular turn, braking and powertrain
The component of system operates the stranded vehicle so that the stranded vehicle is got rid of poverty.
According to embodiment, the coefficient of friction of the estimation between the wheel and the road, which is lower than with empirically determined, is
Permit the value of the operation of the stranded vehicle.
Claims (15)
1. a kind of method, which comprises
Generated based on sensing data map including potential barrier and wheel and stranded vehicle periphery road it
Between estimation coefficient of friction;
Path is determined based on the map;And
Skidding control process based on the path and based on the wheel and operate the stranded vehicle.
2. the method as described in claim 1, wherein the coefficient of friction of the estimation between the wheel and the road is low
In with the empirically determined value to permit the operation of the stranded vehicle.
3. method according to claim 2, the method also includes based on the wheel movement and wheel for being parallel to the road
Rotation is to determine skidding rate with the determination skidding control process.
4. method as claimed in claim 3, the method also includes the determination skidding control processes, described stranded to permit
Vehicle is operated based on the path and the skidding rate determined, regardless of the coefficient of friction of the estimation.
5. method as claimed in claim 4, the method also includes matching the skidding rate of wheel with the determination stranded vehicle
It gets rid of poverty.
6. the method as described in claim 1 includes being based on vehicle sensor data the method also includes generating the map
And the first map is generated, including barrier.
7. including method as claimed in claim 6, being based on vehicle sensor data the method also includes generating the map
And the second map is generated, including the position for the stranded vehicle periphery come estimated friction coefficient.
8. the method for claim 7, the method also includes first map and second map are carried out group
Merge and the first map based on the combination and the second map and determine the path.
9. the method as described in claim 1, wherein vehicle sensor data includes passive sensor data and active sensor
Data.
10. method as claimed in claim 9, wherein the passive sensor data include color video data.
11. method as claimed in claim 10, wherein the active sensor data include laser radar data or radar number
According to.
12. method as claimed in claim 11, wherein by the active sensor data and the passive sensor data into
Row combination is to form the map by rectangular projection.
13. the method as described in claim 1, wherein determined based on the map path include determine the vehicle with certain
The position of operation is able to carry out while kind buffer avoiding obstacles.
14. method as claimed in claim 6, wherein the barrier includes vehicle, pedestrian, fence, curb building, mark
Knowledge, bar or landform.
15. a kind of system, the system comprises the meters for being programmed to execute the method as described in any one of claims 1 to 14
Calculation machine.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US15/978,983 US10759433B2 (en) | 2017-10-16 | 2018-05-14 | Vehicle escape |
US15/978,983 | 2018-05-14 |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110481556A true CN110481556A (en) | 2019-11-22 |
Family
ID=68336980
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910392564.7A Pending CN110481556A (en) | 2018-05-14 | 2019-05-13 | Vehicle is fled from |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN110481556A (en) |
DE (1) | DE102019112446A1 (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113335011A (en) * | 2021-07-21 | 2021-09-03 | 中国第一汽车股份有限公司 | Control method for vehicle getting rid of poverty, vehicle and storage medium |
CN113569692A (en) * | 2021-07-22 | 2021-10-29 | 上汽通用五菱汽车股份有限公司 | Driving assistance method, system, device, and computer-readable storage medium |
EP3943354A1 (en) * | 2020-07-21 | 2022-01-26 | Hyundai Motor Company | Vehicle and method of controlling the same |
-
2019
- 2019-05-13 DE DE102019112446.8A patent/DE102019112446A1/en active Pending
- 2019-05-13 CN CN201910392564.7A patent/CN110481556A/en active Pending
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP3943354A1 (en) * | 2020-07-21 | 2022-01-26 | Hyundai Motor Company | Vehicle and method of controlling the same |
US11427171B2 (en) | 2020-07-21 | 2022-08-30 | Hyundai Motor Company | Vehicle and method of controlling the same |
CN113335011A (en) * | 2021-07-21 | 2021-09-03 | 中国第一汽车股份有限公司 | Control method for vehicle getting rid of poverty, vehicle and storage medium |
CN113569692A (en) * | 2021-07-22 | 2021-10-29 | 上汽通用五菱汽车股份有限公司 | Driving assistance method, system, device, and computer-readable storage medium |
CN113569692B (en) * | 2021-07-22 | 2024-02-09 | 上汽通用五菱汽车股份有限公司 | Driving assistance method, system, apparatus, and computer-readable storage medium |
Also Published As
Publication number | Publication date |
---|---|
DE102019112446A1 (en) | 2019-11-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10759433B2 (en) | Vehicle escape | |
JP7422687B2 (en) | Occlusion awareness planning | |
CN108688656B (en) | Vehicle and method for controlling vehicle | |
JP7391879B2 (en) | Determining the drive envelope | |
CN106080744B (en) | Automatic driving vehicle system | |
US11372424B2 (en) | Generalized 3D inverse sensor model | |
US9454150B2 (en) | Interactive automated driving system | |
US8781670B2 (en) | Controlling vehicle lateral lane positioning | |
US10698406B2 (en) | Autonomous driving device | |
US10347127B2 (en) | Driving mode adjustment | |
CN109990794A (en) | Autonomous driving trajectory planning device with dynamic cost study | |
US20200047747A1 (en) | Vehicle and control method thereof | |
CN110473416A (en) | Controller of vehicle | |
CN110858325A (en) | Vehicle adaptive learning | |
CN111409596A (en) | Droplet path prediction | |
KR102120108B1 (en) | Autonomous vehicle and method of controlling the same | |
CN110481556A (en) | Vehicle is fled from | |
US20180074200A1 (en) | Systems and methods for determining the velocity of lidar points | |
US11827237B2 (en) | Systems and methods for real-time crash detection using telematics data | |
CN107472237A (en) | Adaptive cruise control system and the vehicle including adaptive cruise control system | |
CN103661377A (en) | System and method for automatically controlling vehicle speed | |
CN109664886A (en) | Autonomous vehicle trajectory planning | |
CN109849923A (en) | Vehicle accumulated snow horizontal respone | |
CN108734081A (en) | Vehicle lane angle detecting | |
JP2022068116A (en) | System and method for selectively changing collision warning threshold value |
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 |