CN105857306A - Vehicle autonomous parking path programming method used for multiple parking scenes - Google Patents

Abstract

The invention provides a vehicle autonomous parking path programming method used for multiple parking scenes. The method is used for automatically parking a vehicle in a parking space through an autonomous parking system when the autonomous parking system detects the available parking space. The method includes the steps that target parking space information is detected, and a parking scene is determined; the initial state and target state of the to-be-parked vehicle are determined; a vehicle kinematics differential equation is established; state variables and control variables of the vehicle are segmented, equidistance sampling is performed on each segment according to certain time step, and to-be-optimized variables are obtained; an equality constraint, boundary constraints and inequality constraints of the to-be-optimized variables are formed; motion range constraints of the to-be-parked vehicle are formed according to the motion range limit in the parking process of the vehicle; an optimization objective is determined, and an objective function is established; and by means of a nonlinear programming solver, an optimal solution of a parking path is obtained. The vehicle autonomous parking path programming method is suitable for the multiple parking scenes, the design is reasonable, abundant information can be provided so as to control autonomous parking of the vehicle, and the security coefficient is high.

Description

A kind of vehicle autonomous parking paths planning method for multiple scene of parking

Technical field

The present invention relates to vehicle autonomous parking technical field, specifically a kind of for multiple scene of parking Vehicle autonomous parking paths planning method.

Background technology

In recent years, along with increasing rapidly of domestic automobile recoverable amount, in city, parking stall day is becoming tight With narrow and small.For new hand driver, typically one difficult problem of parking, especially for parking stall excessively Narrow situation, driver is often difficult to control automobile well and parks fast and accurately, by The accident probability caused of parking raises significantly.

Autonomous parking system can help driver to park the most safely, and this system uses one Or multiple sensors detects parking stall size and position thereof, then cook up a feasible road of parking Footpath, the most automatically controls the steering of vehicle, brakes and dynamical system and follows and cook up Path complete to park.In autonomous parking system, path planning is one of key technology.Safety Collisionless, path is feasible is its requirement most basic, most important, the quickest and comfortable Parking path be also autonomous parking system needs.In addition, if parking path program results It is provided that the information of more horn of plenty, to execution system, would be even more beneficial to the tracking to institute's path planning.

Chinese invention patent CN102975715A provide a kind of automobile put down under any attitude The method of row parking path planning, the method traversal connects the dot matrix of vehicle Origin And Destination and simulates SPL, the most therefrom find one meet vehicle kinematics constraint and collision avoidance constraint road Footpath.The most disadvantageously, the method path planning assuming, vehicle only moves to a direction, Not meeting the process of parking needs repeatedly adjust to be actually needed, and is therefore formulated for power the highest.

In the patent of Application No. 201210547981.2, offer one is berthed for auto-paralleling and is The method that system determines vehicle route, the method can provide single cycle handling maneuver or two circulations to turn to The Parallel parking path planning handled.The patent of Application No. 201080064605.7 provides one For making automobile moor the method into vertical parking position progressively.Above method disadvantageously, its method It is only applicable to the parking path planning of a kind of parking position.The most disadvantageously, its method program results In the information such as speed, acceleration is not provided, be unfavorable for the tracking to institute's path planning.

The patent of Application No. 201510737989.9 provide a kind of based on full simultaneous solution plan The dynamic optimization framework of vehicle slightly-environmental integration modeling, effectively eliminates different parking stalls shape pair The impact that trajectory planning strategy causes.The most disadvantageously, the method not range of activity to vehicle Retraining, its planned trajectory may make vehicle invade other tracks and hinder on other tracks Vehicle travel even have an accident.The most disadvantageously, its do not detect discrete after two vehicles Whether collide between state, be likely to result in when vehicle travels along planned trajectory and collide.

Summary of the invention

It is an object of the invention to provide a kind of vehicle autonomous parking road for multiple scene of parking Footpath planing method, to solve the deficiencies in the prior art.

The technical scheme is that

A kind of vehicle autonomous parking paths planning method for multiple scene of parking, the method is used for Autonomous parking system detects that vehicle is automatically berthed in described sky of parking by available parking space In between, comprise the following steps:

(1) detection target parking space information, determines sight of parking；

(2) according to described target parking space information and sight of parking, determine and treat the initial of parked vehicle State and dbjective state；

(3) Ackermam model based on front-wheel steer four-wheel car, sets up vehicle kinematics differential Equation；

(4) vehicle-state variable and control variable are carried out segmentation, and walk according to the regular hour Each segmentation is equidistantly sampled by length, obtains variable to be optimized；

(5) use Lagrange's interpolation by the differential table on each for vehicle-state variable sampled point Being shown as the function of each sampled point in this sampled point place segmentation, function described in simultaneous moves with vehicle Learn the differential equation, make the described vehicle kinematics differential equation be converted into algebraic equation, formed to be optimized The equality constraint of variable；

(6) according to physical restriction and the safety requirements parked of vehicle motion, change to be optimized is formed The boundary constraint of amount；

(7) according to target parking stall peripheral obstacle, formulate collision avoidance requirement, form change to be optimized The inequality constraints of amount；

(8) limit according to vehicle range of movement during parking, formed and treat parked vehicle Range of movement retrains；

(9) determine optimization aim, set up object function；

(10) use nonlinear planning solution device, obtain the optimal solution of parking path.

The described vehicle autonomous parking paths planning method for multiple scene of parking, also include with Lower step:

Use Lagrange's interpolation to the vehicle-state variable in the optimal solution of described parking path It is fitted, with the time step of refinement, the vehicle-state variable after matching is sampled, obtain The vehicle-state sequence of refinement, detects each vehicle-state in the vehicle-state sequence of described refinement Whether collide, the most then increase segments, repeat step (4)～(10) and again enter Row path planning.

The described vehicle autonomous parking paths planning method for multiple scene of parking, step (1) In, described target parking space information include target parking stall towards, position, length and width and mesh Mark parking stall surrounding obstacles object location；Described sight of parking includes vertically parking, oblique park with parallel Park.

The described vehicle autonomous parking paths planning method for multiple scene of parking, step (3) In, the described vehicle kinematics differential equation is:

Wherein, x represents vehicle rear axle center abscissa in cartesian coordinate system, and y represents vehicle Rear shaft center's vertical coordinate in cartesian coordinate system, v represents the translational speed at vehicle rear axle center, θ represent vehicle towards the angle with cartesian coordinate system X-axis,Represent vehicle front wheel angle, a table Showing the acceleration at vehicle rear axle center, ω represents the angular velocity of vehicle front wheel angle, L_mRepresent vehicle Distance between front axle and rear axle；

In step (4), described vehicle-state variable is x, y, v, θ,Described wagon control Variable is a, ω, and vehicle-state variable and control variable are carried out segmentation, if segments is N, often One fragmented packets is containing M equidistant sampled point, each split time a length of (M-1) h, variable x、y、v、θ、In M sampled point of each segmentation after ω is discrete, the sampled point at two ends Sharing with adjacent segmentation, each fragmented packets after variable a is discrete contains M independent sampled point, The most described variable to be optimized is:

Wherein, h express time step-length；x_i, i=0,1 ..., (M-l) N represent variable x discrete after the The value of i sampled point；y_i, i=0,1 ..., (M-1) N represents the discrete rear ith sample point of variable y Value；v_i, i=0,1 ..., (M-1) N represents the value of the discrete rear ith sample point of variable v； θ_i, i=0,1 ..., (M-1) N represents the value of the discrete rear ith sample point of variable θ,Represent variableThe value of discrete rear ith sample point； ω_i, i=0,1 ..., (M-1) N represents the value of the discrete rear ith sample point of variable ω； a_p, p=0,1 ..., MN-l represents the value of discrete rear pth the sampled point of variable a.

The described vehicle autonomous parking paths planning method for multiple scene of parking, uses glug Bright day interpolation method by vehicle-state variable x, y, v, θ,Differential representation on each sampled point is The function of M sampled point in this sampled point place segmentation:

$s_{n,m}^{\′}=\frac{1}{h}\underset{m=0}{\overset{M-1}{\mathrm{\Σ}}}{\mathrm{\γ}}_{n,m}*s_{n,m},n=0,1,...,N-1;m=0,1,\mathrm{...},M-1$

Function described in simultaneous and the vehicle kinematics differential equation, make the described vehicle kinematics differential equation It is converted into algebraic equation, forms the equality constraint of variable to be optimized:

s′_{N, m}*h-f(t_{N, m}) * h=0, n=0,1 ..., N-1；M=0,1 ..., M-1

Wherein, behalf vehicle-state variable x, y, v, θ,s_{N, m}Represent vehicle-state variable At moment t_{N, m}The value at place, γ_{N, m}Represent that vehicle-state variable is at moment t_{N, m}The coefficient at place, s '_{N, m}Table Show that vehicle-state variable is at moment t_{N, m}The derivative at place, t_{N, m}=h* [(M-1) * n+m] represents The moment of m-th sample point in n-th segmentation.

The described vehicle autonomous parking paths planning method for multiple scene of parking, described in treat excellent The boundary constraint changing variable is:

Wherein, h_maxThe threshold limit value of express time step-length, x_lb、y_lb、v_lb、a_lb、θ_lb、 ω_lbRespectively represent variable x, y, v, a, θ,The lower limit of ω, x_ub、y_ub、v_ub、a_ub、 θ_ub、ω_ubRespectively represent variable x, y, v, a, θ,The higher limit of ω；Represent the original state treating parked vehicle,Represent and wait to stop The dbjective state parked；x_i、y_i、v_i、θ_i、ω_iRepresent variable respectively x、y、v、θ、The value of the discrete rear ith sample point of ω, a_pRepresent the discrete rear pth of variable a The value of individual sampled point.

The described vehicle autonomous parking paths planning method for multiple scene of parking, described in treat excellent The inequality constraints changing variable is:

$\left\{\begin{array}{cc}S(C_i,P_{j,k})>\mathrm{\α}*SA,& i=0,1,\mathrm{...},(M-1)N;j=1,2,\mathrm{...},J;k=1,2,3,4\\ S(P_j,C_{i,k})>\mathrm{\α}*{\mathrm{SP}}_j,& i=0,1,\mathrm{...},(M-1)N;j=1,2,\mathrm{...},J;k=1,2,3,4\end{array}\right.$

Wherein, C_iRepresent the tetragon being abstracted into when parked vehicle is in i-th state, P_jTable Showing the tetragon that jth barrier is abstracted into, J represents the quantity of barrier, P_{J, k}Represent tetragon P_jKth angle point, C_{I, k}Represent tetragon C_iKth angle point, S (C_i, P_{J, k}) represent P_{J, k}With four Limit shape C_iThe area of four trianglees formed and, SA represents tetragon C_iArea, S (P_j,C_{I, k}) Represent C_{I, k}With tetragon P_jThe area of four trianglees formed and, SP_jRepresent tetragon P_jFace Long-pending, α is the safety coefficient more than 1.

The described vehicle autonomous parking paths planning method for multiple scene of parking, car to be berthed Range of movement be constrained to:

$\left\{\begin{array}{cc}{x}_{lb}\≤x_{C_{i,k}}\≤x_{ub},& i=1,\mathrm{2...},(M-1)N-1;k=1,2,3,4\\ y_{lb}\≤y_{C_{i,k}}\≤y_{ub},& i=1,2,\mathrm{...},(M-1)N-1;k=1,2,3,4\end{array}\right.$

Wherein,Represent the of the tetragon being abstracted into when parked vehicle is in i-th state The X-axis coordinate of k angle point,Expression was abstracted into when parked vehicle is in i-th state The Y-axis coordinate of the kth angle point of tetragon；x_lb、x_ubRepresent respectively under vehicle-state variable x Limit value and higher limit, y_lb、y_ubRepresent lower limit and the higher limit of vehicle-state variable y respectively.

The described vehicle autonomous parking paths planning method for multiple scene of parking, object function For:

T_f=N* (M-1) h

Wherein, T_fRepresent the path planning time treating parked vehicle；

Optimization aim is shortest time, i.e. minT_f。

The described vehicle autonomous parking paths planning method for multiple scene of parking, M is integer And 4≤M≤8.

The invention have the benefit that

As shown from the above technical solution, the present invention is applicable to the path planning of multiple scene of parking, bag Include and vertically park, oblique park and Parallel parking, it is provided that meet vehicle kinematics constraint and keep away Hit the parking path of constraint, program results safe and feasible, it is also possible to provide the control such as speed, acceleration Information processed is so that tracking to institute's path planning, reasonable in design, using the teaching of the invention it is possible to provide abundant information control Vehicle autonomous parking processed, safety coefficient is high.

Accompanying drawing explanation

Fig. 1 is the block diagram of the autonomous parking system of the application embodiment of the present invention；

Fig. 2 is the vehicle geometric representation of the embodiment of the present invention；

Fig. 3 is the vertical parking path planning schematic diagram of the embodiment of the present invention；

Fig. 4 is the oblique parking path planning schematic diagram of the embodiment of the present invention；

Fig. 5 is the Parallel parking path planning schematic diagram of the embodiment of the present invention.

Detailed description of the invention

The present invention is further illustrated below in conjunction with the accompanying drawings with specific embodiment.

As it is shown in figure 1, autonomous parking system includes turning of sensory perceptual system 1, controller 2 and vehicle To system 31, brakes 32, dynamical system 33.Sensory perceptual system 1 comprises one or more and passes Sensor, such as based on ultrasound wave sensor, the sensor of view-based access control model or biography based on laser Sensor, it can detect the information of peripheral obstacle, and detect parking position information, sets Vehicle target state, then sends information above to controller 2.Controller 2 receives sensory perceptual system 1 obstacle information, target parking space information and the target status information sent, then according to the present invention Method the problem of parking is modeled and solves, the path that last Execution plan goes out.Steering 31, Brakes 32 and dynamical system 33 can receive and perform from controller 2 control command also Feedback information is sent to controller 2.Such as, steering 31 can receive steering wheel angle life Order or vehicle front wheel angle order perform corresponding steering wheel angle or vehicle front wheel angle；Braking System 32 can receive braking percentage command and perform corresponding braking；Dynamical system 33 is permissible Receive engine torque command or speed order and export corresponding engine torque or speed.

As in figure 2 it is shown, by abstract for actual vehicle be the auto model of a rectangle, this auto model Meet Ackermann steering principle.Vehicle has length L and width W.The position of vehicle is with actual car (x y) represents in the position of rear shaft center.Rear shaft center is L to tailstock distance_r, front axle center arrives Headstock distance is L_f, antero posterior axis spacing is L_m.Vehicle rear axle central speed is v, round before vehicle Angle isVehicle is towards being θ with global coordinate system X-axis angle.

The vehicle as it is shown on figure 3, all berthed in typical vertical parking position both sides, it is abstract For tetragon P₁And P₂, record its four angle points by sensory perceptual system {P_{J, k}| j=1,2；K=1,2,3,4}.For vehicle target state, it it is sensory perceptual system Setting out after type by identification parking position, general, dbjective state speed is 0, i.e. v_f=0, vehicle towards with parking stall towards identical, i.e. θ_f=90 °.For vehicle Original state, is vehicle state when starting to apply the present invention, and general, original state speed is 0, i.e. v_z=0.For the i-th state of vehicle in institute's path planning, they are four years old Individual angle point is with { C_{I, 1}, C_{I, 2}, C_{I, 3}, C_{I, 4}Represent.

Meet and there is when the front-wheel steer four-wheel car model of Ackermann steering principle turns to one turn To center, and it is positioned on rear axle extension line.At low speeds, the sliding of tire can be ignored, The vehicle kinematics differential equation can be expressed as:

Vehicle-state variable byRepresenting, control variable is represented by (a, ω).Vehicle shape State variable and control variable are continuous print on time t, engrave its then shape of sampling when a series of Become a series of states of vehicle.Vehicle-state variable and control variable are carried out segmentation, initially, Setting segments as N, such as N=10, each fragmented packets is containing 5 equidistant sampled points, each The a length of 4h of split time, variable x, y, v, θ,5 of each segmentation after ω is discrete In sampled point, the sampled point at two ends shares with adjacent segmentation, each segmentation after variable a is discrete Comprise 5 independent sampled points；Obtaining variable to be optimized is:

Differential on each sampled point can use Lagrange's interpolation to be expressed as in this segmentation 5 The function of individual sampled point:

$\left\{\begin{array}{c}{s^{\′}}_{n,0}=\frac{1}{12h}(-25\·s_{n,0}+48\·s_{n,1}-36\·s_{n,2}+16\·s_{n,3}-3\·s_{n,4})\\ {s^{\′}}_{n,1}=\frac{1}{12h}(-3\·s_{n,0}-10\·s_{n,1}+18\·s_{n,2}-6\·s_{n,3}+\·s_{n,4})\\ {s^{\′}}_{n,2}=\frac{1}{12h}(s_{n,0}-8\·s_{n,1}+8\·s_{n,3}-s_{n,4})\\ {s^{\′}}_{n,3}=\frac{1}{12h}(-s_{n,0}+6\·s_{n,1}-18\·s_{n,2}+10\·s_{n,3}+3\·s_{n,4})\\ {s^{\′}}_{n,4}=\frac{1}{12h}(3\·s_{n,0}-16\·s_{n,1}+36\·s_{n,2}-48\·s_{n,3}+25\·s_{n,4})\\ n=0,1,\mathrm{...},N-1\end{array}\right.---\left(2\right)$

Wherein, s_{N, m}, n=0,1 ..., N-1；M=0,1 ..., 4 for state variable s i.e.At moment t_{N, m}The value at place, S '_{N, m}For state variable s at moment t_n,mThe derivative at place, t_{N, m}=h* (4*n+m) represents the moment of m-th sample point in the n-th segmentation.

It is the function of time t, i.e. can be expressed asForm. Simultaneous formula (1) and (2), in formula (1), shape is such asThe differential equation be converted into as follows Algebraic equation, form the equality constraint of variable to be optimized:

s′_{N, m}*h-f(t_{N, m}) * h=0, (n=0,1 ..., N-1；M=0,1 ..., 4) (3)

Existing in vehicle motor process and limit, i.e. its front wheel angle and front wheel angle angular velocity are positive and negative Both direction has a maximum.Simultaneously for the security consideration of the process of parking, vehicle moves Scope (x, y) and car speed and acceleration should be defined.Time step h and segments N And sampling number 5 has together decided on the used time of parking, generally park and should not expend the long time, because of Time step h is limited by this.Meanwhile, first, vehicle and last shape in program results State should be respectively equal to vehicle original state and dbjective state.In sum, the border of variable to be optimized Retrain as follows:

Wherein, h_maxThe threshold limit value of express time step-length, x_lb、y_lb、v_lb、a_lb、θ_lb、 ω_lbRespectively represent variable x, y, v, a, θ,The lower limit of ω, x_ub、y_ub、v_ub、a_ub、 θ_ub、ω_ubRespectively represent variable x, y, v, a, θ,The higher limit of ω.

It is most important that vehicle does not collides with barrier during parking along path planning.Make Within may determine that whether a point is positioned at a tetragon by area-method: when point be positioned at tetragon it Time interior, four triangle areas that the four edges of this point and this tetragon is formed and equal to tetragon Area；When outside point is positioned at tetragon, the four edges of this point and this tetragon formed four Individual triangle area and the area more than tetragon.If for each state of vehicle, {P_{J, k}| j=1,2；K=1,2,3,4} all outside its tetragon, and for each barrier, car Four angle point { C of each state_{I, 1}, C_{I, 2}, C_{I, 3}, C_{I, 4}All outside its tetragon, the most permissible Each state of judgement vehicle is that safety is the most collisionless.Therefore, the inequality of variable to be optimized It is constrained to:

$\left\{\begin{array}{c}S(C_i,P_{j,k})>\mathrm{\α}*SA,(i=0,1,\mathrm{...},4N;j=1,2;k=1,2,3,4)\\ S(P_j,C_{i,k})>\mathrm{\α}*{\mathrm{SP}}_j,(i=0,1,\mathrm{...},4N;j=1,2;k=1,2,3,4)\end{array}\right.---\left(5\right)$

Wherein, S (C_i, P_{J, k}) represent P_{J, k}With tetragon C_iThe area of four trianglees formed and, SA represents tetragon C_iArea, S (P_j, C_{I, k}) represent C_{I, k}With tetragon P_jFour triangles formed The area of shape and, SP_jRepresent tetragon P_jArea；α is the safety coefficient more than 1, such as 1.05, α is the biggest, and vehicle is the biggest with the safe spacing of barrier.

Set optimization aim as shortest time, i.e. object function is:

T_f=N*4h (6)

Vehicle has a range of activity limited during parking, as it can not invade excessively Another track thus hinder other vehicle pass-throughs, simultaneously also bring potential safety hazard to self.Additionally Road both sides may be the forbidden zone, space of wall one class, therefore should apply fortune to parking path planning Dynamic range constraint.During parking, the range of movement of four impassable restrictions of angle point of vehicle, Therefore the range of movement of vehicle is constrained to:

$\left\{\begin{array}{c}{x}_{lb}\≤x_{C_{i,k}}\≤x_{ub},(i=1,\mathrm{2...},4N-1;k=1,2,3,4)\\ y_{lb}\≤y_{C_{i,k}}\≤y_{ub},(i=1,2,\mathrm{...},4N-1;k=1,2,3,4)\end{array}\right.---\left(7\right)$

Wherein,The X-axis coordinate of kth angle point when representing the i-th state of vehicle,Table The Y-axis coordinate of kth angle point when showing the i-th state of vehicle.WithCan be by vehicle shape The geometric parameter of state parameter and vehicle calculates.

Use nonlinear planning solution device, such as IPOPT, SNOPT, solve band derived above about Bundle nonlinear programming problem:

When environment of parking is excessively harsh, solver cannot be tried to achieve the solution meeting constraint, now be judged Parking path planning failure.Otherwise, solve the result obtained to be and represent shortest time parking path 4N+1 vehicle-state, use segmentation Lagrange's interpolation be fitted i.e. can get vehicle State variable and control variable value at any time.Institute is caused for preventing time step h excessive Between two vehicle-states cooked up exist collision and be not detected at, with 0.01 second be refinement time Between step-length, the vehicle-state sequence more refined.In the vehicle-state sequence of detection refinement Whether each vehicle-state collides, if it is not, then judge that parking path is planned successfully.Otherwise, Then increasing segments, such as making segments is 2N, re-starts path planning.When be repeated 3 times with If yet suffering from colliding afterwards, judge parking path planning failure.

Finally when parking path is planned successfully, park controller to dynamical system, brakes and Steering controls to perform to follow the tracks of the parking path cooked up in real time.

The present invention is applicable to multiple scene of parking, Fig. 4 and Fig. 5 respectively illustrates that vehicle is oblique parks With the parking path obtained by the application present invention in the case of Parallel parking.

It should be noted that, the present invention non-limiting parking position peripheral obstacle are other vehicles, also Can be other barriers, as locked for parking stall, be equally applicable to application scenarios of the present invention. It should be further noted that, the present invention does not limit parking position peripheral obstacle quantity as 2, other The barrier of quantity is also applied for application scenarios of the present invention.

The above embodiment is only to be described the preferred embodiment of the present invention, not The scope of the present invention is defined, on the premise of designing spirit without departing from the present invention, this area Various deformation that technical scheme is made by those of ordinary skill and improvement, all should fall into this In the protection domain that claims of invention determine.

Claims (10)

1. for a vehicle autonomous parking paths planning method for multiple scene of parking, the method Detect that vehicle is automatically berthed in described pool by available parking space for autonomous parking system In car space, it is characterised in that comprise the following steps:

(1) detection target parking space information, determines sight of parking；

(2) according to described target parking space information and sight of parking, determine and treat the initial of parked vehicle State and dbjective state；

(3) Ackermam model based on front-wheel steer four-wheel car, sets up vehicle kinematics differential Equation；

(4) vehicle-state variable and control variable are carried out segmentation, and walk according to the regular hour Each segmentation is equidistantly sampled by length, obtains variable to be optimized；

(5) use Lagrange's interpolation by the differential table on each for vehicle-state variable sampled point Being shown as the function of each sampled point in this sampled point place segmentation, function described in simultaneous moves with vehicle Learn the differential equation, make the described vehicle kinematics differential equation be converted into algebraic equation, formed to be optimized The equality constraint of variable；

(6) according to physical restriction and the safety requirements parked of vehicle motion, change to be optimized is formed The boundary constraint of amount；

(7) according to target parking stall peripheral obstacle, formulate collision avoidance requirement, form change to be optimized The inequality constraints of amount；

(8) limit according to vehicle range of movement during parking, formed and treat parked vehicle Range of movement retrains；

(9) determine optimization aim, set up object function；

(10) use nonlinear planning solution device, obtain the optimal solution of parking path.

Vehicle autonomous parking path for multiple scene of parking the most according to claim 1 Planing method, it is characterised in that further comprising the steps of:

Use Lagrange's interpolation to the vehicle-state variable in the optimal solution of described parking path It is fitted, with the time step of refinement, the vehicle-state variable after matching is sampled, obtain The vehicle-state sequence of refinement, detects each vehicle-state in the vehicle-state sequence of described refinement Whether collide, the most then increase segments, repeat step (4)～(10) and again enter Row path planning.

Vehicle autonomous parking path for multiple scene of parking the most according to claim 1 Planing method, it is characterised in that in step (1), described target parking space information includes target carriage Position towards, position, length and width and peripheral obstacle position, target parking stall；Described park Sight includes vertically parking, oblique park and Parallel parking.

Vehicle autonomous parking path for multiple scene of parking the most according to claim 1 Planing method, it is characterised in that in step (3), the described vehicle kinematics differential equation is:

Wherein, x represents vehicle rear axle center abscissa in cartesian coordinate system, and y represents vehicle Rear shaft center's vertical coordinate in cartesian coordinate system, v represents the translational speed at vehicle rear axle center, θ represent vehicle towards the angle with cartesian coordinate system X-axis,Represent vehicle front wheel angle, a table Showing the acceleration at vehicle rear axle center, ω represents the angular velocity of vehicle front wheel angle, L_mRepresent vehicle Distance between front axle and rear axle；

In step (4), described vehicle-state variable is x, y, v, θ,Described wagon control Variable is a, ω, and vehicle-state variable and control variable are carried out segmentation, if segments is N, often One fragmented packets is containing M equidistant sampled point, each split time a length of (M-1) h, variable x、y、v、θ、In M sampled point of each segmentation after ω is discrete, the sampled point at two ends Sharing with adjacent segmentation, each fragmented packets after variable a is discrete contains M independent sampled point, The most described variable to be optimized is:

Wherein, h express time step-length；x_i, i=0,1 ..., (M-1) N represent variable x discrete after the The value of i sampled point；y_i, i=0,1 ..., (M-1) N represents the discrete rear ith sample point of variable y Value；v_i, i=0,1 ..., (M-1) N represents the value of the discrete rear ith sample point of variable v； θ_i, i=0,1 ..., (M-1) N represents the value of the discrete rear ith sample point of variable θ,Represent variableThe value of discrete rear ith sample point； ω_i, i=0,1 ..., (M-1) N represents the value of the discrete rear ith sample point of variable ω； a_p, p=0,1 ..., MN-1 represents the value of discrete rear pth the sampled point of variable a.

Vehicle autonomous parking path for multiple scene of parking the most according to claim 4 Planing method, it is characterised in that use Lagrange's interpolation by vehicle-state variable x、y、v、θ、Differential representation on each sampled point is that in this sampled point place segmentation, M is individual adopts The function of sampling point:

$s_{n,m}^{\′}=\frac{1}{h}\underset{m=0}{\overset{M-1}{\Σ}}{\mathrm{\γ}}_{n,m}*s_{n,m},n=0,1,...,N-1;m=0,1,...,M-1$

Function described in simultaneous and the vehicle kinematics differential equation, make the described vehicle kinematics differential equation It is converted into algebraic equation, forms the equality constraint of variable to be optimized:

S′_{N, m}*h-f(t_{N, m}) * h=0, n=0,1 ..., N-1；M=0,1 ..., M-1

Wherein, behalf vehicle-state variable x, y, v, θ,s_{N, m}Represent vehicle-state variable At moment t_{N, m}The value at place, γ_{N, m}Represent that vehicle-state variable is at moment t_{N, m}The coefficient at place, s '_{N, m}Table Show that vehicle-state variable is at moment t_{N, m}The derivative at place, t_{N, m}=h* [(M-1) * n+m] represents The moment of m-th sample point in n-th segmentation.

Vehicle autonomous parking path for multiple scene of parking the most according to claim 4 Planing method, it is characterised in that the boundary constraint of described variable to be optimized is:

Wherein, h_maxThe threshold limit value of express time step-length, x_lb、y_lb、v_lb、a_lb、θ_lb、 ω_lbRespectively represent variable x, y, v, a, θ,The lower limit of ω, x_ub、y_ub、v_ub、a_ub、 θ_ub、ω_ubRespectively represent variable x, y, v, a, θ,The higher limit of ω；Represent the original state treating parked vehicle,Represent and wait to stop The dbjective state parked；x_i、y_i、v_i、θ_i、ω_iRepresent variable respectively x、y、v、θ、The value of the discrete rear ith sample point of ω, a_pRepresent the discrete rear pth of variable a The value of individual sampled point.

Vehicle autonomous parking path for multiple scene of parking the most according to claim 4 Planing method, it is characterised in that the inequality constraints of described variable to be optimized is:

$\left\{\begin{array}{cc}S(C_i,P_{j,k})>\mathrm{\α}*SA,& i=0,1,...,(M-1)N;j=1,2,...,J;k=1,2,3,4\\ S(P_j,C_{i,k})>\mathrm{\α}*{\mathrm{SP}}_j,& i=0,1,...,(M-1)N;j=1,2,...,J;k=1,2,3,4\end{array}\right.$

Wherein, C_iRepresent the tetragon being abstracted into when parked vehicle is in i-th state, P_jTable Showing the tetragon that jth barrier is abstracted into, J represents the quantity of barrier, P_{J, k}Represent tetragon P_jKth angle point, C_{I, k}Represent tetragon C_iKth angle point, S (C_i, P_{J, k}) represent P_{J, k}With four Limit shape C_iThe area of four trianglees formed and, SA represents tetragon C_iArea, S (P_j, C_{I, k}) Represent C_{I, k}With tetragon P_jThe area of four trianglees formed and, SP_jRepresent tetragon P_jFace Long-pending, α is the safety coefficient more than 1.

Vehicle autonomous parking path for multiple scene of parking the most according to claim 4 Planing method, it is characterised in that treat that the range of movement of parked vehicle is constrained to:

$\left\{\begin{array}{cc}{x}_{lb}\≤x_{C_{i,k}}\≤x_{ub},& i=1,2,...,\left(M-1\right)N-1;k=1,2,3,4\\ y_{lb}\≤y_{C_{i,k}}\≤y_{ub},& i=1,2,...,\left(M-1\right)N-1;k=1,2,3,4\end{array}\right.$

Wherein,Represent the of the tetragon being abstracted into when parked vehicle is in i-th state The X-axis coordinate of k angle point,Expression was abstracted into when parked vehicle is in i-th state The Y-axis coordinate of the kth angle point of tetragon；x_lb、x_ubRepresent respectively under vehicle-state variable x Limit value and higher limit, y_lb、y_ubRepresent lower limit and the higher limit of vehicle-state variable y respectively.

Vehicle autonomous parking path for multiple scene of parking the most according to claim 4 Planing method, it is characterised in that object function is:

T_f=N* (M-1) h

Wherein, T_fRepresent the path planning time treating parked vehicle；

Optimization aim is shortest time, i.e. min T_f。

Vehicle autonomous parking path for multiple scene of parking the most according to claim 4 Planing method, it is characterised in that M is integer and 4≤M≤8.