CN106379237A - Augmented reality-based lane changing whole-process driver assistant system of vehicle - Google Patents

Abstract

The invention discloses an augmented reality-based lane changing whole-process driver assistant system of a vehicle. The system comprises an image acquisition device, a data processing device and an augmented reality display device, wherein the image acquisition device is used for capturing images of a road and vehicles which surround a lane changing vehicle; the data processing device is used for extracting distance, relative speed and acceleration information between the lane changing vehicle and each surrounding vehicle, inputting the extracted information into a built model, and the model provides a suggestion for the whole process (including decision, preparation and actuation of lane changing) of lane changing of the vehicle; and the augmented reality display device is used for displaying relevant suggestions of the whole process of lane changing. Compared with the prior art, the augmented reality-based lane changing whole-process driver assistant system of the vehicle has the beneficial effects that the position and the speed of the vehicle can be detected in real time, and meanwhile, augmented reality information and augmented reality equipment for the whole process of lane changing are supplied, for example, the augmented reality device displays a symbol or a suggestion for lane changing operation, so that the risk of a traffic accident in the lane changing process can be reduced to the maximum extent.

Description

Vehicle lane-changing overall process DAS (Driver Assistant System) based on augmented reality

Technical field

The invention belongs to automobile, traffic, electronic technology field, the present invention relates to a kind of lane changing accessory system, especially It is related to a kind of new advanced vehicle lane-changing assistance system, using advanced vehicle lane-changing algorithm, including vehicle lane-changing decision making algorithm, The preparative algorithm of lane changing and vehicle lane-changing execution algorithm.Additionally, being also adopted by augmented reality, to show vehicle lane-changing Navigation information.

Background technology

Up to the present it has been proposed that several vehicle lane-changing accessory system.In order to reduce the risk of vehicle lane-changing, existing Some solutions, for example, by BMW and the popular vehicle lane-changing early warning system based on radar.When vehicle driver carries out lane-change During decision-making, in their blind area, prompting can be made based on the lane-change early warning system of radar.However, existing lane-changing assistance system is deposited In some shortcomings, mainly there are two aspects.First, existing vehicle lane-changing accessory system is mainly to provide on target track The information state of vehicle, such as position, speed etc., they are not real " auxiliary ".How to complete vehicle lane-changing process also to depend on Judgement in driver itself.This system is not provided with any suggestion, helps driver safety and is rapidly completed vehicle lane-changing mistake Journey.Therefore, existing vehicle lane-changing accessory system needs to improve.Secondly, current vehicle lane-changing DAS (Driver Assistant System) is mainly The auxiliary of lane-change decision-making level is provided for driver, and other aspects in lane-change, such as lane-change prepare and execution aspect, not It is provided with the auxiliary suggestion of correlation, and lane-change prepares and execution is also to ensure that the important composition content that lane-change can smoothly complete, Unsuitable lane-change prepares and process performing is equally possible causes traffic accident.Finally, existing lane-changing assistance system easily makes Driver diverts one's attention.Existing lane-changing assistance system requires driver that more notices are concentrated and is placed on reading information, and this may be to department Machine brings danger.New technology should be taken, transmittable information needed can reduce driver simultaneously again and divert one's attention.One complete lane-change behavior Process includes three parts：Lane-change decision-making, lane-change prepare, lane-change execution.Lane-change decision-making refer to driver decide whether into Row lane-change.Before lane-change prepares to occur to execute with lane-change after lane-change decision-making, determine lane-change in vehicle and choose target gap Afterwards, realize keeping suitable spacing with the fore-aft vehicle in target track by adjusting the lane-change vehicle speed of itself, under being One step carries out safety, and smoothly lane-change execution is ready.Lane-change execution refers to vehicle from this lanes to target track Process.However, most of at present associated research work, all only focus on lane-change decision part.Not yet have for general In the case of lane-change preparation model.In terms of lane-change process performing research, also only minority scholar is analyzed to it and has been built Mould.Scholars propose horizontal and vertical vehicle lane-changing execution model, the model in two travel directions and True Data Matching is preferable.But current lane-change execution is that the both direction executing lane-change is separated, studies one on the whole Individual complete lane-change execution model.

Content of the invention

In order to overcome the disadvantages mentioned above of prior art, the present invention proposes a kind of vehicle lane-changing overall process based on augmented reality DAS (Driver Assistant System).

The technical solution adopted for the present invention to solve the technical problems is：A kind of full mistake of the vehicle lane-changing based on augmented reality Journey DAS (Driver Assistant System), including image collecting device, data processing equipment and augmented reality display device, wherein：Described image Harvester is used for capturing the image of lane-change vehicle-surroundings road and vehicle；Described data processing equipment is used for extracting lane-change vehicle With the distance of nearby vehicle, relative velocity, acceleration information, and it is that vehicle lane-changing behavior provides suggestion using the information extracted； Described augmented reality display device is used for showing that lane-change is advised.

Compared with prior art, the positive effect of the present invention is：

The present invention utilizes graph processing technique, is capable of position and the speed of real-time detection vehicle, provides the full mistake of lane-change simultaneously The augmented reality information of journey and augmented reality equipment, show building of a symbol or lane-change operation such as on augmented reality device View, the risk of collision can be preferably minimized level just during lane-change.

Compare existing lane-changing assistance system, the present invention has two to improve greatly.First, proposed by the present invention is a replacing vehicle The DAS (Driver Assistant System) of road overall process, on the basis of existing lane-changing assistance system, the lane-change that increased driver prepares and changes Road executes miscellaneous function.Warning against danger not only can be provided for driver, but also other lane-change can be provided for driver Suggestion, helps its adjustment and the relative velocity of surrounding vehicles, following distance etc., is that lane-change execution is ready, and finally helps drive The person of sailing is safely completed the transverse movement from current lane to target track.Secondly, in order to reduce DAS (Driver Assistant System) to driving The interference that member's notice causes, present invention employs augmented reality.Using augmented reality equipment, driver does not need to rotate Head is observing rearview mirror it is only necessary to follow the guidance of the display on augmented reality equipment, so that the efficiency of driver's lane-change increases Plus, and the risk reduction that lane-change is collided.

Specific embodiment

A kind of vehicle lane-changing overall process DAS (Driver Assistant System) based on augmented reality, content mainly includes three aspects：

1) vehicle lane-changing accessory system builds.The system includes image collecting device, the data being arranged on lane-change vehicle Processing meanss and augmented reality display device.First pass through the image that image collecting device captures periphery traffic environment, then, lead to Cross data processing equipment and extract the information such as the distance of vehicle, relative velocity, acceleration, the information of extraction is processed as car Auxiliary lane-change algorithm input, finally by augmented reality display device present vehicle assist lane-change algorithm output lane-change Suggestion.

2) vehicle lane-changing algorithm.Vehicle lane-changing behavior, as the key component in the basic driving behavior of highway, is pacified to traffic Entirely there is significant impact, vehicle lane-changing behavior includes three parts：Lane-change decision-making, lane-change prepares, and lane-change executes.However, mesh In front most research work and vehicle lane-changing DAS (Driver Assistant System), all only focus on lane-change decision part, lane-change has been prepared Research with lane-change executable portion and consideration are seldom.New vehicle lane-changing to be set up in the present system prepare and lane-change execution mould Type, and the algorithm flow of core is provided.

3) image processing algorithm based on augmented reality.Main include lane line positioning and extraction, vehicle detection and recognition, Distance and relative velocity calculate, the augmented reality of lane-change information shows.Need to be converted into ash by the image that camera captures Degree image, extracts the edge of road using related algorithm, that is, lane line.Find suitable feature detection algorithm detection capture Vehicle in image, and then determine vehicle position in the picture, for calculating the distance between vehicle and relative velocity.? Afterwards, last track change tutorial message is shown on augmented reality equipment interface.

Hereinafter the system will be described in detail：

This system includes image collecting device, data processing equipment and augmented reality display device.Using image collector The video camera put is used for catching road and vehicle image, and the Android device of data processing equipment is used for the graphics process collected simultaneously Extract effective information, wherein OpenCV 2.4 is used for processing computer vision.During process, detect track first, then identification is worked as Front track vehicle and fellow road-users.Through information processing, the image procossing output result of traffic collecting unit is to change The vehicle distances of road vehicle periphery and speed.In addition to image processing function, data processing equipment also carries vehicle lane-changing Work, and carry out vehicle lane-changing offer suggestion for driver.Augmented reality device has display function, and lane-change suggestion is displayed on On augmented reality device, for receiving vehicle lane-changing suggestion.

The image collecting device of this system needs three video cameras to cover whole induction region.Three video cameras are pacified respectively It is contained on the dead ahead of vehicle and the mirror of vehicle left and right side, be respectively used to catch current lane front truck, and on adjacent lane Front and back vehicle, that is, have five cars (front truck, the forward and backward car of left-hand lane, the forward and backward car of right-hand lane) need supervised Control.This five classes vehicle has an impact to vehicle lane-changing behavior, will be included in vehicle lane-changing model unit.If in five cars One of them in induction region, is not just not considered.

First, lane detection：Captured picture will be processed to position and to extract lane line.The image of capture is that RGB is color Color image, it needs substantial amounts of process time, so needing to convert them into gray level image.Then position the sense of road area Answer region.Zones of different has different gray scales, and compared with other regions, the gray level of road area is relatively low.Using Sobel method Extract lane line further.Although lane line is to extract from original image, but still there is much noise.Therefore, by adopting Reduce noise with Hough transform method, obtain road edge.Finally, lane detection success output display.

2nd, vehicle detection process：By the vehicle in the image using the detection capture of Haar-like feature detection algorithm, Rate of false alarm will reduce further.Each image that camera is shot (frame) is pre-processed by this system, to improve entirety Efficiency and precision.Although input picture is adjusted, still excessive for property detector.Therefore, it is necessary to positioning input The induction region of image is so that system real-time response.We select the grey center adjusting to detect vehicle as induction region. Meanwhile, always occur from the center of image for guaranteeing our target vehicle, need the position of further calibration camera.For vehicle Detection, we use Haar-like property detector, the AdaBoost algorithm commonly used using Face datection, and its speed is fast, precision High.In order that Haar-like property detector is effective in vehicle detection, we must execute training number using an instrument According to basic operation, herein adopt Imglab training characteristics detector.Wrong report comes from the identification side of Haar-like property detector Method.It identifies that any single object is a chassis, and this does not meet the fact.Wrong report will substantially reduce the accuracy of system, logical for this Cross to reduce wrong report using two kinds of different technology (size filter and aspect ratio wave filter).Resizing filter can by height too Big or too little vehicle filtering falls.Aspect ratio wave filter utilizes the general the ratio of width to height of vehicle.Most of vehicles aspect ratio (width/ Highly) between 0.4 to 1.6.If the aspect ratio of the vehicle detecting is not in this range, it is probably a wrong report, then Ignore this object.After this, system return the bottom coordinate of checking vehicle with further determine that the distance of this vehicle with relative Speed.

3rd, distance and velocity computation process：Display determines distance and the spilling of relative velocity, using inverse perspective mapping (IPM, Inverse Perspective Mapping) technology, achievable coordinate system is from the change of an angle to another angle Change.In order to realize IPM, some parameters should preferentially obtain.For obtaining plane and top view, the height h of video camera need to be obtained, from phase The angle, θ to ground for the machine and the focal length K of camera.H and l passes through to adjust the position of video camera and direction obtains, and K needs by correction Video camera is estimated.Find that the chessboard that camera focus are currently in use is calibrated this and had good grounds, and include in OpenCV storehouse To calculate the focal length K of camera.Once obtaining height h, the focal length K of the angle, θ from camera to ground and camera of camera, conversion can With carry out.In changing image, two pel spacing p will represent their distance.Related to the pixel of the distance of real world, should Calibration in advance.The purpose of this calibration is the length of the real world estimating a pixel.Video camera will be installed in one admittedly Fixed known height h and angle, θ, then length is the character string of l by before the camera being placed on.Camera shoots character After the image of string, the focal length K with camera is admitted to IPM algorithm by image.Pixel p is accounted for by the image after the string conversion of length l According to quantity would indicate that real world length l.By with p divided by l, it is estimated that the length of one of real world pixel Degree.Sustained height and the angle of vehicle after the completion of calibration, should be arranged on.Then, between lane-change vehicle and any target vehicle Relative distance can obtain.Relative velocity can be by calculating to interframe range difference and time difference, because program is Determine vehicle lane-changing condition, need relative velocity rather than absolute velocity.

4th, last track change tutorial message is shown on augmented reality equipment interface：On equipment interface display with The time interval of surrounding vehicles (including main stem preceding vehicle, target track preceding vehicle and delayed vehicle), suitable track become Change point arrow instruction, and the acceleration or deceleration instruction sending to driver etc..

5th, lane-change process：The information of vehicles obtaining will be used for generating vehicle lane-changing guiding.Vehicle lane-changing includes replacing vehicle Road decision-making, vehicle lane-changing prepare and vehicle lane-changing executes three parts, in lane-change decision-making, when driver produces lane-change wish, first First it can be made decision, and determines the selection of track and target range, that is, select track and target distance.The decision model in track makes With the model being proposed by Gipps.Afterwards, driver determines during lane-change it is thus necessary to determine that the spacing of oneself and speed are if appropriate for carrying out Further lane-change, if be suitable for, vehicle can continue executing with lane-change；Otherwise, they need to adjust itself and other cars further Spacing or the synchronous speed with target track surrounding vehicles.

Lane-change decision model：

The decision model in track uses the model being proposed by Gipps, wherein：Lane-change vehicle is n, changing on current lane Vehicle in front of road vehicle n is n-1, and the vehicle in front of the lane-change vehicle n on target track is n-2, on target track The vehicle at lane-change vehicle n rear is n+2.

Lane-change preparation model：

Lane-change prepare behavior be lane-change vehicle n in order to find the expectation time headway and vehicle n-2 and vehicle n+2 between, For the process that the lane-change execution of next step is ready in advance.Therefore, the lane-change preparation model set up herein is based on expectation car Head when away from, the expression formula of model is as follows：

a_n(t+ τ)=α (G_n-2(t)-G_T1(t))+(1-α)(G_n+2(t)-G_T2(t))

$G_{T1}\left(t\right)=\frac{a_1+b_1{V}_n\left(t\right)-c_1\left(V_{n-2}\right(t)-V_n\left(t\right))}{V_n\left(t\right)}$

$G_{T2}\left(t\right)=\frac{a_2-b_2{V}_n\left(t\right)+c_2\left(V_{n+2}\right(t)-V_n\left(t\right))}{V_{n+2}\left(t\right)}$

In formula, τ represents the reaction time of vehicle n, a_n(t+ τ) represents the acceleration in the t+ τ moment for the vehicle n, v_nT () represents t The speed of moment vehicle n, v_n-2T () represents the speed of t vehicle n-2, v_n+2T () represents the speed of t vehicle n+2, G_n-2 T () represents the time headway of t vehicle n and vehicle n-2, G_n+2T () represents the time headway of t vehicle n and vehicle n+2, △v_n-2T () represents the speed difference of t vehicle n and vehicle n-2, △ v_n+2T () represents the speed of t vehicle n and vehicle n+2 Difference, G_T1T () represents the time headway of t driver desired vehicle n-2 and vehicle n, G_T2T () represents t driver's phase The time headway of the vehicle n+2 hoping and vehicle n, α represents the consideration to target track front truck in total acceleration for the lane-change vehicle Degree, a₁, b₁, c₁, a₂, b₂, c₂Represent the linear combination coefficient about car speed in time headway expression formula, there is no actual thing Reason meaning, can change according to different lane-change situations.

Vehicle lane-changing executes model：

1. model framework

Vehicle lane-changing execution model general frame proposed by the present invention is broadly divided into three submodules：Speed generation module, Trajectory planning module and Track Pick-up module.Vehicle perception nearby vehicle transport condition (speed, position etc.), and in this, as being System input, obtains the speed of each step-length of vehicle through speed generation module computing.Meanwhile, nearby vehicle transport condition input To trajectory planning module, the car speed obtaining further according to speed generation module, obtain in each step-length through trajectory planning Optimal trajectory, then again based on speed and optimal trajectory, obtains vehicle lane-changing execution using Track Pick-up module and starts extremely The final track terminated.

The major function of each submodule is as follows：

(1) speed generation module

The target of speed generation module is to obtain the real-time speed of each time step of vehicle.Model considers periphery first The transport condition of vehicle is it is assumed that during current lane-change vehicle n lane-change execution in the vertical as far as possible and before the vehicle n of its current lane The front truck n-2 of the vehicle n on car n-1 and target track and rear car n+2 keep a safe distance, thus calculating in longitudinal direction Travel speed, then further according to the relation of vehicle itself lateral velocity and longitudinal velocity geometry, calculate the general speed of vehicle.

(2) trajectory planning module

Vehicle driver, after decision-making goes out the speed of its lane-change execution, can estimate out a suitable vehicle according to this speed Driving trace, this process, we are referred to as trajectory planning.Trajectory planning is according to the current location of vehicle, lane-change transversely Distance, and according to the calculated present speed of speed generation module, consider lane-change efficiency and lane-change comfortableness, find Optimum lane-change perform track to current step.This trajectory planning module is that each time step planning generation one is new Excellent lane-change track, so vehicle can realize real-time adjustment lane-change track in lane-change execution whole process, high-efficiency comfortable ground is complete Become lane-change implementation procedure.

(3) Track Pick-up module

The lane-change execution speed generating according to speed generation module and trajectory planning module and optimal trajectory, vehicle completes one The traveling of individual time step, final obtains the new position of vehicle and course angle at the end of step-length, using initial as next step-length State.After successive ignition, finally can generate a complete vehicle lane-changing perform track.

2. speed generation module

Lane-change execution rate pattern used herein thinks that lengthwise movement in lane-change implementation procedure for the vehicle is subject to The vehicle n-1 in front of lane-change vehicle n on current lane, the vehicle n-2 in front of the lane-change vehicle n on target track and rear Vehicle n+2 impact, according to the real time running state of these vehicles, itself speed is adjusted.Model introduces Gipps Safe distance rule, that is, assume vehicle lane-change execution in always attempt to and vehicle n-1, n-2 and n+2 keep a security row Sail speed, vehicle is to be weighted by these three speed to obtain in longitudinal final speed.As follows brief introduction is carried out to this model.According to Gipps model, in order to not collide with current lane front truck n-1, vehicle n need to keep following safe speed：

$v_n^{{\mathrm{PV}}_c}(t+\mathrm{\τ})=b_n\mathrm{\τ}+\sqrt{b_n^2{\mathrm{\τ}}^2-b_n\[2\left(x_{n-1}\right(t)-l_{n-1}-x_n\left(t\right))-v_n\left(t\right)\mathrm{\τ}-v_{n-1}^2\left(t\right)/b_{n-1}\]}---\left(1\right)$

In formula,It is longitudinal safe speed of n vehicle relatively current track front truck n-1, l_n-1Before being current lane The vehicle commander of car n-1, b_n-1And b_nIt is the respective maximum brake acceleration of current lane front truck n-1 and Ben Che.x_nT () is that t is changed The lengthwise position of road vehicle n, x_n-1T () is the lengthwise position of t current lane front truck n-1, τ is the reaction time.

Meanwhile, it is deduced the safety speed that should not keep when colliding with front truck n-2 and rear car n+2 in target track Degree, its expression formula is as follows：

$v_n^{{\mathrm{PV}}_t}(t+\mathrm{\τ})=b_n\mathrm{\τ}+\sqrt{b_n^2{\mathrm{\τ}}^2-b_n\[2\left(x_{n-2}\right(t)-l_{n-2}-x_n\left(t\right))-v_n\left(t\right)\mathrm{\τ}-v_{n-2}^2\left(t\right)/b_{n-2}\]}---\left(2\right)$

$v_n^{{\mathrm{LV}}_t}(t+\mathrm{\τ})=\frac{b_n\mathrm{\τ}}{2}+\sqrt{{\left(\frac{b_n\mathrm{\τ}}{2}\right)}^2+b_n\[2\left(x_n\right(t)-l_n-x_{n+2}\left(t\right))+v_n\left(t\right)\mathrm{\τ}-2v_{n+2}\left(t\right)\mathrm{\τ}+\frac{v_{n+2}^2\left(t\right)}{b_{n+2}}\]}---\left(3\right)$

In formulaWithIt is lane-change vehicle n relative target track front truck n-2 and target track rear car n+2 Longitudinal safe speed, l_n-2It is target track front truck n-2 vehicle commander, b_n-2It is the maximum brake acceleration of target track front truck n-2, x_n-2T () is the lengthwise position of t target track front truck n-2, l_nIt is the vehicle commander of lane-change vehicle n, b_n+2It is target track rear car n + 2 maximum brake acceleration, x_n+2T () is the lengthwise position of t target track rear car n+2.

Then, consider vehicle n-1, the impact to vehicle n speed for n-2 and n+2, obtain final the indulging of lane-change vehicle As follows to speed：

$v_n(t+\mathrm{\τ})={{\mathrm{\αv}}_n}^{{\mathrm{PV}}_c}(t+\mathrm{\τ})+(1-\mathrm{\α})\[{{\mathrm{\βv}}_n}^{{\mathrm{PV}}_t}(t+\mathrm{\τ})+(1-\mathrm{\β}){v_n}^{{\mathrm{LV}}_t}(t+\mathrm{\τ})\]---\left(4\right)$

In formula, v_n(t+ τ) is the longitudinal velocity of lane-change vehicle n, α and β is weight coefficient.

When in formula (4), β represents vehicle lane-changing execution, the weight that the front truck n-2 in target track is considered, and 1- β is right The weight that rear car n+2 in target track considers.

In formula (4), α is the weight coefficient that this track vehicle affects on lane-change vehicle n speed, then 1- α is the car of target vehicle On vehicle n speed impact weight coefficient.α reduces with the minimizing away from target track lateral attitude for the lane-change vehicle, it Span is between 0 and 1.Before lane-change vehicle is across lane line, the value of α is more than 0.5 that is to say, that the car of current lane To lane-change execute vehicle impact larger；After lane-change vehicle is across lane line, the value of α will less than 0.5, target track The impact that vehicle executes vehicle to lane-change becomes much larger.The computing formula of α is shown below：

α=tan (p_n(t))/k (5)

p_n(t)=y_f(t)/Y (6)

Wherein, y_fT () represents lane-change vehicle n in the lateral attitude of t, that is, current location from home horizontal away from From Y is the lateral separation in target track and initial track, and k is a parameter needing to demarcate.

In the vehicle lane-changing implementation procedure obtaining on the basis of longitudinal velocity, according to consolidating of longitudinal velocity and lateral velocity There is geometrical relationship, the general speed obtaining vehicle traveling is as follows：

$u_n(t+\mathrm{\τ})=\frac{v_n(t+\mathrm{\τ})}{cos\left(\mathrm{\θ}\right(t+\mathrm{\τ}))}---\left(7\right)$

In formula, u_n(t+ τ) is the general speed of lane-change vehicle n vehicle, and θ is the course angle of automobile, that is, vehicle motion side To the angle with x coordinate axle.

3. trajectory planning module

Lane-change vehicle carries out the planning of track when lane-change executes, and each time step can find a new optimum rail Mark.The present invention uses cubic polynomial using polynomial curve come the driving trace in simulating vehicle lane-change implementation procedure, The linear and vehicle lane-changing execution curve of cubic polynomial is very close, and cubic polynomial track had both had second-order smooth degree Feature, turn avoid high-order moment track and need to introduce abstract parameter and carry out the complexity of constrained trajectory, its expression is shown as follows：

$y_n\left(x\right)=a_0+a_1{x}_n+a_2{x}_n^2+a_3{x}_n^3---\left(8\right)$

Wherein, a₀, a₁, a₂, a₃It is parameter to be determined, x_nFor the position of longitudinally upper lane-change vehicle n, y_nFor transversely changing The position of road vehicle n.In lane-change implementation procedure, vehicle can cook up a new lane-change execution rail in each step-length Mark, this track with current location as starting point, with a certain position on the center line of target track as terminal, vehicle when terminal The direction of motion is parallel with track.Here, we use moving coordinate system, the start position of each step-length is defined as (0,0) point, Final position is defined as (x_f,y_f), the vehicle course angle of arbitrary step-length starting point is θ_i, the vehicle course angle of terminal is 0, then have：

y′_n(0)=tan θ_i(9)

y′_n(x_f)=0 (10)

By the starting point coordinate (0,0) of above vehicle lane-changing execution and terminal point coordinate (x_f,y_f) and formula (9) and (10) substitution The expression formula that equation of locus (8) obtains each parameter in equation of locus is as follows：

a₁=tan θ_i(11)

a₀=0 (12)

$a_2=\frac{3y_f-2x_f{\mathrm{tan\θ}}_i}{{x_f}^2}---\left(13\right)$

$a_3=\frac{x_f{\mathrm{tan\θ}}_i-2y_f}{{x_f}^3}---\left(14\right)$

Formula (11)-(14) are updated to equation of locus (8) obtain：

$y_n\left(x\right)=x_n{\mathrm{tan\θ}}_i+\frac{3y_f-2x_f{\mathrm{tan\θ}}_i}{x_f^2}{x_n}^2+\frac{x_f{\mathrm{tan\θ}}_i-2y_f}{x_f^3}{x_n}^3---\left(15\right)$

It should be noted that different from step-length afterwards, in first step-length of vehicle lane-changing execution, vehicle starts to sail out of Current lane, now the direction of motion of vehicle is parallel with track direction, and its original heading angle is also 0.

In formula (15), tan θ_iCan be determined with starting point course angle, y_fFor current location lateral separation from home, for known Amount.So, the maximum position x that when equation of locus is completed by lane-change execution, vehicle can reach in the vertical_fUnique determination.And During true lane-change, driver always expects to complete lane-change execution with the shorter time during lane-change execution, and Ensure that the comfortableness of lane-change process is maximum as far as possible.But, this two factors are conflicting, and driver is in lane-change implementation procedure In this two factors need to be carried out with balance consider.Therefore, construct the cost function of lane-change execution herein to express driving Member's consideration to efficiency and comfortableness in lane-change execution.Represent comfort level with lateral peak acceleration in this cost function, Lateral peak acceleration is bigger, illustrates that comfort level is poorer, with maximum lengthwise position x of vehicle lane-changing execution_fRepresent efficiency, x_fMore Greatly, illustrate that the efficiency of lane-change execution is lower.In order that two factors are estimated in an order of magnitude, we are carried out to them Normalized, expression is as follows：

$J=\mathrm{\ω}{\left(\frac{a_s^{last}}{a_s^{\max }}\right)}^2+(1-\mathrm{\ω})\frac{x_f}{x_f^{\max }}---\left(16\right)$

In formula, J is cost function,Refer to the side acceleration of destination county in a lane-change perform track, be track In maximum side acceleration,Refer to the maximum side acceleration of all vehicles in lane-change implementation procedure,Refer to own The maximum fore-and-aft distance running in lane-change implementation procedure of vehicle, ω is the weight parameter needing to demarcate.

Cost function J is in the side acceleration of destination countyCan be calculated by following formula：

$a_s^{last}=u_n^{2}K\left(x_f\right)---\left(17\right)$

In formula, K () is the curvature function with regard to longitudinal direction of car position of lane-change perform track.And establish in formula (17) K () must be related to equation of locus：

$K=\left|\frac{y_n^{\′\′}}{{(1+y_n^{\′2})}^{\frac{3}{2}}}\right|---\left(18\right)$

Wherein, y ' and y " is respectively single order and the second dervative of lane-change perform track equation (15), and their expression formula is such as Under：

$y_n^{\′}\left(x\right)={\mathrm{tan\θ}}_i+2\frac{3y_f-2x_f{\mathrm{tan\θ}}_i}{{x_f}^2}{x}_n+3\frac{x_f{\mathrm{tan\θ}}_i-2y_f}{{x_f}^3}{x}_n^2---\left(19\right)$

$y_n^{\′\′}\left(x\right)=2\frac{3y_f-2x_f{\mathrm{tan\θ}}_i}{{x_f}^2}+6\frac{x_f{\mathrm{tan\θ}}_i-2y_f}{{x_f}^3}{x}_n---\left(20\right)$

Formula (19) and (20) are updated in formula (18) and draw the relevant x of curvature K_nFunction (0≤x_n≤x_f), its expression formula As follows：

$K=\left|\frac{2\frac{3y_f-2x_f{\mathrm{tan\θ}}_i}{{x_f}^2}+6\frac{x_f{\mathrm{tan\θ}}_i-2y_f}{{x_f}^3}{x}_n}{{(1+{({\mathrm{tan\θ}}_i+2\frac{3y_f-2x_f{\mathrm{tan\θ}}_i}{{x_f}^2}{x}_n+3\frac{x_f{\mathrm{tan\θ}}_i-2y_f}{{x_f}^3}{x}_n^2)}^2)}^{\frac{3}{2}}}\right|---\left(21\right)$

Then formula (17) is：

$a_s^{last}=u_n^2\left|\frac{2x_f{\mathrm{tan\θ}}_i-6y_f}{{x_f}^2}\right|---\left(22\right)$

Again formula (22) is updated to cost function J, that is, in formula (16), obtains the final expression formula of cost function J, such as Under：

$J={\mathrm{\ω}}_1{\left(\frac{{u_n}^2\left|\frac{2x_f{\mathrm{tan\θ}}_i-6y_f}{{x_f}^2}\right|}{a_c^{\max }}\right)}^2+{\mathrm{\ω}}_2\frac{x_f}{x_f^{\max }}---\left(23\right)$

From formula (23) as can be seen that cost function J is with regard to x_fFunction of a single variable, J gets corresponding x during minimum of a value_fI.e. Lengthwise position for this step-length optimal trajectory terminal.In lane-change implementation procedure, each step-length can produce an optimum rail Mark.

4. Track Pick-up module

Vehicle lane-changing implementation procedure needs to complete through multiple time steps, and the above-mentioned speed of each step-length generates Module and trajectory planning module will execute once, thus realize vehicle moving to target track from the center line of current lane Line.In this part, we will introduce the movement how lane-change vehicle is realized in a step-length, calculates vehicle at the end of step-length New position and course angle.

In a step-length, arc length L that vehicle runs over along the optimal trajectory planning out is as follows：

L=u_n(t)τ (24)

Meanwhile, arc length can be calculated using rectangular area integration quad method to lane-change perform track equation (15) as follows：

$\begin{array}{c}L={\∫}_0^{X}dx={\∫}_0^X\sqrt{1+y^{\′2}}dx\\ ={\∫}_0^X\sqrt{1+{({\mathrm{tan\θ}}_i+2\frac{3y_f-2x_f{\mathrm{tan\θ}}_i}{{x_f}^2}{x}_n+3\frac{x_f{\mathrm{tan\θ}}_i-2y_f}{{x_f}^3}{x}_n^2)}^2}dx\end{array}---\left(25\right)$

In formula, X is the longitudinal coordinate that this step-length terminates, and is also the longitudinal coordinate that next step-length starts.

Then have:

$u_n\mathrm{\τ}={\∫}_0^X\sqrt{1+{({\mathrm{tan\θ}}_i+2\frac{3y_f-2x_f{\mathrm{tan\θ}}_i}{{x_f}^2}{x}_n+3\frac{x_f{\mathrm{tan\θ}}_i-2y_f}{{x_f}^3}{x}_n^2)}^2}dx---\left(26\right)$

The value of X can be tried to achieve according to formula (26), and be updated to can get in equation of locus (15) horizontal at the end of step-length Coordinate.

Next the course angle of the vehicle at the end of material calculation, the single order derived function (19) making equation of locus is：

$tan\mathrm{\θ}={\mathrm{tan\θ}}_i+2\frac{3y_f-2x_f{\mathrm{tan\θ}}_i}{{x_f}^2}{x}_n+3\frac{x_f{\mathrm{tan\θ}}_i-2y_f}{{x_f}^3}{x}_n^2---\left(27\right)$

In formula, θ is the vehicle course angle of any point in lane-change perform track.

Make x_n=X, then can obtain the course angle of vehicle at the end of current step, as follows：

${\mathrm{\θ}}_f=arctan({\mathrm{tan\θ}}_i+2\frac{3y_f-2x_f{\mathrm{tan\θ}}_i}{{x_f}^2}X+3\frac{x_f{\mathrm{tan\θ}}_i-2y_f}{{x_f}^3}{X}^2)---\left(28\right)$

Arrive this, the vehicle-state at the end of current step be can be obtained by according to formula (26) and (28), thus also obtain The state that next step-length starts, after successive ignition, the local path of each step-length is end to end, reaches in target vehicle Line, defines a complete lane-change perform track.

Claims (9)

1. a kind of vehicle lane-changing overall process DAS (Driver Assistant System) based on augmented reality it is characterised in that：Including image collector Put, data processing equipment and augmented reality display device, wherein：Described image harvester is used for capturing lane-change vehicle-surroundings road Road and the image of vehicle；Described data processing equipment is used for extracting the distance of lane-change vehicle and nearby vehicle, relative velocity, acceleration Degree information, and provide suggestion using the information extracted for vehicle lane-changing overall process；Described augmented reality display device is used for showing Lane-change is advised.

2. the vehicle lane-changing overall process DAS (Driver Assistant System) based on augmented reality according to claim 1 it is characterised in that： Described image harvester includes three shootings being separately mounted in lane-change right ahead and lane-change vehicle left and right side mirror Machine, is respectively used to catch current lane front truck, the forward and backward car of left-hand lane, and the image of the forward and backward car of right-hand lane.

3. the vehicle lane-changing overall process DAS (Driver Assistant System) based on augmented reality according to claim 1 it is characterised in that： Described vehicle lane-changing overall process includes lane-change decision process, lane-change set-up procedure and lane-change implementation procedure.

4. the vehicle lane-changing overall process DAS (Driver Assistant System) based on augmented reality according to claim 3 it is characterised in that： In described lane-change set-up procedure, between setting up based on the lane-change vehicle forward and backward side vehicle on lane-change vehicle-to-target track Expect the lane-change preparation model of time headway：

a_n(t+ τ)=α (G_n-2(t)-G_T1(t))+(1-α)(G_n+2(t)-G_T2(t))

$G_{T1}\left(t\right)=\frac{a_1+b_1{V}_n\left(t\right)-c_1\left(V_{n-2}\right(t)-V_n(t\left)\right)}{V_n\left(t\right)}$

$G_{T2}\left(t\right)=\frac{a_2-b_2{V}_n\left(t\right)+c_2\left(V_{n+2}\right(t)-V_n(t\left)\right)}{V_{n+2}\left(t\right)}$

In formula, n is lane-change vehicle, and n-2 is the vehicle in front of the lane-change vehicle n on target track, and n+2 is changing on target track The vehicle at road vehicle n rear；τ represents the reaction time of vehicle n, a_n(t+ τ) represents the acceleration in the t+ τ moment for the vehicle n, v_n(t) Represent the speed of t vehicle n, v_n-2T () represents the speed of t vehicle n-2, v_n+2T () represents the speed of t vehicle n+2 Degree, G_n-2T () represents the time headway of t vehicle n and vehicle n-2, G_n+2T () represents the car of t vehicle n and vehicle n+2 Away from △ v during head_n-2T () represents the speed difference of t vehicle n and vehicle n-2, △ v_n+2T () represents t vehicle n and vehicle n+ 2 speed difference, G_T1T () represents the time headway of t driver desired vehicle n-2 and vehicle n, a₁, b₁, c₁Represent G_T1 About the linear combination coefficient of car speed in (t)；G_T2T () represents the car of t driver desired vehicle n+2 and vehicle n Away from a during head₂, b₂, c₂Represent G_T2About the linear combination coefficient of car speed in (t)；α represents lane-change vehicle in total acceleration In consideration degree to target track front truck.

5. the vehicle lane-changing overall process DAS (Driver Assistant System) based on augmented reality according to claim 3 it is characterised in that： In described lane-change implementation procedure, the lane-change execution model of foundation includes speed generation module, trajectory planning module and track life Become module.

6. the vehicle lane-changing overall process DAS (Driver Assistant System) based on augmented reality according to claim 5 it is characterised in that： Described speed generation module implements function such as：

(1) calculate the safe speed that lane-change vehicle n and current lane front truck n-1 does not collide：

$v_n^{{\mathrm{PV}}_c}(t+\mathrm{\τ})=b_n\mathrm{\τ}+\sqrt{b_n^2{\mathrm{\τ}}^2-b_n\[2\left(x_{n-1}\right(t)-l_{n-1}-x_n(t\left)\right)-v_n\left(t\right)\mathrm{\τ}-v_{n-1}^2\left(t\right)/b_{n-1}\]}$

In formula,It is longitudinal safe speed of vehicle n relatively current track front truck n-1, l_n-1It is current lane front truck n- 1 vehicle commander, b_n-1And b_nIt is the respective maximum brake acceleration of current lane front truck n-1 and vehicle n, x_nT () is t lane-change car The lengthwise position of n, x_n-1T () is the lengthwise position of t current lane front truck n-1, τ is the reaction time；

(2) calculate the safe speed that the front truck n-2 on lane-change vehicle n and target track and rear car n+2 do not collide：

$v_n^{{\mathrm{PV}}_t}(t+\mathrm{\τ})=b_n\mathrm{\τ}+\sqrt{b_n^2{\mathrm{\τ}}^2-b_n\[2\left(x_{n-2}\right(t)-l_{n-2}-x_n(t\left)\right)-v_n\left(t\right)\mathrm{\τ}-v_{n-2}^2\left(t\right)/b_{n-2}\]}$

$v_n^{{\mathrm{LV}}_t}(t+\mathrm{\τ})=\frac{b_n\mathrm{\τ}}{2}+\sqrt{{\left(\frac{b_n\mathrm{\τ}}{2}\right)}^2+b_n\[2\left(x_n\right(t)-l_n-x_{n+2}(t\left)\right)+v_n\left(t\right)\mathrm{\τ}-2v_{n+2}\left(t\right)\mathrm{\τ}+\frac{v_{n+2}^2\left(t\right)}{b_{n+2}}\]}$

In formula,WithIt is the vertical of lane-change vehicle n relative target track front truck n-2 and target track rear car n+2 To safe speed, l_n-2It is target track front truck n-2 vehicle commander, b_n-2It is the maximum brake acceleration of target track front truck n-2, x_n-2 T () is the lengthwise position of t target track front truck n-2, l_nIt is the vehicle commander of lane-change vehicle n, b_n+2It is target track rear car n+2 Maximum brake acceleration, x_n+2T () is the lengthwise position of t target track rear car n+2；

(3) calculate the final longitudinal velocity of lane-change vehicle：

$v_n(t+\mathrm{\τ})={{\mathrm{\αv}}_n}^{{\mathrm{PV}}_c}(t+\mathrm{\τ})+(1-\mathrm{\α})\[{{\mathrm{\βv}}_n}^{{\mathrm{PV}}_t}(t+\mathrm{\τ})+(1-\mathrm{\β}){v_n}^{{\mathrm{LV}}_t}(t+\mathrm{\τ})\]$

In formula, v_n(t+ τ) is the longitudinal velocity of lane-change vehicle n, α and β is weight coefficient, when β represents vehicle lane-changing execution, to mesh The weight that front truck n-2 in mark track considers, 1- β is the weight that rear car n+2 in target track is considered；α is this track car Weight coefficient on the impact of lane-change vehicle n speed, 1- α is the vehicle weight coefficient that vehicle n speed is affected of target vehicle；

(4) calculate the general speed that vehicle travels：

$u_n(t+\mathrm{\τ})=\frac{v_n(t+\mathrm{\τ})}{cos\left(\mathrm{\θ}\right(t+\mathrm{\τ}\left)\right)}$

In formula, u_n(t+ τ) is the general speed of lane-change vehicle n, and θ is the course angle of lane-change vehicle.

7. the vehicle lane-changing overall process DAS (Driver Assistant System) based on augmented reality according to claim 6 it is characterised in that： The computing formula of described α is as follows：

α=tan (p_n(t))/k

p_n(t)=y_f(t)/Y

Wherein, y_fT () represents lane-change vehicle n in the lateral attitude of t, Y is the lateral separation in target track and initial track, k It is calibrating parameters.

8. the vehicle lane-changing overall process DAS (Driver Assistant System) based on augmented reality according to claim 5 it is characterised in that： Described trajectory planning module implements function such as：

(1) set up parameter equation of locus undetermined：

$y_n\left(x\right)=a_0+a_1{x}_n+a_2{x}_n^2+a_3{x}_n^3$

Wherein, a₀, a₁, a₂, a₃It is parameter to be determined, x_nFor the position of longitudinally upper lane-change vehicle n, y_nFor transversely lane-change car The position of n；

(2) calculate each parameter of equation of locus：

The start position of each step-length is defined as (0,0) point, final position is defined as (x_f,y_f), the vehicle of arbitrary step-length starting point Course angle is θ_i, the vehicle course angle of terminal is 0, then have y '_n(0)=tan θ_i, y '_n(x_f)=0；Then substitute into equation of locus to ask The expression formula obtaining each parameter is as follows：

a₁=tan θ_i

a₀=0

$a_2=\frac{3y_f-2x_f{\mathrm{tan\θ}}_i}{{x_f}^2}$

$a_3=\frac{x_f{\mathrm{tan\θ}}_i-2y_f}{{x_f}^3}$

(3) obtain the equation of locus of parameter determination：

$y_n\left(x\right)=x_n{\mathrm{tan\θ}}_i+\frac{3y_f-2x_f{\mathrm{tan\θ}}_i}{x_f^2}{x_n}^2+\frac{x_f{\mathrm{tan\θ}}_i-2y_f}{x_f^3}{x_n}^3$

(4) cost function of construction lane-change execution：

$J=\mathrm{\ω}{\left(\frac{a_s^{last}}{a_s^{\max }}\right)}^2+(1-\mathrm{\ω})\frac{x_f}{x_f^{\max }}$

In formula, J is cost function,Refer to the side acceleration of destination county in a lane-change perform track, be in track Maximum side acceleration,Refer to the maximum side acceleration of all vehicles in lane-change implementation procedure,Refer to all vehicles The maximum fore-and-aft distance running in lane-change implementation procedure, ω is the weight parameter needing to demarcate；

Cost function J is in the side acceleration of destination countyCalculated by following formula：

$a_s^{last}=u_n^{2}K\left(x_f\right)$

In formula, K () is the curvature function with regard to longitudinal direction of car position of lane-change perform track；K () and the pass of equation of locus It is to be：

$K=\left|\frac{y_n^{\′\′}}{{(1+y_n^{\′2})}^{\frac{3}{2}}}\right|$

Wherein, y ' and y " is respectively single order and the second dervative of lane-change perform track equation：

$y_n^{\′}\left(x\right)={\mathrm{tan\θ}}_i+2\frac{3y_f-2x_f{\mathrm{tan\θ}}_i}{{x_f}^2}{x}_n+3\frac{x_f{\mathrm{tan\θ}}_i-2y_f}{{x_f}^3}{x}_n^2$

$y_n^{\′\′}\left(x\right)=2\frac{3y_f-2x_f{\mathrm{tan\θ}}_i}{{x_f}^2}+6\frac{x_f{\mathrm{tan\θ}}_i-2y_f}{{x_f}^3}{x}_n$

Substitute into one, second dervative to K () x relevant with drawing curvature K in the relation of equation of locus_nFunction (0≤x_n≤x_f)：

$K=\left|\frac{2\frac{3y_f-2x_f{\mathrm{tan\θ}}_i}{{x_f}^2}+6\frac{x_f{\mathrm{tan\θ}}_i-2y_f}{{x_f}^3}{x}_n}{{(1+{\left({\mathrm{tan\θ}}_i+2\frac{3y_f-2x_f{\mathrm{tan\θ}}_i}{{x_f}^2}{x}_n+3\frac{x_f{\mathrm{tan\θ}}_i-2y_f}{{x_f}^3}{x}_n^2\right)}^2)}^{\frac{3}{2}}}\right|$

Then obtain：

$a_s^{last}=u_n^2\left|\frac{2x_f{\mathrm{tan\θ}}_i-6y_f}{{x_f}^2}\right|$

The final expression formula obtaining cost function J further is as follows：

$J={\mathrm{\ω}}_1{\left(\frac{{u_n}^2\left|\frac{2x_f{\mathrm{tan\θ}}_i-6y_f}{{x_f}^2}\right|}{a_c^{\max }}\right)}^2+{\mathrm{\ω}}_2\frac{x_f}{x_f^{\max }}$

The corresponding x when J gets minimum of a value_fIt is the lengthwise position of this step-length optimal trajectory terminal；In lane-change implementation procedure, Each step-length can produce an optimal trajectory.

9. the vehicle lane-changing overall process DAS (Driver Assistant System) based on augmented reality according to claim 8 it is characterised in that： Described Track Pick-up module implements function such as：

(1) calculate in a step-length, arc length L that vehicle runs over along the optimal trajectory planning out：

L=u_n(t)τ

Meanwhile, lane-change perform track equation is used rectangular area integration quad method calculate arc length：

$\begin{array}{c}L={\∫}_0^{X}ds={\∫}_0^X\sqrt{1+y^{\′2}}dx\\ ={\∫}_0^X\sqrt{1+{({\mathrm{tan\θ}}_i+2\frac{3y_f-2x_f{\mathrm{tan\θ}}_i}{{x_f}^2}{x}_n+3\frac{x_f{\mathrm{tan\θ}}_i-2y_f}{{x_f}^3}{x}_n^2)}^2}dx\end{array}$

In formula, X is the longitudinal coordinate that this step-length terminates, and is also the longitudinal coordinate that next step-length starts simultaneously；

Then basisTry to achieve the value of X, and It is updated to the lateral coordinates being calculated in equation of locus at the end of step-length；

(2) course angle of the vehicle at the end of material calculation：

The single order derived function making equation of locus is：

$tan\mathrm{\θ}={\mathrm{tan\θ}}_i+2\frac{3y_f-2x_f{\mathrm{tan\θ}}_i}{{x_f}^2}{x}_n+3\frac{x_f{\mathrm{tan\θ}}_i-2y_f}{{x_f}^3}{x}_n^2$

In formula, θ is the vehicle course angle of any point in lane-change perform track；

Make x_n=X, then obtain the course angle of vehicle at the end of current step：

${\mathrm{\θ}}_f=\arctan \left({\mathrm{tan\θ}}_i+2\frac{3y_f-2x_f{\mathrm{tan\θ}}_i}{{x_f}^2}X+3\frac{x_f{\mathrm{tan\θ}}_i-2y_f}{{x_f}^3}{X}^2\right)$

So far, obtaining the vehicle-state at the end of current step, thus obtaining vehicle-state when next step-length starts, passing through After successive ignition, the local path of each step-length is end to end, reaches target vehicle center line, forms a complete lane-change execution Track.