A preview decision method for speed following of autonomous vehicles

Abstract

The present invention discloses a preview decision method for speed following of an autonomous vehicle, comprising the following steps: selecting a preview target and determining a preview distance; based on the preview target and the preview distance, using the mileage on the target path as the longitudinal motion distance, calculating a preview tracking error based on the preview speed and the expected speed; controlling the preview tracking error within a preset following accuracy error threshold using a multi-segment preview method based on motion primitives to obtain a preview result; determining an expected longitudinal acceleration based on the preview result, following the vehicle's longitudinal acceleration using an expected longitudinal acceleration following method, and performing load compensation on the vehicle's longitudinal acceleration; and selecting an actuation phase of the vehicle's chassis mechanism based on the compensated following acceleration to determine a control command. The present invention is capable of following a speed change driving plan planned based on space requirements.

Description

Technical Field

[0001] The present invention relates to the technical field of autonomous driving vehicles, and in particular to a preview decision method for speed following of an autonomous driving vehicle. Background Art

[0002] With the advancement of science and technology, autonomous vehicles have become one of the fastest-growing and most anticipated technological growth areas in the automotive industry. Speed ​​tracking is a key feature required by autonomous vehicles. Autonomous vehicles are already commercially deployed in urban settings, where they must effectively navigate various traffic regulations, such as speed limit signs, stop signs, and traffic lights. These regulations impose speed requirements based on spatial locations. Autonomous vehicles in urban settings must adhere to these spatial speed requirements. Current autonomous driving technologies mostly implement speed tracking methods based on time to complete driving plans. However, these spatial speed requirements are less than ideal, primarily due to significant speed tracking lag, large speed fluctuations, and significant speed errors at specific locations. Summary of the Invention

[0003] The present invention proposes a preview decision method for speed following of an autonomous vehicle to address the shortcomings of existing autonomous vehicles in their ability to follow a speed target set based on space.

[0004] To achieve the above objectives, the present invention provides a preview decision method for speed following of an autonomous vehicle, comprising:

[0005] Selecting a preview target and determining a preview distance, wherein the preview target is any segment in the to-be-followed speed list provided by the upstream vehicle speed planning function module of the system;

[0006] Based on the preview target and the preview distance, the mileage on the target path is used as the longitudinal movement distance, and a preview tracking error is calculated according to the preview vehicle speed and the expected vehicle speed. The preview tracking error is controlled within a preset following accuracy error threshold range through a multi-segment preview method based on motion primitives to obtain a preview result.

[0007] Determining an expected longitudinal acceleration according to the preview result, following the vehicle longitudinal acceleration using an expected longitudinal acceleration following method, and performing load compensation on the vehicle longitudinal acceleration;

[0008] The actuation phase of the vehicle chassis mechanism is selected according to the compensated following acceleration, and the control command amount is determined.

[0009] Preferably, selecting the preview target and determining the preview distance includes:

[0010] The Frenet coordinate system is established with the expected following path as the s-axis. The geodetic coordinates of the main vehicle at the current moment are (X e , Y e ), the vehicle speed is V, and the preview time is T P ;

[0011] According to the conversion relationship between the current earth coordinate system and the frenet coordinate system, the earth coordinate of the main vehicle is converted into the frenet coordinate (s e , d e ), the preview target is obtained as [s e , s Tp ], where s Tp =s e +d; d is the preview distance, d=max(VT P , Δs min ), Δs min The shortest sight distance.

[0012] Preferably, calculating the preview tracking error includes:

[0013] err i =|V i -V preview，i |;

[0014] Where err i is the preview tracking error, V i is the expected speed corresponding to the mileage, V preview，i To preview the vehicle speed;

[0015]

[0016] Where V start is the starting speed, a x is the preview acceleration, s i is the mileage of the preview point, s1 is the starting mileage, V N is the speed of the vehicle at the last speed point in the speed list, and Δs is the mileage difference between the start and end points.

[0017] Preferably, controlling the preview tracking error within a preset error range by a multi-segment preview method based on motion primitives includes:

[0018] Update the list of speeds to be followed according to the preview speed points;

[0019] Get the preset shortest distance information and determine whether the mileage covered by the list is greater than the shortest distance. If so, proceed to the next step, otherwise return;

[0020] Set a following accuracy error threshold, perform a preview of the motion primitive, i.e., the expected acceleration, and determine whether the following error is less than the threshold under the description of the motion primitive. If so, continue inserting subsequent speed points. Otherwise, insert the preview result of the previous step into the acceleration / deceleration-mileage list, and delete all speed points except the last two in the following speed point list.

[0021] Repeat the above steps until the speed list to be followed is empty, and the acceleration / deceleration intensity-mileage list [(a x1 ,s1)),…),(a xn , s n ))] is the decision result, where a xi is the preview acceleration of the i-th segment, s i is the starting mileage of the i-th segment, i=1,2,…,n;

[0022] Finally, the current mileage signal of the controlled vehicle is obtained, and the expected longitudinal acceleration is determined based on the acceleration and deceleration intensity-mileage list to be executed and the current mileage of the autonomous driving vehicle.

[0023] Preferably, determining the expected longitudinal acceleration comprises:

[0024]

[0025] Where, is the expected longitudinal acceleration, a xi is the preview acceleration of the i-th segment, s i is the mileage of the preview point, s e It is the current mileage signal of the controlled vehicle.

[0026] Preferably, following the vehicle longitudinal acceleration by the expected longitudinal acceleration following method includes:

[0027] A closed-loop acceleration tracking module with direct feedforward is used. When there is an error in acceleration tracking, the error is eliminated through closed-loop tracking. When the feedback sensor fails, the closed-loop module is disconnected, and the acceleration tracking method is degraded to open-loop compensation correction to ensure normal driving of the vehicle.

[0028] Among them, anti-integral saturation PID is used as the closed-loop tracking method, specifically:

[0029]

[0030] Where, e is the acceleration error, K p , K I , K D are proportional coefficient, integral coefficient, differential coefficient, K b is the inverse calculation coefficient, Δu is the difference between the output control quantity and the directly calculated control quantity, Δax It is the interpolation between the ideal acceleration and the current actual acceleration.

[0031] Preferably, performing load compensation on the longitudinal acceleration of the vehicle includes:

[0032] F c =F slope +F wind +F friction +F access_load ;

[0033]

[0034] Where, F c F is the total resistance that needs to be overcome for the longitudinal movement of the vehicle. slope is the slope resistance, F wind is the air resistance, F friction is the rolling resistance, F access_load is the equivalent resistance generated by the electrical load, f c is the load specific force, and M is the vehicle mass.

[0035] Preferably, selecting the actuation phase of the vehicle chassis mechanism according to the compensated following acceleration includes:

[0036]

[0037] Where phase is the working phase of the vehicle's longitudinal motion chassis system, f * For the comparison after compensation, is the minimum specific force of the driving phase, is the maximum specific force of the driving phase, is the minimum specific force of the braking phase, is the maximum specific force of the braking phase.

[0038] Preferably, the final drive force is determined or brake specific force According to the driving force and the vehicle speed steady value V to determine the accelerator pedal opening α T , according to the braking force and the vehicle speed steady value V to determine the brake master cylinder pressure P B .

[0039] Preferably, the final drive force is determined or brake specific force Specifically:

[0040]

[0041] Determine α T Specifically:

[0042]

[0043] Determine P B Specifically:

[0044]

[0045] Where, VSD -1 is the inverse model of vehicle system dynamics, α T is the accelerator pedal opening, P B is the brake master cylinder pressure.

[0046] Compared with the prior art, the present invention has the following advantages and technical effects:

[0047] This invention proposes a preview decision-making method for speed following in autonomous vehicles. This method is capable of following speed-changing driving plans planned based on spatial requirements. Using a spatial preview method, the expected vehicle speed is decomposed into multiple segments in space using motion primitives. The segmentation points are determined based on the preview error to ensure that the preview following error is within the allowable error range, thereby reducing the degradation of driving stability and comfort caused by frequent driving operations. The expected longitudinal acceleration is determined based on the preview results, and an expected longitudinal acceleration following method is established to follow the vehicle's longitudinal acceleration, ensuring the rapid achievement of the expected speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0049] Figure 1 A schematic diagram of preview target selection according to an embodiment of the present invention;

[0050] Figure 2 Schematic diagram of target speed and preview speed in single-stage preview following according to an embodiment of the present invention;

[0051] Figure 3 Schematic diagram of speed preview error in single-stage preview following according to an embodiment of the present invention;

[0052] Figure 4 Flowchart of a multi-segment preview method based on motion primitives according to an embodiment of the present invention;

[0053] Figure 5 This is an overall block diagram of the longitudinal acceleration following according to an embodiment of the present invention. DETAILED DESCRIPTION

[0054] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0055] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0056] This embodiment proposes a preview decision method for speed following of an autonomous vehicle, including:

[0057] Select a preview target and determine the preview distance, where the preview target is any segment in the list of speeds to be followed provided by the upstream vehicle speed planning function module of the system;

[0058] Based on the preview target and the preview distance, the mileage on the target path is used as the longitudinal movement distance, the preview tracking error is calculated according to the preview vehicle speed and the expected vehicle speed, and the preview tracking error is controlled within a preset following accuracy error threshold range through a multi-segment preview method based on motion primitives to obtain a preview result;

[0059] Determine the expected longitudinal acceleration based on the preview result, follow the vehicle's longitudinal acceleration using the expected longitudinal acceleration following method, and perform load compensation on the vehicle's longitudinal acceleration;

[0060] The actuation phase of the vehicle chassis mechanism is selected according to the compensated following acceleration, and the control command amount is determined.

[0061] Specifically, this embodiment can follow speed-varying driving plans planned based on spatial requirements. Using a spatial preview method, the expected speed is decomposed into multiple spatial segments using motion primitives. The segmentation points are determined based on the preview error to ensure that the preview-following error is within the allowable error range, thereby reducing the degradation of driving stability and comfort caused by frequent driving operations. Based on the preview results, the expected longitudinal acceleration is determined, and an expected longitudinal acceleration following method is established to track the vehicle's longitudinal acceleration, ensuring rapid achievement of the expected speed.

[0062] Furthermore, selecting a preview target and determining a preview distance include:

[0063] The Frenet coordinate system is established with the expected following path as the s-axis. The geodetic coordinates of the main vehicle at the current moment are (X e ,Y e ), the vehicle speed is V, and the preview time is T P ;

[0064] According to the conversion relationship between the current earth coordinate system and the frenet coordinate system, the earth coordinate of the main vehicle is converted into the frenet coordinate (s e , d e ), the preview target is obtained as [se ,s Tp ], where s Tp =s e +d; d is the preview distance, d=max(VT P ,Δs min ), Δs min The shortest sight distance.

[0065] Specifically, the speeds that the driver wants to follow vary. In order to uniformly express the target speed, this embodiment describes the desired speed in the form of a discrete point sequence. When the target speed to be followed is given in the form of a spatial sequence, (V i ),s i ) to describe the expected vehicle speed, s i is the mileage on the desired path, V i is the expected speed corresponding to the mileage.

[0066] When a person is driving, the expected target will move forward along the expected driving path. Therefore, it is assumed that the preview target is always selected along the direction of the follow-up path, and a preview target selection method is proposed based on this. Figure 1 As shown in the figure, the Frenet coordinate system is established with the expected following path as the s axis. The current earth coordinate of the main vehicle is (X e ,Y e ), the vehicle speed is V, the preview time is T P , d=max(VT P ,Δs min ) is the preview distance, where Δs min is the shortest sight distance. First, the main vehicle's geodetic coordinates are converted into frenet coordinates (s e , d e ). The preview target is [s e ,s Tp ] The following target between mileages, where s Tp =s e +d. This means the mileage distance in the Frenet coordinate system is used as the preview distance. Clearly, when the mileage difference remains constant and the curvature of the expected following path increases, the straight-line distance from the driver to the preview point decreases. This is consistent with the characteristic that when a driver is driving a vehicle, the greater the road curvature, the smaller the field of view, and the shorter the forward distance.

[0067] Furthermore, the preview tracking error is calculated, including:

[0068] For vehicle speed following, constant longitudinal acceleration motion is used as the longitudinal motion primitive. When the speed following target is in the speed-mileage sequence [(V1, s1)),…), (V N , s NGiven , the lateral motion follows the target path with little error, so the mileage on the target path is used as the longitudinal motion distance. The longitudinal acceleration is calculated as shown in Equation (1). The preview speed formula is shown in Equation (2).

[0069]

[0070] Where V start is the starting speed, a x is the preview acceleration, V N is the speed of the vehicle at the last speed point in the speed list, and Δs is the mileage difference between the start and end points.

[0071]

[0072] Where a x is the preview acceleration, s i is the mileage of the preview point, s1 is the starting mileage, V preview，i To preview the vehicle speed.

[0073] Calculate the preview tracking error based on the preview speed and the expected target speed:

[0074] err i =|V i -V preview , i |(3);

[0075] Where err i is the preview tracking error, V i is the expected speed corresponding to the mileage.

[0076] When using a single-segment motion primitive for previewing, if the target to be followed does not match the motion primitive, there may be a large error between the preview speed and the target speed, such as Figure 2 and Figure 3 shown.

[0077] Furthermore, the preview tracking error is controlled within a preset error range by a multi-segment preview method based on motion primitives, including:

[0078] Update the list of speeds to be followed according to the preview speed points;

[0079] Get the preset shortest distance information and determine whether the mileage covered by the list is greater than the shortest distance. If so, proceed to the next step, otherwise return;

[0080] Set a following accuracy error threshold, perform a preview of the motion primitive, i.e., the expected acceleration, and determine whether the following error is less than the threshold under the description of the motion primitive. If so, continue inserting subsequent speed points. Otherwise, insert the preview result of the previous step into the acceleration / deceleration-mileage list, and delete all speed points except the last two in the following speed point list.

[0081] Repeat the above steps until the speed list to be followed is empty, and the acceleration / deceleration intensity-mileage list [(a x1 ,s1)),…),(a xn , s n ))] is the decision result; the acceleration and deceleration intensity-mileage list is obtained in sequence according to the advancement of the preview target, s i is the starting mileage of the i-th segment, i=1,2,…,n, a xi is the preview acceleration of the i-th segment.

[0082] Finally, the current mileage signal of the controlled vehicle is obtained, and the expected longitudinal acceleration is determined based on the acceleration and deceleration intensity-mileage list to be executed and the current mileage of the autonomous driving vehicle.

[0083] Specifically, to address the problem of excessive error when the tracking target and the motion primitives do not match, a multi-segment preview method based on motion primitives is proposed. Given a preview target, the preview target is previewed using motion primitives, and the segmentation points are determined based on the preview error. This process can positively control the preview error within a reasonable error range.

[0084] First, the speed-mileage point (V i ,s i ) is inserted into the following speed mileage list. Since the car needs to travel a sufficient distance to perform maneuvers, it is determined whether the difference between the first and last mileage points of the following speed point exceeds the minimum distance required for maneuvering. If the distance is insufficient, speed points are continuously inserted until the mileage difference exceeds the minimum distance required for maneuvering. When the distance is sufficient, the speed is previewed. The preview method is shown in Equations (1) and (2).

[0085] If the preview tracking error (calculated as formula (3)) is less than the allowable error, continue to insert more speed points, otherwise the preview result of the previous step ((a xi , s i )), where a xi is the acceleration of the i-th segment, s i Insert the acceleration / deceleration intensity-mileage list for the starting mileage of the i-th segment, and delete the remaining speed points except the last two speed points in the follow speed point list. Repeat the above steps until the speed list to be followed is empty. Acceleration / deceleration intensity-mileage list [(a x1,s1),…,(a xn , s n )] is the decision result.

[0086] According to the acceleration and deceleration intensity mileage list to be executed [(a x1 ,s1),…,(a xn , s n )], and the current mileage of the autonomous vehicle s e Determine the expected longitudinal acceleration.

[0087]

[0088] Where, is the expected longitudinal acceleration, a xi is the preview acceleration of the i-th segment, s i is the mileage of the preview point, s e It is the current mileage signal of the controlled vehicle.

[0089] Furthermore, the vehicle longitudinal acceleration is followed by an expected longitudinal acceleration following method, including:

[0090] A closed-loop acceleration tracking module with direct feedforward is used. When there is an error in acceleration tracking, the error can be eliminated through closed-loop tracking. When the feedback sensor fails, due to the use of a closed-loop method with direct feedforward, the acceleration tracking method can be degraded to an open-loop compensation correction by disconnecting the closed-loop module, ensuring that the car can drive normally.

[0091] This embodiment uses anti-integral windup PID as a closed-loop tracking method, specifically:

[0092]

[0093] Where, e is the acceleration error, K p , K I , K D are proportional coefficient, integral coefficient, differential coefficient, K b is the inverse calculation coefficient, Δu is the difference between the output control quantity and the directly calculated control quantity, Δa x It is the interpolation between the ideal acceleration and the current actual acceleration.

[0094] Furthermore, load compensation is performed on the longitudinal acceleration of the vehicle, including:

[0095] A car needs to overcome various resistances during driving. When the car works in different working conditions, the driving force of the car is the same when the control quantity is the same, but the acceleration is different at this time. In order to make the method applicable to all working conditions, load compensation is performed on the acceleration, specifically:

[0096] Fc =F slope +F wind +F friction +F access_load (6);

[0097]

[0098] Where, F c F is the total resistance that needs to be overcome for the longitudinal movement of the vehicle. slope is the slope resistance, F wind is the air resistance, F friction is the rolling resistance, F access_load is the equivalent resistance generated by the electrical load, f c is the load specific force, and M is the vehicle mass.

[0099] Furthermore, based on the compensated specific force, the current steady-state speed of the autonomous vehicle, and the operating range of each operating phase, the operating phase is determined to select and activate different assemblies of the vehicle chassis to achieve different operating states such as driving, coasting, and braking. This includes:

[0100]

[0101] Where phase is the working phase of the vehicle's longitudinal motion chassis system, f * For the comparison after compensation, is the minimum specific force of the driving phase, is the maximum specific force of the driving phase, is the minimum specific force of the braking phase, is the maximum specific force of the braking phase.

[0102] Furthermore, the final drive force is determined or brake specific force According to the driving force and the vehicle speed steady value V to determine the accelerator pedal opening α T , according to the braking force and the vehicle speed steady value V to determine the brake master cylinder pressure P B .

[0103] Determine the final drive force or brake specific force Specifically:

[0104]

[0105] Determine α T Specifically:

[0106]

[0107] Determine PB Specifically:

[0108]

[0109] Where, VSD -1 is the inverse model of vehicle system dynamics, α T is the accelerator pedal opening, P B is the brake master cylinder pressure.

[0110] In order to more clearly express the technical solution of the present invention, the following specific embodiments are provided to introduce the solution:

[0111] S1. Preview the expected acceleration:

[0112] S2. Select a preview point based on the expected speed plan to be followed;

[0113] First, obtain the following information: the system preset preview time T P , shortest viewing distance Δs min ; The conversion relationship between the current geodetic coordinate system and the frenet coordinate system.

[0114] Secondly, the following signals are obtained: the expected speed plan to be followed is obtained from the leading module, which consists of a sequence of mileage and corresponding speed data pairs; the current vehicle status, such as position and speed, is obtained from the controlled vehicle.

[0115] Calculate preview distance d = max (VT P ,Δs min ), which means the mileage length in the Frenet coordinate system. Based on the preview distance, the corresponding preview speed (V i ,s i ).

[0116] S3, determine the ideal acceleration based on the preview point velocity;

[0117] like Figure 4 As shown:

[0118] First, update the speed list to be followed according to the preview speed point.

[0119] Secondly, obtain the system preset shortest travel distance information and determine whether the mileage covered by the list is greater than the shortest travel distance. If it is greater, continue to the next step, otherwise return.

[0120] Then, the system preset tracking accuracy error threshold information is obtained, and the motion primitive, i.e., the expected acceleration, is previewed according to formulas (1) and (2). It is determined whether the following error is less than the threshold under the description of the primitive. If it is, the subsequent speed points are inserted. Otherwise, the preview result of the previous step is inserted into the acceleration / deceleration-mileage list, and the remaining speed points in the following speed point list except the last two speed points are deleted.

[0121] Repeat the above steps until the speed list to be followed is empty. Acceleration / deceleration intensity-mileage list [(a x1 ,s1),…,(a xn , s n )] is the decision result.

[0122] Finally, obtain the current mileage signal s of the controlled vehicle e ,determine the expected longitudinal acceleration based on the acceleration / deceleration intensity-mileage list to be executed and the current mileage of the autonomous vehicle.

[0123] S4. Acceleration following:

[0124] like Figure 5 The figure shows the overall framework of the longitudinal acceleration following of this embodiment, specifically S5-S7.

[0125] Among them, V is the steady speed value of the autonomous vehicle; is the expected longitudinal acceleration; a x is the actual longitudinal acceleration of the vehicle; f c is the load compensation specific force; f * is the total specific force demand; The driving force of the distribution; is the distributed braking force; α T is the accelerator pedal opening; P B is the brake master cylinder pressure; specific force is the ratio of force to vehicle mass, and has the same dimension as acceleration.

[0126] S5. Determine PID following acceleration based on the ideal acceleration;

[0127] The longitudinal acceleration signal of the current controlled vehicle is obtained, the error between it and the expected longitudinal acceleration is calculated, and the closed-loop acceleration tracking module with direct feedforward is executed to decide the following acceleration.

[0128] S6. Determine the driving force requirement corresponding to the current load and perform following acceleration compensation;

[0129] Execute the load compensation function module, first obtain the current road slope signal i, and calculate the slope resistance:)

[0130] F slope =Mg sin i(13);

[0131] Where g is the standard acceleration due to gravity.

[0132] Then calculate the rolling resistance:

[0133] F friction =Mgk cos i(14);

[0134] Where k is the pre-calibrated rolling resistance coefficient.

[0135] After that, get the current vehicle speed V and calculate the air resistance:

[0136]

[0137] Where C D is the drag coefficient, and A is the frontal area.

[0138] Finally, obtain the curve of the equivalent resistance generated by the pre-calibrated electrical load according to the vehicle speed, and determine F by looking up the table based on the vehicle speed. access_load .

[0139] The load compensation amount is calculated according to formulas (6) and (7).

[0140] S7. Select the actuation phase of the vehicle chassis mechanism based on the compensated following acceleration and determine the control command amount.

[0141] Get the pre-calibrated driving, coasting and braking threshold value curves with vehicle speed, and determine the value based on the current vehicle speed. and

[0142] Determine the working phase to be selected at present and determine the final driving force according to formulas (9) and (10) or brake specific force

[0143] According to the accelerator pedal opening α calibrated in advance based on the vehicle field test T The inverse model of the controlled vehicle powertrain system identified by the longitudinal acceleration change curve is as shown in formula (11), which realizes the driving steady-state compensation and determines α T .

[0144] According to the brake master cylinder pressure P pre-calibrated according to the vehicle field test B The inverse model of the controlled vehicle power transmission system identified by the longitudinal deceleration change curve is as shown in formula (12) to achieve steady-state compensation and determine P B , output α T or P B。

[0145] The above are merely preferred embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A preview decision method for speed following of an autonomous vehicle, characterized in that: include: Selecting a preview target and determining a preview distance, wherein the preview target is any segment in the to-be-followed speed list provided by the upstream vehicle speed planning function module of the system; Based on the preview target and the preview distance, the mileage on the target path is used as the longitudinal movement distance, and a preview tracking error is calculated according to the preview vehicle speed and the expected vehicle speed. The preview tracking error is controlled within a preset following accuracy error threshold range through a multi-segment preview method based on motion primitives to obtain a preview result. Determining an expected longitudinal acceleration according to the preview result, following the vehicle longitudinal acceleration using an expected longitudinal acceleration following method, and performing load compensation on the vehicle longitudinal acceleration; Select the actuation phase of the vehicle chassis mechanism according to the compensated following acceleration and decide the control command amount; Calculating the preview tracking error includes: ； Where, is the preview tracking error, is the expected speed corresponding to the mileage, To preview the vehicle speed; ； ； Where, is the starting speed, is the preview acceleration, For the preview point mileage, is the starting mileage, The speed of the vehicle at the last speed point in the following speed list, is the difference in mileage between the start and end points; Controlling the preview tracking error within a preset error range by a multi-segment preview method based on motion primitives includes: Update the list of speeds to be followed according to the preview speed points; Get the preset shortest distance information and determine whether the mileage covered by the list is greater than the shortest distance. If so, proceed to the next step, otherwise return; Set a following accuracy error threshold, perform a preview of the motion primitive, i.e., the expected acceleration, and determine whether the following error is less than the threshold under the description of the motion primitive. If so, continue inserting subsequent speed points. Otherwise, insert the preview result of the previous step into the acceleration / deceleration-mileage list, and delete all speed points except the last two in the following speed point list. Repeat the above steps until the speed list to be followed is empty, and the acceleration / deceleration-mileage list [( ), ,( )] is the decision result, where For the Segment preview acceleration, For the Segment starting mileage, ; Finally, the current mileage signal of the controlled vehicle is obtained, and the expected longitudinal acceleration is determined based on the acceleration / deceleration-mileage list to be executed and the current mileage of the autonomous driving vehicle.

2. The preview decision method for speed following of an autonomous vehicle according to claim 1, characterized in that: Selecting the preview target and determining the preview distance includes: The Frenet coordinate system is established with the expected following path as the s-axis. The geodetic coordinates of the host vehicle at the current moment are ( , ), the vehicle speed is V , the preview time interval is ; According to the conversion relationship between the current earth coordinate system and the frenet coordinate system, the earth coordinate of the main vehicle is converted into the frenet coordinate ( , ), the preview target is obtained as [ , ],in, ; d is the preview distance, , The shortest sight distance.

3. The preview decision method for speed following of an autonomous vehicle according to claim 1, characterized in that: Determining the expected longitudinal acceleration includes: ； Where, is the expected longitudinal acceleration, For the Segment preview acceleration, For the preview point mileage, It is the current mileage signal of the controlled vehicle.

4. The preview decision method for speed following of an autonomous vehicle according to claim 3, characterized in that: Following the vehicle longitudinal acceleration by the expected longitudinal acceleration following method includes: A closed-loop acceleration tracking module with direct feedforward is used. When there is an error in acceleration tracking, the error is eliminated through closed-loop tracking. When the feedback sensor fails, the closed-loop module is disconnected, and the acceleration tracking method is degraded to open-loop compensation correction to ensure normal driving of the vehicle. Among them, anti-integral saturation PID is used as the closed-loop tracking method, specifically: ； Where, is the acceleration error, , , are the proportional coefficient, integral coefficient, and differential coefficient respectively. is the inverse calculation coefficient, is the difference between the output control quantity and the directly calculated control quantity, It is the difference between the ideal acceleration and the current actual acceleration.

5. The preview decision method for speed following of an autonomous vehicle according to claim 1, characterized in that: Performing load compensation on the longitudinal acceleration of the vehicle includes: ； ； Where, It is the total resistance that needs to be overcome for the longitudinal movement of the vehicle. is the slope resistance, is the air resistance, is the rolling resistance, is the equivalent resistance generated by the electrical load, is the load specific force, For the vehicle quality.

6. The preview decision method for speed following of an autonomous vehicle according to claim 1, characterized in that: The selection of the vehicle chassis mechanism actuation phase according to the compensated following acceleration includes: ； Where, It is the working phase of the vehicle's longitudinal motion chassis system. For the comparison after compensation, is the minimum specific force of the driving phase, is the maximum specific force of the driving phase, is the minimum specific force of the braking phase, is the maximum specific force of the braking phase.

7. The preview decision method for speed following of an autonomous vehicle according to claim 6, characterized in that: Determine the final drive force or brake specific force , according to the driving force and vehicle speed stability value Determine the accelerator pedal opening , according to the braking force and vehicle speed stability value Determining the brake master cylinder pressure .

8. The preview decision method for speed following of an autonomous vehicle according to claim 7, characterized in that: Determine the final drive force or brake specific force Specifically: ； ； Sure Specifically: ； Sure Specifically: ； Where, is the inverse dynamics model of the vehicle system, is the accelerator pedal opening, is the brake master cylinder pressure.

Images