Dynamic speed planning method for autonomous vehicle and corresponding system

Abstract

Dynamic speed planning method (100) for an autonomous vehicle, executed to plan the best speed curve (110) of the autonomous vehicle, wherein the dynamic speed planning method (100) for the autonomous vehicle comprises: the execution of an information storage step (S02) to cause a memory to store obstacle information (102) about an obstacle, road information (102) and vehicle information about the autonomous vehicle, wherein the vehicle information includes a jerk threshold and a jerk switching frequency threshold, the execution of a step to calculate an acceleration limit (S04) to cause a data processing unit to receive the vehicle information (102) from memory and to calculate the vehicle information after a calculation procedure (104) to generate an acceleration limit range of the autonomous vehicle, the execution of a step to generate acceleration combinations (S06) to cause the data processing unit to receive the obstacle information and the road information from memory and to plan an acceleration interval of the autonomous vehicle according to the obstacle information, the road information and the acceleration limit range (104) and then to generate several acceleration combinations (106) of the autonomous vehicle according to the acceleration interval, the execution of an acceleration filtering step (S08) to cause the data processing unit to filter the acceleration combinations (106) according to the jerk threshold and the jerk switching frequency threshold to obtain a selected acceleration combination (108), and the execution of an acceleration smoothing step (S10) to cause the data processing unit to execute a driving behavior procedure to adjust the selected acceleration combination (108) to generate the best speed curve (110).

Description

BACKGROUNDField of expertise

[0001] This description relates to a dynamic speed planning method for an autonomous vehicle and a corresponding system. In particular, this description relates to a dynamic speed planning method for an autonomous vehicle with human-like driving behavior and a corresponding system. Description of the previously known technology

[0002] As autonomous vehicles receive increased attention, many automakers have invested in autonomous vehicle development, and several governments are planning operational mass transit systems using autonomous vehicles. Experimental autonomous vehicles have been approved in some countries.

[0003] The advanced driver assistance systems (ADAS) and autonomous driving systems (ADS) of current autonomous vehicles lack adaptability to the external environment. For example, the cruise control of ADAS relies primarily on fully automated longitudinal guidance (ACC), but ACC lacks speed or acceleration planning and cannot predict future interactions with other vehicles. Furthermore, while ADS also has speed planning, it only considers the vehicle's operating limits, while ADS does not consider acceleration and the limits of jerky movement of the vehicle's dynamics. Patent DE 11 2009 005 449 T5 describes a driving control device that controls the movement of a vehicle.The travel control device includes an acquisition unit that acquires a stabilization speed pattern indicating an upper speed limit up to which the vehicle can travel without deviating from a planned travel route. Furthermore, the travel control device includes a travel control setting unit that sets a travel control using a target speed pattern indicating a target speed based on the stabilization speed limit pattern and an actual speed.Patent DE 11 2017 001 905 T5 describes a vehicle speed control system with a lateral inclination detection system. The lateral inclination detection system comprises detection means that receives terrain information indicating the topography of an area extending in front of the vehicle from a terrain data acquisition means arranged to acquire data relating to terrain in front of the vehicle using one or more sensors. The processing means is configured to generate, depending on a predicted path of the vehicle through the terrain, information for the predicted path of the vehicle indicating a lateral inclination angle of the predicted path, which corresponds to the inclination of the predicted path transverse to the direction of travel of the vehicle.The vehicle speed control system is configured to control the vehicle speed at least partly in dependence on the information indicating the angle of lateral inclination for the anticipated path generated by the lateral inclination detection system.

[0004] It is therefore clear that the existing ADAS and ADS on the market lack the ability to plan and respond to acceleration. How to develop a dynamic speed planning method for an autonomous vehicle and a system based on vehicle dynamics and the external environment is highly anticipated by the public and will be the target of considerable industry effort. SUMMARY

[0005] In one aspect of the present description, a dynamic speed planning method for an autonomous vehicle is implemented to plan the optimal speed curve of the autonomous vehicle. The dynamic speed planning method for an autonomous vehicle includes executing an information storage step, an acceleration limit calculation step, an acceleration combination generation step, an acceleration filtering step, and an acceleration smoothing step. The information storage step is to cause a memory to store obstacle information about an obstacle, road information, and vehicle information about the autonomous vehicle. The vehicle information includes a jerk threshold and a jerk switching frequency threshold.The acceleration limit calculation step is to cause a data processing unit to receive the vehicle information from the memory and calculate the vehicle information according to a calculation procedure to generate an acceleration limit range of the autonomous vehicle. In addition, the acceleration combination generation step is to cause the data processing unit to receive the obstacle information and the road information from the memory and plan an acceleration interval of the autonomous vehicle according to the obstacle information, the road information, and the acceleration limit range, and then generate multiple acceleration combinations of the autonomous vehicle according to the acceleration interval.The acceleration filtering step causes the data processing unit to filter the acceleration combinations according to the jerk threshold and the jerk switching frequency threshold to obtain a selected acceleration combination. The acceleration smoothing step causes the data processing unit to execute a driving behavior procedure to adjust the selected acceleration combination to generate the optimal speed curve.

[0006] According to the aforementioned dynamic speed planning method for the autonomous vehicle, the aforementioned acceleration limit calculation step includes executing a lateral acceleration calculation step, a longitudinal acceleration calculation step, and a longitudinal speed and lateral speed calculation step. The lateral acceleration calculation step is to cause the data processing unit to calculate the vehicle information according to a dynamic calculation model to generate a lateral acceleration of the autonomous vehicle. The longitudinal acceleration calculation step is to cause the data processing unit to calculate the lateral acceleration according to a friction circle calculation model to generate a longitudinal acceleration of the autonomous vehicle.The step of calculating the longitudinal velocity and the lateral velocity serves to cause the data processing unit to calculate the lateral acceleration and the longitudinal acceleration according to a kinematic calculation model in order to generate a longitudinal velocity and a lateral velocity of the autonomous vehicle, respectively.

[0007] In another aspect of the present description, a dynamic speed planning system for an autonomous vehicle is implemented to plan the optimal speed curve of the autonomous vehicle. The dynamic speed planning system includes a memory and a data processing unit. The memory is configured to access obstacle information about an obstacle, road information and vehicle information about the autonomous vehicle, a calculation procedure, and a driving behavior procedure. The vehicle information includes a jerk threshold and a jerk switching frequency threshold. The data processing unit is electrically connected to the memory.The data processing unit is configured to implement a dynamic speed planning method for the autonomous vehicle, comprising executing an acceleration limit calculation step, an acceleration combination generation step, an acceleration filtering step, and an acceleration smoothing step. The acceleration limit calculation step calculates the vehicle information according to the calculation procedure to generate an acceleration limit range of the autonomous vehicle. The acceleration combination generation step plans an acceleration interval of the autonomous vehicle according to the obstacle information, the road information, and the acceleration limit range, and then generates multiple acceleration combinations of the autonomous vehicle according to the acceleration interval.The acceleration filtering step filters the acceleration combinations according to the jerk threshold and the jerk switching frequency threshold to obtain a selected acceleration combination. The acceleration smoothing step executes the driving behavior procedure to adjust the selected acceleration combination to generate the optimal speed curve.

[0008] According to the aforementioned dynamic speed planning system for the autonomous vehicle, the aforementioned memory contains a dynamic calculation model, a friction circle calculation model, and a kinematic calculation model, and the aforementioned acceleration limit calculation step includes the execution of a lateral acceleration calculation step, a longitudinal acceleration calculation step, and a longitudinal velocity and lateral velocity calculation step. The lateral acceleration calculation step is to calculate the vehicle information according to the dynamic calculation model to generate a lateral acceleration of the autonomous vehicle. The longitudinal acceleration calculation step is to calculate the lateral acceleration according to a friction circle calculation model to generate a lateral acceleration of the autonomous vehicle.The step of calculating the longitudinal velocity and the lateral velocity is used to calculate the lateral acceleration and the longitudinal acceleration according to a kinematic calculation model in order to generate a lateral velocity and a longitudinal velocity of the autonomous vehicle, respectively. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] The present description will be better understood from the detailed description of the embodiments with reference to the following attached figures: Fig. 1 shows a flowchart of a dynamic speed planning method for an autonomous vehicle according to a first embodiment of the present description. Fig. 2 shows a flowchart of a dynamic speed planning method for an autonomous vehicle according to a second embodiment of the present description. Fig. Figure 3 shows a schematic view of an information storage step of the dynamic speed planning method for the autonomous vehicle of the Fig. 2. Fig. 4 shows a schematic view of a step for calculating an acceleration limit of the dynamic speed planning method for the autonomous vehicle of the Fig. 2 Fig. Figure 5 shows a schematic view of an acceleration interval generation step of the dynamic speed planning method for the autonomous vehicle of the Fig. 2. Fig. Figure 6 shows a schematic view of the dynamic speed planning method for the autonomous vehicle of the Fig. 2, applied to avoiding an object on the same lane. Fig. Figure 7 shows a schematic view of an acceleration distribution step of the dynamic speed planning method for the autonomous vehicle of the Fig. 2. Fig. Figure 8 shows a schematic view of an acceleration filtering step of the dynamic speed planning method for the autonomous vehicle of the Fig. 2. Fig. Figure 9 shows a schematic view of a progressive model, a normal model and a conservative model of the dynamic speed planning method for the autonomous vehicle of the Fig. 2. Fig. 10 shows a schematic view of an adjustment step of the dynamic speed planning method for the autonomous vehicle of the Fig. 2. Fig. 11 shows a schematic view of an adaptation step of the dynamic speed planning method for the autonomous vehicle of the Fig. 2. Fig. 12 shows a block diagram of a dynamic speed planning system for the autonomous vehicle according to a third embodiment of the present description. DETAILED DESCRIPTION

[0010] The embodiment will be described with reference to the drawings. For clarity, some practical details are described below. However, it should be noted that the present description should not be limited by the practical details; that is, in some embodiments, the practical details are not required. Furthermore, to simplify the drawings, some conventional structures and parts are simply illustrated, and repeated parts may be denoted by the same reference numerals.

[0011] It should be noted that when a part (or device) is stated to be "connected" to another part, it may be directly connected to the other part or indirectly connected to the other part, i.e., there may be interposed parts. Conversely, when a part is stated to be "directly connected" to another part, there are no interposed parts. Furthermore, the terms first, second, third, etc., are used herein to describe various parts or components, and these parts or components are not intended to be limited by these terms. Accordingly, a first part or component discussed below could also be referred to as a second part or component.

[0012] See Fig. 1. Fig. 1 shows a flowchart of a dynamic speed planning method 100 for an autonomous vehicle according to a first embodiment of the present description. The dynamic speed planning method 100 for the autonomous vehicle is used to plan a best speed curve 110 of the autonomous vehicle. The dynamic speed planning method 100 for the autonomous vehicle includes the execution of an information storage step S02, an acceleration limit calculation step S04, an acceleration combination generation step S06, an acceleration filtering step S08, and an acceleration smoothing step S10.

[0013] The information storage step S02 serves to cause a memory to store obstacle information about an obstacle, road information, and vehicle information 102 about the autonomous vehicle. The vehicle information 102 includes a jerk threshold and a jerk switching frequency threshold. The acceleration limit calculation step S04 serves to cause a data processing unit to receive the vehicle information 102 from the memory and calculate the vehicle information 102 according to a calculation procedure to generate an acceleration limit range 104 of the autonomous vehicle.In addition, the acceleration combination generation step S06 serves to cause the data processing unit to receive the obstacle information and the road information from the memory and to schedule an acceleration interval of the autonomous vehicle according to the obstacle information, the road information, and the acceleration threshold range 104, and then generate a plurality of combinations 106 for accelerating the autonomous vehicle according to the acceleration interval. The acceleration filtering step S08 serves to cause the data processing unit to filter the acceleration combinations 106 according to the jerk threshold and the jerk switching frequency threshold to obtain a selected acceleration combination 108.The acceleration smoothing step S10 serves to cause the data processing unit to execute a driving behavior procedure to adjust the selected acceleration combination 108 to generate the best speed curve 110. Thus, the inventive dynamic speed planning method 100 for the autonomous vehicle described herein obtains the acceleration limit range 104 using the vehicle information 102 and the calculation method, and then plans the acceleration combinations 106 by integrating the acceleration limit range 104 with the obstacle information and the road information, using acceleration as a standard, and simultaneously considering the vehicle's operating limits, vehicle dynamics, and human driving behavior to adapt to changes in the environment, so that the future behavior of the autonomous vehicle is predictable.The details of the above-mentioned steps will be described below in more detailed embodiments.

[0014] See Fig. 2 to 11. Fig. 2 shows a flowchart of a dynamic speed planning method 100a for an autonomous vehicle HV according to a second embodiment of the present description. Fig. 3 shows a schematic view of an information storage step S12 of the dynamic speed planning method 100a for the autonomous vehicle HV of the Fig. 2. Fig. Figure 4 shows a schematic view of a step for calculating an acceleration limit S14 of the dynamic speed planning method 100a for the autonomous vehicle HV of the Fig. 2. Fig. Figure 5 shows a schematic view of a step for generating an acceleration interval 162 of the dynamic speed planning method for the autonomous vehicle HV of the Fig. 2. Fig. Figure 6 shows a schematic view of the dynamic speed planning method 100a for the autonomous vehicle HV of the Fig. 2, applied to avoiding an object on the same lane. Fig. Fig. 7 shows a schematic view of an acceleration distribution step S164 of the dynamic speed planning method 100a for the autonomous vehicle HV of the Fig. 2. Fig. Figure 8 shows a schematic view of an acceleration filtering step S18 of the dynamic speed planning method 100a for the autonomous vehicle HV of the Fig. 2. Fig. Figure 9 shows a schematic view of a progressive model M1, a normal model M2 and a conservative model M3 of the dynamic speed planning method 100a for the autonomous vehicle HV of the Fig. 2. Fig. 10 shows a schematic view of an adjustment step S202 of the dynamic speed planning method 100a for the autonomous vehicle HV of the Fig. 2. Fig. 11 shows a schematic view of an adaptation step S204 of the dynamic speed planning method 100a for the autonomous vehicle HV of the Fig. 2. In the Fig. 2 to 11, the dynamic speed planning method 100a for the autonomous vehicle HV is executed to plan a best speed curve 110 of the autonomous vehicle HV. The dynamic speed planning method 100a for the autonomous vehicle HV includes the execution of an information storage step S12, an acceleration limit calculation step S14, an acceleration combination generation step S16, an acceleration filtering step S18, an acceleration smoothing step S20, and a control step S22.

[0015] The information storage step S12 serves to cause a memory to store obstacle information about an obstacle Obj, road information, and vehicle information 102 about the autonomous vehicle HV. Specifically, the autonomous vehicle HV includes a sensor module configured to detect the obstacle information, the road information, and the vehicle information 102, and stores the obstacle information, the road information, and the vehicle information 102 in the memory. The obstacle information includes an obstacle speed V Obj , an obstacle acceleration a Obj and an obstacle acceleration range 103 of the obstacle Obj. The road information includes a maximum speed limit V max and a minimum speed limit V minThe vehicle information 102 includes a jerk threshold, a jerk switching frequency threshold, a front wheel cornering stiffness, a rear wheel cornering stiffness, a front wheelbase, a rear wheelbase, a vehicle inertia, and a vehicle mass.

[0016] The step of calculating an acceleration limit S14 serves to cause a data processing unit to receive the vehicle information 102 from the memory and to calculate the vehicle information 102 according to a calculation procedure in order to generate an acceleration limit range 104 of the autonomous vehicle HV. In detail, the step of calculating an acceleration limit S14 comprises the execution of a step of calculating a lateral acceleration S142, a step of calculating a longitudinal acceleration S144, and a step of calculating a longitudinal speed and a lateral speed S146. The step of calculating a lateral acceleration S142 serves to cause the data processing unit to calculate the vehicle information 102 according to a dynamic calculation model in order to generate a lateral acceleration a yof the autonomous vehicle HV. The step of calculating a longitudinal acceleration S144 serves to cause the data processing unit to calculate the lateral acceleration a y according to a friction circle calculation model in order to calculate a longitudinal acceleration a x of the autonomous vehicle HV. The step of calculating the longitudinal and lateral speeds S146 serves to cause the data processing unit to calculate the lateral acceleration a y and the longitudinal acceleration a x to be calculated according to a kinematic calculation model in order to generate a lateral velocity v or a longitudinal velocity u of the autonomous vehicle HV.

[0017] In more detail, the calculation method includes the dynamic calculation model, the friction circle calculation model, and the kinematic calculation model. The dynamic calculation model first contains a transverse force F y, the vehicle mass m, an acceleration v̇, the longitudinal velocity u, a yaw rate r, a yaw angular acceleration ṙ, a front wheel lateral force F yf , a rear wheel lateral force F yr and the vehicle inertia I z and satisfies the following equation (1): {Fy=m(v˙+ur)=Fyf+FyrIz⋅r˙=Fyf+Fyr

[0018] The data processing unit substitutes the front wheel cornering stiffness C in equation (1) αf , the rear wheel cornering stiffness C αr , the front wheelbase a, the rear wheelbase b, the vehicle inertia I z and the vehicle mass m from the vehicle information 102 and derives the following equation (2) according to the dynamic calculation model: ddt[vr]=[−(Cαf+Cαr)mubCαr−aCαrmubCαr−aCαrIzu−(a2Cαf+b2Cαr)Izu][vr]+[CαfmaCαfIz]δf v is the lateral velocity, δ fa front wheel angle and t a time. The data processing unit performs a matrix multiplication and an expansion to equation (2), and after rearrangement, the lateral acceleration a y which satisfies the following equation (3): dvdt=(−(Cαf+Cαr)mu)v+(bCαr−aCαrmu−u0)r / (Cαfm)δf =−Cαfm(δf−v−aru)+Cαrm(−v−bru)−ur=ay

[0019] The friction circle calculation model then contains a maximum available longitudinal force F x, max , a longitudinal force F x , a maximum available shear force F y, max , a transverse force F y , a maximum longitudinal acceleration a x, max , the longitudinal acceleration a x , and the lateral acceleration a y , and satisfies the following equation (4). The data processing unit shifts and eliminates equation (4) to generate the longitudinal acceleration a x which satisfies the following equation (5): (FxFx,max)2+(FyFy,max)2=(maxmax,max)2+(maymay,max)2=1 ax=ax,max×1−(ayay,max)2

[0020] Finally, the kinematic calculation model includes a velocity V, an initial velocity V0, the acceleration V̇, and the time t, and satisfies the following equation (6). The data processing unit generates the longitudinal velocity u and the lateral velocity v, which satisfy the following equation (7), according to the kinematic calculation model: V=V0+V˙t {u=u02+2axSv=v02+2ayS

[0021] S is a distance, u0 is an initial longitudinal velocity, and v0 is an initial lateral velocity. The dynamic velocity planning method 100a for the autonomous vehicle HV of the present description thus generates the lateral acceleration a y with the vehicle information 102 and the dynamic calculation model and then generates the longitudinal acceleration a xwith the friction circle calculation model and finally generates the longitudinal velocity u and the lateral velocity v with the kinematic calculation model. It should be noted that the two accelerations a HV the vehicle information 102 in a longitudinal direction X and a transverse direction Y on a future path the above-mentioned longitudinal acceleration a x or the lateral acceleration a y Two speeds V HV of the vehicle information 102 in the longitudinal direction X and the transverse direction Y are the above-mentioned longitudinal speed u and transverse speed v, respectively. The range of accelerations a HV the vehicle information 102 is the acceleration limit range 104 of the autonomous vehicle HV.

[0022] The acceleration combination generation step S16 includes the execution of an acceleration interval generation step S162 and an acceleration distribution step S164. The acceleration interval generation step S162 includes the execution of an obstacle restriction step S1622 and a road restriction step S1624. The obstacle restriction step S1622 is executed to restrict the acceleration limit range 104 of the autonomous vehicle HV according to the obstacle information to generate an initial acceleration interval 104a. The road restriction step S1624 is executed to restrict the initial acceleration interval 104a according to the road information to generate an acceleration interval 105. In detail, the data processing unit restricts the acceleration limit range 104 based on the obstacle acceleration range 103 (iethe range of obstacle acceleration a. Obj ) of the obstacle information to generate the initial acceleration interval 104a. Then, the data processing unit extracts the acceleration limit range 104 based on the maximum speed limit V max and the minimum speed limit V min to generate the acceleration interval 105. Thus, the dynamic speed planning method 100a for the autonomous vehicle HV of the present description uses the road information and the obstacle information in an ordinary roadway to further restrict the acceleration limit range 104 of the autonomous vehicle HV to calculate the applicable acceleration limit range for the autonomous vehicle HV (ie, the acceleration interval 105).

[0023] Furthermore, the acceleration distribution step S164 comprises the execution of a distribution step S1642 and a target point combination step S1644. The distribution step S1642 serves to generate a plurality of acceleration groups G1, G2, G3, G4 according to a specified time interval and the acceleration interval 105 and to distribute each of the acceleration groups G1, G2, G3, G4 according to a specified acceleration interval to create at least one acceleration target point a T The target point combination step S1644 serves to successively generate the at least one acceleration target point a T each of the acceleration groups G1, G2, G3, G4 to generate several acceleration combinations 106. For example, if the specified acceleration interval is 1 m / s 2and the specified time interval is 0.1 s as standard, the acceleration interval 105 of a first waypoint of the autonomous vehicle is HV [1, 1] m / s 2 and has only one acceleration value (ie the acceleration target point a T ). Then the next waypoint of an acceleration value is calculated according to the first waypoint and the next acceleration interval 105 is [-3, 5] m / s 2 , which is available in 9 types of acceleration values ​​-3, -2, -1, 0, 1, 2, 3, 4 and 5 m / s 2 Similarly, another waypoint of an acceleration value is calculated according to the current waypoint and the other acceleration interval 105 is [-4, 8] m / s 2 etc. (as in Fig. 7) and is not described in detail, but is not limited to the present description.

[0024] The acceleration filtering step S18 serves to cause the data processing unit to filter the acceleration combinations 106 according to the jerk threshold and the jerk switching frequency threshold to obtain a selected acceleration combination 108. In particular, each of the aforementioned acceleration combinations 106 includes a maximum jerk. The maximum jerk of each of the acceleration combinations 106 is less than or equal to the jerk threshold. The maximum jerk of each of the acceleration combinations 106 is denoted by J max The jerk threshold is denoted by J threshold and satisfies the following equation (8): Jmax≤Jthreshold

[0025] In detail, the acceleration combinations 106 comprise an acceleration combination 1061 and an acceleration combination 1062. The jerk threshold J threshold can reach 20 m / s2 In the acceleration combination 1061, the first jerk of approximately 10 m / s 2 of 1 m / s 2 up to 2 m / s 2 generated and the last jerk (ie the maximum jerk J max the acceleration combination 1061) of approximately 20 m / s 2 of 3 m / s 2 up to 1 m / s 2 In the acceleration combination 1062, the first jerk (ie the maximum jerk J max the acceleration combination 1062) of approximately 40 m / s 2 of 1 m / s 2 down to -3 m / s 2 and the last jerk of about 20 m / s 2 of 1 m / s 2 down to -1 m / s 2 Therefore, the data processing unit separates the above-mentioned acceleration combination 1062 based on the jerk threshold J threshold out of.

[0026] Furthermore, the vehicle information 102 also includes the jerk switching frequency threshold stored in the memory. Each of the above-mentioned acceleration combinations 106 also includes a jerk switching frequency. In particular, the jerk switching frequency of each of the acceleration combinations 106 is less than or equal to the jerk switching frequency threshold and is denoted by Jfrequency, and the jerk switching frequency threshold is denoted by f threshold and satisfies the following equation (9): ∑Jfrequency≤fthreshold

[0027] In detail, the jerk switching frequency threshold f threshold 2. When the jerk is switched between positive and negative values, the jerk switching frequency threshold Jfrequency is accumulated once per time. In the acceleration combination 1061, the acceleration target point a T from the initial value 1 m / s 2 to 2 m / s2 . Then the acceleration target point changes to a T from the initial value 1 m / s 2 to 2 m / s 2 . Then the acceleration target point changes to a T of 2 m / s 2 to 3 m / s 2 and then from 3 m / s 2 to 1 m / s 2 . The jerk switching frequency threshold Jfrequency of the acceleration combination 1061 is 0. In the acceleration combination 1062, the acceleration target point a T from the initial value 1 m / s 2 to -3 m / s 2 . Then the acceleration target point changes to a T from the initial value -3 m / s 2 to 1 m / s 2 and then from 1 m / s 2 to -1 m / s 2 . The jerk switching frequency threshold Jfrequency of the acceleration combination 1062 is 3. The data processing unit therefore separates the above-mentioned acceleration combination 1062 based on the jerk switching frequency threshold f thresholdThe dynamic speed planning method 100a for the autonomous vehicle HV of the present description thus filters the acceleration combinations 106 of the vehicle dynamics according to the jerk threshold J threshold and the jerk switching frequency threshold f threshold to obtain the selected acceleration combination 108 (ie, acceleration combination 1061).

[0028] The acceleration smoothing step S20 serves to cause the data processing unit to execute the driving behavior procedure to adjust the selected acceleration combination 108 to generate the best speed curve 110. The driving behavior procedure is classified into the progressive model M1, the normal model M2, and the conservative model M3 according to an acceleration a, a speed V, and a steering wheel angle θ. Furthermore, the acceleration smoothing step S20 includes an adjustment step S202 and an adaptation step S204. The adjustment step S202 serves to adjust the selected acceleration combination 108 according to one of the aforementioned models: the progressive model M1, the normal model M2, or the conservative model M3, to generate an artificial acceleration combination 108a. The artificial acceleration combination 108a includes several optimal accelerations a f1 , a f2, a f3 , a f4 . The adjustment step S204 is used to integrate and smooth each of the optimal accelerations a f1 , a f2 , a f3 , a f4 the artificial acceleration combination 108a to adapt it to the best speed curve 110.

[0029] In detail, each of the above-mentioned models, the progressive model M1, the normal model M2 and the conservative model M3, contains a trend curve C. The data processing unit adjusts the acceleration target points a T1 , a T2 , a T3 , a T4 the artificial acceleration combination 108 to the optimal accelerations a according to the trend curve C f1 , a f2 , a f3 , a f4one of the models, the progressive model M1, the normal model M2 or the conservative model M3, each of which is similar to the selected acceleration combination 108. For example, the acceleration target point a T2 (-3 m / s 2 ) to the best acceleration a f2 (-2 m / s 2 ). Finally, the artificial acceleration combination 108a is converted into a speed combination 108b using an integration process. The speed combination 108b is curve-fitted to smooth the speed combination 108b and generate the best speed curve 110. The control step S22 serves to control the autonomous vehicle HV using an automatic driving parameter based on the best speed curve 110. The details of the control step S22 are conventional and will not be described again here.

[0030] The dynamic speed planning method 100a for the autonomous vehicle HV of the present description thus changes an acceleration change rate through the driving behavior procedure to obtain the artificial acceleration combination 108a, and then adjusts the speed combination 108b to a smooth curve to eliminate the control shock caused by the problems of discontinuous speed. Furthermore, the collision time between the autonomous vehicle HV and the obstacle Obj can be estimated according to the best speed curve 110, so that the interaction relationship between the autonomous vehicle HV and the obstacle Obj can be predicted.

[0031] See the Fig. 2 to 12. Fig.Figure 12 shows a block diagram of a dynamic speed planning system 200 for the autonomous vehicle HV according to a third embodiment of the present description. The dynamic speed planning system 200 for the autonomous vehicle HV is configured to plan the optimal speed curve 110 of the autonomous vehicle HV and includes a sensor module 300, a memory 400, and a data processing unit 500.

[0032] The sensor module 300 is configured to detect the obstacle information, the road information, and the vehicle information 102 and to store the obstacle information, the road information, and the vehicle information 102 in the memory 400. The obstacle information includes the obstacle speed V Obj , the obstacle acceleration a Obj . The road information includes the maximum speed limit V max and the minimum speed limit V minThe vehicle information 102 includes the jerk threshold, the jerk switching frequency threshold, the front wheel cornering stiffness, the rear wheel cornering stiffness, the front wheelbase, the rear wheelbase, the vehicle inertia, and the vehicle mass of the autonomous vehicle HV. The above-mentioned sensor module 300 may include a GPS, a gyroscope, an odometer, a speedometer, an inertial measurement unit (IMU), a LiDAR, a radar, and a camera. The sensor module 300 is conventional and will not be described again herein.

[0033] The memory 400 is configured to access obstacle information about the obstacle Obj, the road information 102, and the vehicle information about the autonomous vehicle HV, the calculation method, and the driving behavior procedure. The driving behavior procedure is classified according to the acceleration a, the speed V, and the steering wheel angle θ into the progressive model M1, the normal model M2, and the conservative model M3. The vehicle information 102 includes the jerk threshold J threshold and the jerk switching frequency threshold f threshold of the autonomous vehicle HV.

[0034] The data processing unit 500 is electrically connected to the memory 400 and the sensor module 300. The data processing unit 500 is configured to implement the dynamic speed planning method 100, 100a for the autonomous vehicle HV. The data processing unit 500 may be a microprocessor, an electronic control unit (ECU), a computer, a mobile device, or other computing device.

[0035] Thus, the dynamic speed planning system 200 for the autonomous vehicle HV of the present description uses the obstacle information, the road information, and the vehicle information 102 to plan the acceleration combinations 106 that are possible for the autonomous vehicle HV, and filters the acceleration combinations 106 into the artificial acceleration combination 108a that are suitable for driving behavior that conforms to the jerk limits in order to adapt to environmental changes in the future and improve the comfort of passengers.

[0036] In summary, the present description has the following advantages: First, the acceleration threshold range is obtained from the vehicle information and the calculation method, and the acceleration combination is planned by integrating the obstacle information with the road information. Acceleration is used as a standard, and the present description simultaneously considers the vehicle's operating limits, vehicle dynamics, and human driving behavior to adapt to environmental changes. Second, the possible acceleration combinations are filtered according to the jerk threshold and the jerk switching frequency threshold to reduce the acceleration change rate and improve passenger comfort.Third, speed planning that can take environmental changes into account can handle situations that cannot be handled by conventional commercial vehicle systems, such as obstacle encounters, lane changes, etc. The present description makes a planning system for an autonomous vehicle more robust and safer against environmental changes.

Claims

[1] Dynamic speed planning method (100) for an autonomous vehicle, executed to plan the best speed curve (110) of the autonomous vehicle, the dynamic speed planning method (100) for the autonomous vehicle comprising: performing an information storage step (S02) to cause a memory to store obstacle information (102) about an obstacle, road information (102) and vehicle information about the autonomous vehicle, wherein the vehicle information includes a jerk threshold and a jerk switching frequency threshold, performing a step of calculating an acceleration limit (S04) to cause a data processing unit to receive the vehicle information (102) from the memory and to calculate the vehicle information according to a calculation procedure (104) to generate an acceleration limit range of the autonomous vehicle, performing an acceleration combination generation step (S06) to cause the data processing unit to receive the obstacle information and the road information from the memory and to plan an acceleration interval of the autonomous vehicle according to the obstacle information, the road information and the acceleration limit range (104), and then to generate several acceleration combinations (106) of the autonomous vehicle according to the acceleration interval, performing an acceleration filtering step (S08) to cause the data processing unit to filter the acceleration combinations (106) according to the jerk threshold and the jerk switching frequency threshold to obtain a selected acceleration combination (108), and performing an acceleration smoothing step (S10) to cause the data processing unit to execute a driving behavior procedure to adjust the selected acceleration combination (108) to generate the best speed curve (110). [2] Dynamic speed planning method (100a) for the autonomous vehicle (HV) according to claim 1, in which the vehicle information (102) further comprises a front wheel cornering stiffness, a rear wheel cornering stiffness, a front wheelbase, a rear wheelbase, a vehicle inertia and a vehicle mass. [3] Dynamic speed planning method (100a) for the autonomous vehicle (HV) according to claim 1, in which the step of calculating an acceleration limit (S14) comprises: performing a step (S142) for calculating a lateral acceleration to cause the data processing unit to calculate the vehicle information (102) according to a dynamic calculation model in order to calculate a lateral acceleration (a y ) of the autonomous vehicle (HV); performing a step for calculating a longitudinal acceleration (S144) to inform the data processing unit (a y ) to calculate the vehicle information according to a friction circle calculation model in order to determine a longitudinal acceleration (a x ) of the autonomous vehicle (HV), and performing a step for calculating a longitudinal velocity (S146) and a lateral velocity to cause the data processing unit to calculate the lateral acceleration and the longitudinal acceleration according to a kinematic calculation model in order to generate a longitudinal velocity (u) and a lateral velocity of the autonomous vehicle (HV), respectively. [4] Dynamic speed planning method (100a) for the autonomous vehicle (HV) according to claim 1, in which the step of generating acceleration combinations (S16) comprises: the execution of a step for generating an acceleration interval (S162), wherein the step for generating an acceleration interval (S162) is implemented by the data processing unit and comprises: executing an obstacle restriction step (S1622) for restricting the acceleration limit range (104) of the autonomous vehicle (HV) according to the obstacle information to generate an initial acceleration interval (104a), and executing a road restriction step (S1624) for restricting the initial acceleration interval (104a) according to the road information to generate the acceleration interval (105). [5] Dynamic speed planning method (100a) for the autonomous vehicle (HV) according to claim 4, in which the step of generating acceleration combinations (S16) further comprises: the execution of an acceleration distribution step (S164), wherein the acceleration distribution step (S164) is implemented by the data processing unit and comprises: performing a distribution step (S1624) to generate a plurality of acceleration groups (G1, G2, G3, G4) according to a specified time interval and the acceleration interval (105) and to distribute each of the acceleration groups (G1, G2, G3, G4) according to a specified acceleration interval to determine at least one acceleration target point (a T ) and executing a target point combining step (S1644) to successively select the at least one acceleration target point (a T ) of each of the acceleration groups (G1, G2, G3, G4) to produce the acceleration combinations (106, 1061, 1062). [6] Dynamic speed planning method (100a) for the autonomous vehicle (HV) according to claim 1, in which each of the acceleration combinations (106, 1061, 1062) comprises a maximum jerk and a jerk switching frequency, the jerk threshold being J thresholdis called, the maximum jerk of each of the acceleration combinations (106, 1061, 1062) with J max is denoted, the jerk switching frequency threshold is denoted by Jfrequency, and the jerk switching frequency threshold is denoted by f threshold and satisfies the following equation: Jmax≤Jthreshold and ∑Jfrequency≤fthreshold. [7] Dynamic speed planning method (100a) for the autonomous vehicle (HV) according to claim 1, in which the acceleration smoothing step (S20) comprises: performing an adjustment step (S202) to adjust the selected acceleration combination (108) according to one of a progressive model (M1), a normal model (M2) and a conservative model (M3) to generate an artificial acceleration combination (108a), wherein the artificial acceleration combination (108a) comprises a plurality of optimal accelerations (a f1 , a f2 , a f3, a f4 ), and the execution of an adaptation step (S204) to determine the optimal accelerations (a f1 , a f2 , a f3 , a f4 ) of the artificial acceleration combination (108a) and smooth it to adapt it to the best speed curve (110). [8] Dynamic speed planning system (200) for an autonomous vehicle (HV), designed to plan the best speed curve (110) of the autonomous vehicle (HV), the dynamic speed planning system comprising: a memory (400) configured to access obstacle information about an obstacle (Obj), road information, and vehicle information (102) about the autonomous vehicle (HV), a calculation procedure, and a driving behavior procedure, wherein the vehicle information (102) comprises a jerk threshold and a jerk switching frequency threshold, and a data processing unit (500) electrically connected to the memory (400), the data processing unit (500) being configured to implement a dynamic speed planning method (100a) for an autonomous vehicle (HV), comprising: performing an acceleration limit calculation step (S14) to calculate the vehicle information (102)' according to the calculation procedure to generate an acceleration limit range (104) of the autonomous vehicle (HV), executing an acceleration combination generation step (S16) to plan an acceleration interval (105) of the autonomous vehicle (HV) according to the obstacle information, the road information and the acceleration limit range (104) and then generating a plurality of acceleration combinations (106, 1061, 1062) of the autonomous vehicle (HV) according to the acceleration interval, performing an acceleration filtering step (S18) to filter the acceleration combinations (106, 1061, 1062) according to the jerk threshold and the jerk switching frequency threshold to obtain a selected acceleration combination (108), and performing an acceleration smoothing step (S20) to execute the driving behavior procedure to adjust the selected acceleration combination (108) to generate the best speed curve (110). [9] Dynamic speed planning system (200) for the autonomous vehicle (HV) according to claim 8, in which the vehicle information (102) further comprises a front wheel cornering stiffness, a rear wheel cornering stiffness, a front wheelbase, a rear wheelbase, a vehicle inertia and a vehicle mass. [10] Dynamic speed planning system (200) for the autonomous vehicle (HV) according to claim 8, in which the memory (400) contains a dynamic calculation model, a friction circle calculation model and a kinematics calculation model, and the step of calculating the acceleration limit (S14) comprises: performing a lateral acceleration calculation step (S142) to calculate the vehicle information (102) according to the dynamic calculation model to calculate a lateral acceleration (a y ) of the autonomous vehicle (HV), the execution of a step for calculating a longitudinal acceleration (S144) in order to calculate the vehicle information according to the friction circle calculation model (a y ) to achieve a longitudinal acceleration (a x ) of the autonomous vehicle (HV), and the execution of a step for calculating a longitudinal velocity (S146) and a lateral velocity in order to determine the lateral acceleration (ay ) and the longitudinal acceleration (a x ) according to the kinematic calculation model in order to generate a lateral velocity (v) or a longitudinal velocity (u) of the autonomous vehicle (HV). [11] Dynamic speed planning system (200) for the autonomous vehicle (HV) according to claim 8, in which the data processing unit (500) is arranged to implement a step of generating an acceleration interval (S162), the step of generating an acceleration interval (S162) comprising: performing an obstacle restriction step (1622) for restricting the acceleration limit range (104) of the autonomous vehicle (HV) according to the obstacle information to generate an initial acceleration interval (104a), and executing a road restriction step (S1624) for restricting the initial acceleration interval (104a) according to the road information to generate an acceleration interval (105). [12] Dynamic speed planning system (200) for the autonomous vehicle (HV) according to claim 11, in which the data processing unit (500) is arranged to implement an acceleration distribution step (S164), the acceleration distribution step (S164) comprising: Carrying out a distribution step (S1642) to generate a plurality of acceleration groups (G1, G2, G3, G4) according to a specified time interval and the acceleration interval (105) and to distribute each of the acceleration groups (G1, G2, G3, G4) according to a specified acceleration interval to determine at least one acceleration target point (a T ) and performing a target point combination step (1644) to successively select the at least one acceleration target point (a T ) of each of the acceleration groups (G1, G2, G3, G4) to produce the acceleration combinations (106, 1061, 1062). [13] Dynamic speed planning system (200) for the autonomous vehicle (HV) according to claim 8, in which each of the acceleration combinations (106, 1061, 1062) comprises a maximum jerk and a jerk switching frequency, the jerk threshold being J threshold is called, the maximum jerk of each of the acceleration combinations (106, 1061, 1062) with J max is denoted, the jerk switching frequency of each of the acceleration combinations (106, 1061, 1062) is denoted by Jfrequency, and the jerk switching frequency threshold is denoted by f threshold and satisfies the following equation: Jmax≤Jthreshold; and ∑Jfrequency≤fthreshold. [14] Dynamic speed planning system (200) for the autonomous vehicle (HV) according to claim 8, in which the data processing unit (500) is arranged to implement the acceleration smoothing step (S20), the acceleration smoothing step (S20) comprising: performing an adjustment step (S202) to adjust the selected acceleration combination (108) according to one of a progressive model (M1), a normal model (M2) and a conservative model (M3) to generate an artificial acceleration combination (108a), wherein the artificial acceleration combination (108a) comprises a plurality of optimal accelerations (a f1 , a f2 , a f3 , a f4 ), and the execution of an adaptation step (S204) to determine the optimal accelerations (a f1 , a f2 , a f3 , a f4) of the artificial acceleration combination (108a) and smooth it to adapt it to the best speed curve (110).

Images