Longitudinal control method, apparatus, electronic device, and storage medium

Abstract

Embodiments of the present disclosure disclose a longitudinal control method and device, electronic equipment and storage medium, the method comprising: determining a reference vehicle speed sequence according to a planning speed of a reference trajectory point, and determining a reference acceleration sequence according to a planning acceleration of the reference trajectory point; determining a desired acceleration increment sequence by minimizing a pre-established target function under the constraint of a preset constraint condition, according to a desired acceleration increment correlation, a current vehicle speed, a current vehicle acceleration, a desired acceleration of the vehicle at a previous time of a current time, a current vehicle speed error integral, a current vehicle acceleration error integral, the reference vehicle speed sequence and the reference acceleration sequence; and determining a sum value of a first term in the desired acceleration increment sequence and the desired acceleration of the vehicle at the previous time of the current time as a desired acceleration at the current time, so as to achieve safer, more accurate and smoother longitudinal control.

Description

Technical Field

[0001] This disclosure relates to the field of autonomous driving technology, and in particular to a longitudinal control method, apparatus, electronic device, and storage medium. Background Technology

[0002] The primary task of longitudinal control is to track the longitudinal trajectory provided by the planning module and issue corresponding acceleration / deceleration commands to the drive-by-wire chassis. A common control architecture is a hierarchical architecture: the upper-level controller calculates the desired acceleration based on the current vehicle state and the reference longitudinal trajectory, while the lower-level controller calculates the torque (opening) / brake pressure (braking force) based on the desired acceleration. If the drive-by-wire chassis supplier has provided the corresponding longitudinal control interface, then acceleration tracking control can be performed by the autonomous driving system, which simply issues the torque (opening) / brake pressure (braking force). If the relevant longitudinal control interface is not provided, the autonomous driving system only needs to issue the desired acceleration.

[0003] Therefore, regardless of the type of longitudinal control interface, the desired acceleration must be accurately determined for tracking. However, because the vehicle's physical response takes time and there may be issues such as acceleration overshoot or insufficient acceleration response, precise and smooth longitudinal control of the vehicle may not be possible. Summary of the Invention

[0004] To address, or at least partially address, the aforementioned technical problems, embodiments of this disclosure provide a longitudinal control method, apparatus, electronic device, and storage medium to achieve safer, more precise, and smoother longitudinal control.

[0005] In a first aspect, embodiments of this disclosure provide a longitudinal control method, the method comprising:

[0006] Based on the planned speed of the reference trajectory points, a reference vehicle speed sequence is determined, and based on the planned acceleration of the reference trajectory points, a reference acceleration sequence is determined; wherein, the reference trajectory points are the planned trajectory points for the subsequent tracking of the vehicle, the planned speed of the reference trajectory points is the planned speed of the vehicle when it reaches the reference trajectory points, and the planned acceleration of the reference trajectory points is the planned acceleration of the vehicle when it reaches the reference trajectory points.

[0007] Based on the expected acceleration increment correlation, the vehicle's current speed, vehicle's current acceleration, the vehicle's expected acceleration at the previous moment, the vehicle's current speed error integral, the vehicle's current acceleration error integral, the reference speed sequence, and the reference acceleration sequence, the pre-established objective function is minimized under preset constraints to determine the expected acceleration increment sequence; wherein, the expected acceleration increment correlation is the correlation between the actual speed recursive sequence, the actual acceleration recursive sequence, the expected acceleration recursive sequence, the speed error integral sequence, the acceleration error integral sequence, the reference speed sequence, the reference acceleration sequence, and the expected acceleration increment sequence; the actual speed recursive sequence is the correlation between the vehicle's current speed, current acceleration, current acceleration, the vehicle's current acceleration at the previous moment, the vehicle's current speed error integral, the vehicle's current acceleration error integral, the reference speed sequence, the reference acceleration sequence, and the expected acceleration increment sequence; the actual speed recursive sequence is the correlation between the vehicle's current speed, current acceleration, current acceleration at the previous moment, the vehicle's current speed error integral, the vehicle's current acceleration error integral, the reference speed sequence, the reference acceleration sequence, and the expected acceleration increment sequence; The objective function is a sequence composed of the actual vehicle speed recursive values ​​at each moment, the actual acceleration recursive sequence being a sequence composed of the actual acceleration recursive values ​​of the vehicle at each moment, and the desired acceleration recursive sequence being a sequence composed of the desired acceleration recursive values ​​of the vehicle at each moment from the previous moment; the objective function is a function constructed based on the actual vehicle speed recursive sequence, the reference vehicle speed sequence, the actual acceleration recursive sequence, the reference acceleration sequence, the desired acceleration recursive sequence, the vehicle speed error integral sequence, the acceleration error integral sequence, and the desired acceleration increment sequence; the preset constraints include desired acceleration constraints, desired acceleration constraints, and actual acceleration constraints.

[0008] The sum of the first term in the expected acceleration increment sequence and the expected acceleration of the vehicle at the previous time is determined as the expected acceleration at the current time.

[0009] Secondly, embodiments of this disclosure also provide a longitudinal control device, comprising:

[0010] The reference value sequence determination module is used to determine a reference vehicle speed sequence based on the planned speed of the reference trajectory points, and to determine a reference acceleration sequence based on the planned acceleration of the reference trajectory points; wherein, the reference trajectory points are the trajectory points for the planned subsequent tracking of the vehicle, the planned speed of the reference trajectory points is the planned speed of the vehicle when it reaches the reference trajectory points, and the planned acceleration of the reference trajectory points is the planned acceleration of the vehicle when it reaches the reference trajectory points;

[0011] The objective function solving module is used to minimize a pre-established objective function under preset constraints, based on the expected acceleration increment correlation, the vehicle's current speed, the vehicle's current acceleration, the vehicle's expected acceleration at the previous time step, the vehicle's current speed error integral, the vehicle's current acceleration error integral, the reference speed sequence, and the reference acceleration sequence, to determine the expected acceleration increment sequence. The expected acceleration increment correlation is the correlation between the actual speed recursive sequence, the actual acceleration recursive sequence, the expected acceleration recursive sequence, the speed error integral sequence, the acceleration error integral sequence, the reference speed sequence, the reference acceleration sequence, and the expected acceleration increment sequence. The objective function is a sequence composed of the actual vehicle speed recursive values ​​at each moment, the actual acceleration recursive sequence is a sequence composed of the actual acceleration recursive values ​​at each moment, and the desired acceleration recursive sequence is a sequence composed of the desired acceleration recursive values ​​of the vehicle at each moment from the previous moment. The objective function is a function constructed based on the actual vehicle speed recursive sequence, the reference vehicle speed sequence, the actual acceleration recursive sequence, the reference acceleration sequence, the desired acceleration recursive sequence, the vehicle speed error integral sequence, the acceleration error integral sequence, and the desired acceleration increment sequence. The preset constraints include desired acceleration constraints, desired acceleration constraints, and actual acceleration constraints.

[0012] The desired acceleration determination module is used to determine the desired acceleration at the current moment as the sum of the first item in the desired acceleration increment sequence and the desired acceleration of the vehicle at the previous moment.

[0013] Thirdly, embodiments of this disclosure also provide an electronic device, the electronic device comprising: one or more processors; a storage device for storing one or more programs; and when the one or more programs are executed by the one or more processors, causing the one or more processors to implement the longitudinal control method as described above.

[0014] Fourthly, embodiments of this disclosure also provide a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the longitudinal control method as described above.

[0015] This disclosure provides a longitudinal control method that determines a reference vehicle speed sequence based on the planned speed of reference trajectory points and a reference acceleration sequence based on the planned acceleration of reference trajectory points. This expands the tracking time points and improves subsequent tracking performance. Furthermore, based on the expected acceleration increment correlation, the vehicle's current speed, the vehicle's current acceleration, the vehicle's expected acceleration at the previous moment, the vehicle's current speed error integral, the vehicle's current acceleration error integral, the reference vehicle speed sequence, and the reference acceleration sequence, a pre-established objective function is minimized under preset constraints to determine the expected acceleration increment sequence. The sum of the first term in the expected acceleration increment sequence and the vehicle's expected acceleration at the previous moment is determined as the expected acceleration at the current moment. This achieves safer, more accurate, and smoother longitudinal control. Attached Figure Description

[0016] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.

[0017] Figure 1 This is a flowchart of a longitudinal control method according to an embodiment of the present disclosure;

[0018] Figure 2 This is a schematic diagram of a timeline according to an embodiment of the present disclosure;

[0019] Figure 3 This is a schematic diagram of the structure of a longitudinal control device according to an embodiment of the present disclosure;

[0020] Figure 4 This is a schematic diagram of the structure of an electronic device according to an embodiment of this disclosure. Detailed Implementation

[0021] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0022] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0023] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0024] The primary task of longitudinal control is to track the longitudinal trajectory provided by the planning module and issue corresponding acceleration / deceleration commands to the drive-by-wire chassis. A common control architecture is a hierarchical architecture: the upper-level controller calculates the desired acceleration based on the current vehicle state and the reference longitudinal trajectory, while the lower-level controller calculates the torque (opening) / brake pressure (braking force) based on the desired acceleration. If the drive-by-wire chassis supplier has opened the corresponding longitudinal control interface, then acceleration tracking control is implemented by the autonomous driving system, which needs to issue the torque (opening) / brake pressure (braking force). If the relevant longitudinal control interface is not opened, the autonomous driving system only needs to issue the desired acceleration. Regardless of the longitudinal control interface, the acceleration tracking stage has drawbacks. The vehicle's physical response requires a certain amount of time and may experience acceleration overshoot or insufficient acceleration response. However, existing longitudinal control methods rarely consider the motion response characteristics of autonomous vehicles.

[0025] To address the aforementioned issues, this disclosure provides a longitudinal control method that takes into account the vehicle's own motion response characteristics during the longitudinal motion control process, thereby achieving safer, more precise, and smoother longitudinal control.

[0026] Figure 1 This is a flowchart illustrating a longitudinal control method according to an embodiment of this disclosure. The method can be executed by a longitudinal control device, which can be implemented in software and / or hardware, and can be configured in an electronic device. Figure 1 As shown, the method may specifically include the following steps:

[0027] S110. Based on the planned velocity of the reference trajectory points, determine the reference vehicle speed sequence, and based on the planned acceleration of the reference trajectory points, determine the reference acceleration sequence.

[0028] Here, the reference trajectory point is the trajectory point along which the planned vehicle will subsequently travel; the planned velocity of the reference trajectory point is the speed of the planned vehicle when it reaches the reference trajectory point; and the planned acceleration of the reference trajectory point is the acceleration of the planned vehicle when it reaches the reference trajectory point. The reference speed sequence is a speed sequence obtained by processing and expanding the planned velocity. The reference acceleration sequence is an acceleration sequence obtained by processing and expanding the planned acceleration.

[0029] Specifically, the planned speed and planned acceleration of the vehicle as it travels to the reference trajectory point are obtained. The planned speed is processed and expanded to obtain the reference speed sequence, and the planned acceleration is processed and expanded to obtain the reference acceleration sequence.

[0030] For example, if the number of reference trajectory points is 1, then each item in the reference vehicle speed sequence can be the planned speed of the reference trajectory point, and each item in the reference acceleration sequence can be the planned acceleration of the reference trajectory point. If the number of reference trajectory points is 2 or more, then each planned speed can be processed by interpolation or other methods to obtain the reference vehicle speed sequence, and each planned acceleration can be processed by interpolation or other methods to obtain the reference acceleration sequence.

[0031] Based on the above example, the reference vehicle speed sequence can be determined from the planned velocity of the reference trajectory points, and the reference acceleration sequence can be determined from the planned acceleration of the reference trajectory points in the following ways:

[0032] Based on the current time, the aiming time, the control period, and the maximum and minimum timestamps among the planned timestamps corresponding to the reference trajectory points, determine each predicted timestamp;

[0033] For each prediction timestamp, determine the adjacent planning timestamps corresponding to the prediction timestamp, as well as the adjacent planning velocity and adjacent planning acceleration corresponding to each adjacent planning timestamp;

[0034] Based on the predicted timestamp, adjacent planning timestamps, and adjacent planning speeds, determine the reference vehicle speed corresponding to the predicted timestamp, and based on the predicted timestamp, adjacent planning timestamps, and adjacent planning accelerations, determine the reference acceleration corresponding to the predicted timestamp.

[0035] Based on the reference vehicle speed corresponding to each prediction timestamp, a reference vehicle speed sequence is constructed, and based on the reference acceleration corresponding to each prediction timestamp, a reference acceleration sequence is constructed.

[0036] The preview time is the sum of the execution and communication times of each module during the vehicle's perception and control process. For example, it includes the execution and communication times of the localization perception, environmental prediction, decision planning, and motion control modules. The control cycle is the period at which a command is issued to control the vehicle's movement. The planning timestamp is the timestamp corresponding to each reference trajectory point. The prediction timestamp is the timestamp derived by previewing a period of time backward from the current moment, with the control cycle as the time interval. Adjacent planning timestamps are the two planning timestamps adjacent to the prediction timestamp, one before and one after. Adjacent planning velocities and adjacent planning accelerations are the planning velocities and accelerations corresponding to adjacent planning timestamps.

[0037] Specifically, based on the maximum and minimum timestamps among the planned timestamps corresponding to the reference trajectory points, the timestamp intervals requiring interpolation can be determined. Interpolation is performed within these intervals, while extension processing can be directly applied outside of them. By combining the current time, the aiming time, and different numbers of control cycles, the predicted timestamps obtained through interpolation are determined. Then, for each predicted timestamp, interpolation is performed using the two adjacent planned timestamps and their corresponding adjacent planned velocities to obtain the reference velocity corresponding to that predicted timestamp. Similarly, interpolation is performed using the two adjacent planned timestamps and their corresponding adjacent planned accelerations to obtain the reference acceleration corresponding to that predicted timestamp. Furthermore, if there are predicted timestamps before the minimum timestamp, these can be planned as minimum planned timestamps; if there are predicted timestamps after the maximum timestamp, these can be planned as maximum timestamps. Finally, a reference vehicle speed sequence is constructed based on the reference vehicle speeds corresponding to each predicted timestamp, and a reference acceleration sequence is constructed based on the reference accelerations corresponding to each predicted timestamp.

[0038] It's important to note that a typical architecture for autonomous driving consists of perception and localization, environmental prediction, decision planning, and motion control. The process from the vehicle entering a specific environment to adapting and performing actions such as acceleration, deceleration, and steering is sequential. Assuming the vehicle is moving forward at a certain moment, the perception and localization module first detects a cardboard box in front of the vehicle. The environmental prediction module predicts the box is a stationary obstacle. The decision planning module plans a (longitudinal) reference trajectory for deceleration. The motion control module calculates the desired acceleration based on the (longitudinal) reference trajectory, or converts it into torque (aperture) / brake pressure (braking force) and sends it to the drive-by-wire chassis. The total time for this process should include the execution time of each module and the communication time between them. The starting point of the reference trajectory points that the motion control module needs to track should be pre-aimed for a period of time based on the current time point (the current moment), i.e., the pre-aiming time. The (longitudinal) reference trajectory is a discrete sequence of reference trajectory points, each with attributes such as a timestamp, planned speed, and planned acceleration.

[0039] Based on the above example, the predicted timestamps can be determined using the following method: (Based on the current time, aiming time, control period, and the maximum and minimum timestamps among the planned timestamps corresponding to the reference trajectory points.)

[0040] Each predicted timestamp is determined using the following formula:

[0041]

[0042] in, For the current moment, The first time after the current moment A predicted timestamp, For the aiming time, To control the cycle, The minimum timestamp among all the planned timestamps corresponding to the reference trajectory point. This is the largest timestamp among all the planned timestamps corresponding to the reference trajectory point.

[0043] Specifically, based on the current time, the aiming time, and each control cycle, the predicted timestamps are initially determined. If any of these predicted timestamps is earlier than the minimum timestamp, then all predicted timestamps before the minimum timestamp are adjusted to the minimum timestamp. If any of these predicted timestamps is later than the maximum timestamp, then all predicted timestamps later than the maximum timestamp are adjusted to the maximum timestamp. Furthermore, the portion of predicted timestamps between the minimum and maximum timestamps is retained to obtain the final predicted timestamps.

[0044] For example, Figure 2 This is a schematic diagram of a timeline according to an embodiment of this disclosure. For example... Figure 2 As shown, at the minimum timestamp and maximum timestamp The time of the first predicted timestamp between The second predicted timestamp is And so on, the time of the i-th predicted timestamp is .

[0045] After obtaining the predicted timestamps, the velocity and acceleration can be interpolated according to time, which can be achieved through the following two formulas.

[0046] Based on the above example, the reference speed corresponding to the predicted timestamp can be determined using the following method, based on the predicted timestamp, adjacent planning timestamps, and adjacent planning speeds:

[0047]

[0048] in, The first time after the current moment The reference vehicle speed corresponding to each predicted timestamp The first time after the current moment A predicted timestamp, For the position located at the The prediction timestamp before and the one before the first prediction timestamp The planned timestamps are adjacent to the predicted timestamps. For the position located at the After the predicted timestamp and related to the first The planned timestamps are adjacent to the predicted timestamps. It is located at the The adjacent planning velocities corresponding to adjacent planning timestamps prior to the predicted timestamp. It is located at the The adjacent planning velocity corresponding to the adjacent planning timestamp after the prediction timestamp.

[0049] Based on the above example, the reference acceleration corresponding to the predicted timestamp can be determined using the predicted timestamp, adjacent planning timestamps, and adjacent planning accelerations in the following ways:

[0050]

[0051] in, The first time after the current moment Reference acceleration corresponding to each predicted timestamp The first time after the current moment A predicted timestamp, For the position located at the The prediction timestamp before and the one before the first prediction timestamp The planned timestamps are adjacent to the predicted timestamps. For the position located at the After the predicted timestamp and related to the first The planned timestamps are adjacent to the predicted timestamps. It is located at the The adjacent planning acceleration corresponding to the planning timestamp before the predicted timestamp. It is located at the The adjacent planning acceleration corresponding to the adjacent planning timestamp after the prediction timestamp.

[0052] Understandably, current mainstream longitudinal control methods are based on single-point feedback. The aforementioned approach proposes using reference vehicle speed and acceleration sequences extended from the tracking reference trajectory points, utilizing more information and achieving more accurate tracking. Furthermore, this approach considers the execution time and communication time of each module, setting a preview time to acquire the reference vehicle speed and acceleration sequences. This can compensate for the impact of computational time and communication latency in autonomous driving systems, thereby improving vehicle driving safety.

[0053] S120. Based on the expected acceleration increment correlation, the vehicle's current speed, the vehicle's current acceleration, the vehicle's expected acceleration at the previous moment, the vehicle's current speed error integral, the vehicle's current acceleration error integral, the reference speed sequence, and the reference acceleration sequence, minimize the pre-established objective function under the constraints of the preset constraints to determine the expected acceleration increment sequence.

[0054] Among them, the vehicle's current speed error integral is the speed error integral at the current moment. The vehicle's current acceleration error integral is the acceleration error integral at the current moment. The previous moment is the previous control cycle. The expected acceleration increment correlation is the correlation between the actual speed recursive sequence, the actual acceleration recursive sequence, the expected acceleration recursive sequence, the speed error integral sequence, the acceleration error integral sequence, the reference speed sequence, the reference acceleration sequence, and the expected acceleration increment sequence; it can be understood as a correlation between the above eight sequences. The actual speed recursive sequence is a sequence composed of the recursive values ​​of the vehicle's actual speed at each moment. The actual acceleration recursive sequence is a sequence composed of the recursive values ​​of the vehicle's actual acceleration at each moment. The expected acceleration recursive sequence is a sequence composed of the recursive values ​​of the vehicle's expected acceleration at the previous moment. The speed error integral sequence is a sequence composed of the cumulative speed error of the vehicle at each moment. The acceleration error integral sequence is a sequence composed of the cumulative acceleration error of the vehicle at each moment. The desired acceleration increment sequence is a sequence consisting of the differences between the desired acceleration of the vehicle at each time step and the desired acceleration at the previous time step. The objective function is constructed based on the actual vehicle speed recursive sequence, the reference vehicle speed sequence, the actual acceleration recursive sequence, the reference acceleration sequence, the desired acceleration recursive sequence, the vehicle speed error integral sequence, the acceleration error integral sequence, and the desired acceleration increment sequence. Preset constraints include desired acceleration constraints, desired acceleration constraints, and actual acceleration constraints.

[0055] Specifically, by substituting the expected acceleration increment correlation, the vehicle's current speed, the vehicle's current acceleration, the vehicle's expected acceleration at the previous moment, the vehicle's current speed error integral, the vehicle's current acceleration error integral, the reference speed sequence, and the reference acceleration sequence into a pre-established objective function, and minimizing this objective function under preset constraints, the expected acceleration increment sequence can be obtained.

[0056] Based on the above example, the expected acceleration increment correlation could be:

[0057]

[0058] in, This is a set of vectors constructed from the recursive sequences of actual vehicle speed, actual acceleration, expected acceleration, vehicle speed error integral, and acceleration error integral. , The first time after the current moment For each predicted timestamp, the recursive values ​​of actual vehicle speed, actual acceleration, and expected acceleration, the vehicle speed error integral, and the acceleration error integral, 1 ≤ i ≤ , To predict the step size, , , The vehicle's current speed. The vehicle's current acceleration. The expected acceleration of the previous moment in the current moment. Integral of the vehicle's current speed error Let C be the integral of the vehicle's current acceleration error, and C be the first coefficient matrix. , For the next prediction timestamp at the current moment, the actual vehicle speed, actual acceleration, expected acceleration at the previous moment, vehicle speed error integral, and acceleration error integral are given. , The expected acceleration increment at the current moment, For reference only. , This is the reference speed at the current moment. Let F be the reference acceleration at the current moment, and let F be the second coefficient matrix. This is the third coefficient matrix. A is the fourth coefficient matrix, A is the first kinematic matrix, and B is the second kinematic matrix. Let U be the third kinematic matrix, and U be the desired acceleration increment sequence. , To control the period, k is the preset gain. As a preset time constant, For the reference value sequence, .

[0059] The second, third, and fourth coefficient matrices, as well as the first, second, and third kinematic matrices, can be obtained through kinematic model reasoning. The preset gain and preset time constant are pre-defined vehicle kinematic response parameters.

[0060] For example, the above correlation of expected acceleration increments can be derived in the following way:

[0061] A vehicle kinematics model is a model that describes the process from the issuance of a desired acceleration to the vehicle's response acceleration or deceleration. It is basically described by a first-order inertial element.

[0062]

[0063] in, For the desired acceleration, Let the rate of change of the desired acceleration be... For the response acceleration, k is the preset gain. This is the preset time constant.

[0064] Based on the above formula, an acceleration-vehicle speed integrator is added, and the above model is rewritten in discrete form:

[0065]

[0066] in, It is a control cycle. These are the vehicle's current speed, current acceleration, and expected acceleration at the current moment, respectively. These represent the vehicle's speed and acceleration at the next moment, respectively.

[0067] To better account for the desired acceleration increment constraint, the above model can be rewritten in incremental form:

[0068]

[0069] in, These are the expected acceleration of the previous moment and the expected acceleration increment of the current moment, respectively.

[0070] However, this model still has the following problems: the vehicle response has a pure time delay, which is not reflected in the above model; although, realistic parameters are set... and However, under the same load and road gradient, even without a pure time lag, the vehicle's kinematic response still differs somewhat from this model. Vehicle load is not reflected in this model; once the vehicle load changes, the vehicle's motion response will change accordingly. Furthermore, when going uphill or downhill, the vehicle's acceleration response is also affected by gravitational acceleration, a factor not included in this model.

[0071] To address the issues of mismatch between model parameters and actual vehicle response, as well as the omission of some dynamics in the model, an integral term for error can be added, analogous to the integral term in a PID (Proportional-Integral-Differential) controller. This is done because if the model and reality do not match, trajectory tracking will be biased. Reducing this bias can achieve a certain degree of anti-interference effect. The expanded state-space model is as follows:

[0072]

[0073] in, , These are the vehicle's current speed error integral and the vehicle's current acceleration error integral, respectively. , These are the reference vehicle speed and reference acceleration at the current moment, respectively, obtained from various reference trajectory points. The above formula can be simplified as:

[0074]

[0075]

[0076] in, , as well as This can be derived through the above reasoning and related kinematic models.

[0077] Let the prediction step size be... Control step size is The control sequence, i.e., the desired acceleration increment sequence, is... The predicted sequence (a vector set constructed from the actual vehicle speed recursive sequence, the actual acceleration recursive sequence, the expected acceleration recursive sequence, the vehicle speed error integral sequence, and the acceleration error integral sequence) is: Reference value sequence The current state x(n) is known (the vehicle's current speed and acceleration are measurable; if the current acceleration is not measurable, an estimated value can be used instead), and the reference speed and acceleration within the prediction time period are known. The prediction sequence can be represented as a function of the control sequence:

[0078]

[0079] in, , , .

[0080] It should be noted that, It is a preset time constant; the smaller the value, the faster the response. This is the preset gain, which, under steady-state conditions, is greater than 1 when acceleration overshoots and less than 1 when acceleration lags behind. The preset gain k and preset time constant are also relevant. These are the parameters that need to be identified. Only when these two parameters are determined can the above model be fully known. Generally, the acceleration and deceleration actuators of a drive-by-wire chassis are different, so it is necessary to distinguish between the acceleration and deceleration processes. Theoretically, there are two models (driving model and braking model). Therefore, a total of four parameters need to be identified.

[0081] Based on the above example, for the preset gain k and preset time constant It can be identified in the following ways:

[0082] Acquire multiple acceleration and deceleration segments of the vehicle;

[0083] Based on the preset longitudinal driving model and the sample expected acceleration sequence, sample vehicle speed sequence and sample longitudinal acceleration sequence corresponding to the acceleration segment, the driving model parameters to be adjusted are adjusted to obtain the preset driving model parameters.

[0084] Based on the preset longitudinal braking model and the sample expected deceleration sequence, sample vehicle speed sequence, and sample longitudinal deceleration sequence corresponding to the deceleration segment, the braking model parameters to be adjusted are obtained to obtain the preset braking model parameters.

[0085] The acceleration / deceleration segments are extracted from the autonomous driving data transmitted back from the vehicle. These segments are continuous processes with either non-negative or non-positive expected acceleration. An acceleration segment contains three sequences: a sample expected acceleration sequence, a sample vehicle speed sequence, and a sample longitudinal acceleration sequence. A deceleration segment contains three sequences: a sample expected deceleration sequence, a sample vehicle speed sequence, and a sample longitudinal deceleration sequence. If the transmitted autonomous driving data does not record sample longitudinal acceleration or deceleration in real time, it can be estimated using methods such as time difference analysis, differential trackers, and Kalman filters. The parameters of the driving model to be adjusted include the time constant and gain. The preset driving model parameters include the time constant and gain. The parameters of the braking model to be adjusted include the time constant and gain. The preset braking model parameters include the time constant and gain. The preset time constant includes the time constant of the preset driving model and the time constant of the preset braking model; the preset gain includes the gain of the preset driving model and the gain of the preset braking model. The preset longitudinal drive model is a pre-built drive model, which can be a first-order inertial drive model. The preset longitudinal braking model is a pre-built braking model, which can be a first-order inertial braking model.

[0086] Specifically, multiple acceleration and deceleration segments of the vehicle are acquired. The expected acceleration sequence, sample vehicle speed sequence, and sample longitudinal acceleration sequence corresponding to each acceleration segment are calculated and compared according to a preset longitudinal drive model. The results of these calculations and comparisons are then used to adjust the parameters of the drive model to be adjusted, resulting in the preset drive model parameters. Similarly, the expected deceleration sequence, sample vehicle speed sequence, and sample longitudinal deceleration sequence corresponding to each deceleration segment are calculated and compared according to a preset longitudinal braking model. The results of these calculations and comparisons are then used to adjust the parameters of the braking model to be adjusted, resulting in the preset braking model parameters.

[0087] Based on the above example, the parameters of the driving model to be adjusted can be obtained by adjusting the parameters of the driving model to be adjusted according to the preset longitudinal driving model and the sample expected acceleration sequence, sample vehicle speed sequence, and sample longitudinal acceleration sequence corresponding to the acceleration segment, in the following way:

[0088] Based on the sample expected acceleration sequence, sample vehicle speed sequence, sample longitudinal acceleration sequence, and the preset longitudinal driving model minimizing the pre-established driving model loss function, the driving model parameters to be adjusted corresponding to minimizing the pre-established driving model loss function are determined as the preset driving model parameters.

[0089] The preset longitudinal driving model is used to determine the output vehicle speed sequence and output acceleration sequence corresponding to the acceleration segment based on the driving model parameters to be adjusted, the first term of the sample vehicle speed sequence, and the sample expected acceleration sequence corresponding to the acceleration segment. The pre-established driving model loss function is a loss function constructed based on the output vehicle speed sequence, output acceleration sequence, sample vehicle speed sequence corresponding to the acceleration segment, and sample longitudinal acceleration sequence, and is used to measure the response deviation of the preset longitudinal driving model.

[0090] Specifically, by processing the parameters of the driving model to be adjusted, the first term of the sample vehicle speed sequence, and the expected acceleration sequence of the sample corresponding to the acceleration segment using a preset longitudinal driving model, the output vehicle speed sequence and the output acceleration sequence corresponding to the acceleration segment can be obtained. The driving model loss function, established by the loss between the output vehicle speed sequence and the sample vehicle speed sequence, and the loss between the output acceleration sequence and the sample longitudinal acceleration sequence, determines the parameters of the driving model to be adjusted when the loss is minimized as the preset driving model parameters.

[0091] It should be noted that the process of determining the preset braking model parameters is similar to that of determining the preset driving model parameters. It can be as follows: Based on the sample expected deceleration sequence, sample vehicle speed sequence, sample longitudinal deceleration sequence, and the preset longitudinal braking model minimizing the pre-established braking model loss function, the braking model parameters to be adjusted corresponding to minimizing the pre-established braking model loss function are determined as the preset braking model parameters. Specifically, the preset longitudinal braking model is used to determine the output vehicle speed sequence and output deceleration sequence corresponding to the deceleration segment based on the braking model parameters to be adjusted, the first term of the sample vehicle speed sequence, and the sample expected deceleration sequence corresponding to the deceleration segment. The pre-established braking model loss function is a loss function constructed based on the output vehicle speed sequence, output deceleration sequence, sample vehicle speed sequence corresponding to the deceleration segment, and sample longitudinal deceleration sequence. The specific process is similar to that of determining the preset driving model parameters and will not be elaborated here.

[0092] For example, for the acceleration process, given an acceleration segment, we know the initial vehicle speed (the first item in the sample vehicle speed sequence) and the expected acceleration sequence within this segment (the sample expected acceleration sequence). Assuming τ and k (the parameters of the driving model to be adjusted) are known, we can simulate the vehicle's response speed and acceleration during this time period (output speed sequence and output acceleration sequence) based on the preset longitudinal driving model. The simulated output speed and acceleration sequences will inevitably deviate from the sample speed and longitudinal acceleration sequences recorded in the real vehicle log. We construct a loss function for the speed and acceleration deviation sequences to describe the difference between the preset longitudinal driving model and the system's actual response. Considering the deviations of all acceleration segments, we can obtain the driving model loss function:

[0093]

[0094] In the formula, , These are the initial vehicle speed (the first item in the sample vehicle speed sequence) and the expected acceleration sequence of the i-th acceleration segment, respectively. It is the sample vehicle speed sequence and sample longitudinal acceleration sequence of the i-th acceleration segment. This represents the simulation output of the preset longitudinal drive model. The loss function can be SSE, MSE, RMSE, or MAE.

[0095] As can be seen from the above equation, the loss function of the driving model is and If a function is defined, then minimizing this function will find the closest approximation to the vehicle's kinematic drive response. and Regarding the deceleration process, and The identification process is similar. In the process of minimizing this function, we can also give... and Add constraints to ensure they are within a reasonable range.

[0096] From the perspective of the implementation time of parameter (preset time constant and preset gain) identification, it can be divided into offline identification and online identification. Offline identification is performed when the autonomous driving system is not running, while online identification is performed simultaneously with the autonomous driving system. Theoretically, online identification can reflect changes in vehicle response in the vehicle's longitudinal kinematics model in a timely manner. However, online identification faces challenges such as maintaining historical data, filtering new data, switching acceleration / deceleration models, and solving optimization problems. When offline identification parameters are applied online, the vehicle response may change, but the difference can be compensated for by other means. This example can use the offline parameter identification method.

[0097] The calibration of the vehicle's longitudinal kinematic model parameters (preset time constant and preset gain) can be performed during the drive-by-wire chassis acceptance phase, using the step response identification method for offline parameter identification to determine the values ​​of parameters τ and k. However, the vehicle's response performance may change over time, and in many cases, drive-by-wire chassis calibration and acceptance operations are no longer suitable after the vehicle is put into operation. Therefore, using autonomous driving feedback data for offline parameter identification allows consideration of the vehicle's response characteristics across the entire speed range, and re-identification can be performed when needed without requiring separate sites and testing procedures. Furthermore, during drive-by-wire chassis acceptance, lateral control is in an open-loop state, and performing longitudinal control interface performance testing carries certain risks; therefore, it is often required to be conducted on straight roads. However, this example avoids this problem by directly using feedback data for parameter identification, because the vehicle's lateral control in autonomous driving mode can already track the desired path, and the scenario can be achieved on both straight and curved roads.

[0098] Building upon the above example, the preset constraints are primarily considered from two perspectives: ride comfort and actuator performance. Ride comfort is characterized by the desired jerk; the smaller the desired jerk, the better the ride comfort, thus requiring constraints on the desired jerk. The actuator performance is mainly reflected in the vehicle's maximum driving acceleration and maximum braking deceleration. These two parameters are physically limited, therefore, constraints on both actual and desired acceleration are necessary.

[0099] The desired jerk constraints include:

[0100]

[0101] in, This represents the expected acceleration increment corresponding to the m-th step size after the current time. To control the cycle, To control the step size, To preset the minimum desired jerk, The preset maximum expected accelerometer.

[0102] Actual acceleration constraints include:

[0103]

[0104] in, The first time after the current moment The actual acceleration recursive value corresponding to each predicted timestamp To predict the step size, To preset the minimum acceleration, This is the preset maximum acceleration.

[0105] Desired acceleration constraints include:

[0106]

[0107] in, The first time after the current moment The expected acceleration recursive value corresponding to each predicted timestamp To predict the step size, To preset the minimum acceleration, This is the preset maximum acceleration.

[0108] The optimization objectives are mainly considered from three perspectives: tracking performance, fuel economy, and smoothness. Tracking performance is primarily reflected by the sum of squared errors between the actual vehicle speed / acceleration and the reference vehicle speed / acceleration. Fuel economy is reflected by the magnitude of the expected acceleration; the greater the expected acceleration, the more aggressively the accelerator is pressed, resulting in higher fuel consumption. Smoothness is reflected by the magnitude of the increment in expected acceleration; the smaller the increment in expected acceleration, the slower the change in actual acceleration, and the smoother the ride. Therefore, combining the requirements of minimizing the integral of vehicle speed error and the integral of acceleration error, the objective function is constructed as follows:

[0109]

[0110] in, Let be the objective function. The first time after the current moment The actual vehicle speed recursive value corresponding to each predicted timestamp The first time after the current moment The reference vehicle speed corresponding to each predicted timestamp The first time after the current moment The actual acceleration recursive value corresponding to each predicted timestamp The first time after the current moment Reference acceleration corresponding to each predicted timestamp The first time after the current moment -1 expected acceleration recursive value corresponding to the predicted timestamp The first time after the current moment The vehicle speed error integral corresponding to each predicted timestamp The first time after the current moment The acceleration error integral corresponding to each predicted timestamp The first time after the current moment The expected acceleration increment corresponding to each predicted timestamp To predict the step size, To control the step size, , , , as well as All of these are preset weight parameters, and each preset weight parameter needs to be adjusted according to the actual application.

[0111] In addition to the constraints and performance indicators mentioned above, slack variables can be used to rewrite the constraints on the reference acceleration and actual acceleration as soft constraints, and penalty terms for slack variables can be added to the optimization objective. The objective function and constraints define an optimization model, and solving this model yields the value of the sequence U (the desired acceleration increment sequence).

[0112] Understandably, the above method, when applying model predictive control to the longitudinal control of a vehicle, considers the vehicle's kinematic response characteristics. Therefore, specific compensation can be made based on the speed of the vehicle's transient response and the magnitude of its steady-state response; a slower or lagging response will trigger corresponding compensation, theoretically leading to more accurate tracking. To address the possibility of a mismatch between the model and the actual vehicle response, a state error integral term (vehicle speed error integral and acceleration error integral) is added as a state, and a corresponding penalty term is included in the objective function. If the identified parameters do not match the vehicle response, such as due to load changes or inclines / declines, there will be errors in speed and acceleration tracking. Once this mismatch persists, the error will increase. By reducing the error integral term, the tracking error can be gradually reduced, ultimately achieving unbiased tracking.

[0113] S130. The sum of the first term in the expected acceleration increment sequence and the expected acceleration of the vehicle at the previous time is determined as the expected acceleration at the current time.

[0114] Specifically, the first term of the expected acceleration increment sequence is taken as the expected acceleration increment at the current moment. Adding the expected acceleration of the previous moment to the current moment gives the expected acceleration at the current moment.

[0115] The longitudinal control method provided in this embodiment determines a reference vehicle speed sequence based on the planned speed of the reference trajectory points and a reference acceleration sequence based on the planned acceleration of the reference trajectory points, thereby expanding the tracking time points and improving subsequent tracking performance. Furthermore, based on the expected acceleration increment correlation, the vehicle's current speed, the vehicle's current acceleration, the vehicle's expected acceleration at the previous moment, the vehicle's current speed error integral, the vehicle's current acceleration error integral, the reference vehicle speed sequence, and the reference acceleration sequence, the method minimizes a pre-established objective function under preset constraints to determine the expected acceleration increment sequence. The sum of the first term in the expected acceleration increment sequence and the vehicle's expected acceleration at the previous moment is determined as the expected acceleration at the current moment, achieving safer, more accurate, and smoother longitudinal control.

[0116] Figure 3This is a schematic diagram of the structure of a longitudinal control device according to an embodiment of this disclosure. Figure 3 As shown, the device includes: a reference value sequence determination module 210, an objective function solution module 220, and a desired acceleration determination module 230.

[0117] The reference value sequence determination module 210 is used to determine a reference vehicle speed sequence based on the planned speed of the reference trajectory points, and to determine a reference acceleration sequence based on the planned acceleration of the reference trajectory points. The reference trajectory points are the trajectory points for the planned subsequent tracking of the vehicle; the planned speed of the reference trajectory points is the planned speed of the vehicle when it reaches the reference trajectory point; and the planned acceleration of the reference trajectory points is the planned acceleration of the vehicle when it reaches the reference trajectory point. The objective function solving module 220 is used to minimize a pre-established objective function under preset constraints, based on the expected acceleration increment correlation, the vehicle's current speed, the vehicle's current acceleration, the vehicle's expected acceleration at the previous moment, the vehicle's current speed error integral, the vehicle's current acceleration error integral, the reference vehicle speed sequence, and the reference acceleration sequence, to determine the expected acceleration increment sequence. The expected acceleration increment correlation consists of the actual vehicle speed recursive sequence, the actual acceleration recursive sequence, the expected acceleration recursive sequence, the vehicle speed error integral sequence, and the acceleration error... The correlation between the difference integral sequence, the reference vehicle speed sequence, the reference acceleration sequence, and the expected acceleration increment sequence; the actual vehicle speed recursive sequence is a sequence composed of the recursive values ​​of the actual speed of the vehicle at each time moment, the actual acceleration recursive sequence is a sequence composed of the recursive values ​​of the actual acceleration of the vehicle at each time moment, and the expected acceleration recursive sequence is a sequence composed of the recursive values ​​of the expected acceleration of the vehicle at each time moment compared to the previous time moment; the objective function is a function constructed based on the actual vehicle speed recursive sequence, the reference vehicle speed sequence, the actual acceleration recursive sequence, the reference acceleration sequence, the expected acceleration recursive sequence, the vehicle speed error integral sequence, the acceleration error integral sequence, and the expected acceleration increment sequence; the preset constraints include expected acceleration constraints, expected acceleration constraints, and actual acceleration constraints; the expected acceleration determination module 230 is used to determine the expected acceleration at the current time as the sum of the first term in the expected acceleration increment sequence and the expected acceleration of the vehicle at the previous time moment.

[0118] Based on the above example, optionally, the reference value sequence determination module 210 is further configured to determine each predicted timestamp based on the current time, the aiming time, the control cycle, and the maximum and minimum timestamps among the planned timestamps corresponding to each reference trajectory point; for each predicted timestamp, determine the adjacent planned timestamps corresponding to the predicted timestamp, as well as the adjacent planned speed and adjacent planned acceleration corresponding to each adjacent planned timestamp; determine the reference vehicle speed corresponding to the predicted timestamp based on the predicted timestamp, the adjacent planned timestamps, and the adjacent planned speeds, and determine the reference acceleration corresponding to the predicted timestamp based on the predicted timestamp, the adjacent planned timestamps, and the adjacent planned accelerations; construct a reference vehicle speed sequence based on the reference vehicle speeds corresponding to each predicted timestamp, and construct a reference acceleration sequence based on the reference accelerations corresponding to each predicted timestamp.

[0119] Building upon the example above, optionally, the reference value sequence determination module 210 is also used to determine each predicted timestamp using the following formula:

[0120]

[0121] in, For the current moment, The first time after the current moment A predicted timestamp, For the aiming time, To control the cycle, The minimum timestamp among all the planned timestamps corresponding to the reference trajectory point. The maximum timestamp among all the planned timestamps corresponding to the reference trajectory point;

[0122] The reference value sequence determination module 210 is further configured to determine the reference vehicle speed corresponding to the predicted timestamp using the following formula:

[0123]

[0124] in, The first time after the current moment The reference vehicle speed corresponding to each predicted timestamp The first time after the current moment A predicted timestamp, For the position located at the The prediction timestamp before and the one before the first prediction timestamp The planned timestamps are adjacent to the predicted timestamps. For the position located at the After the predicted timestamp and related to the first The planned timestamps are adjacent to the predicted timestamps. It is located at the The adjacent planning velocities corresponding to adjacent planning timestamps prior to the predicted timestamp. It is located at the The adjacent planning velocity corresponding to the adjacent planning timestamp after the prediction timestamp;

[0125] The reference value sequence determination module 210 is further configured to determine the reference acceleration corresponding to the predicted timestamp using the following formula:

[0126]

[0127] in, The first time after the current moment Reference acceleration corresponding to each predicted timestamp The first time after the current moment A predicted timestamp, For the position located at the The prediction timestamp before and the one before the first prediction timestamp The planned timestamps are adjacent to the predicted timestamps. For the position located at the After the predicted timestamp and related to the first The planned timestamps are adjacent to the predicted timestamps. It is located at the The adjacent planning acceleration corresponding to the planning timestamp before the predicted timestamp. It is located at the The adjacent planning acceleration corresponding to the adjacent planning timestamp after the prediction timestamp.

[0128] Based on the above example, optionally, the expected acceleration increment correlation includes:

[0129]

[0130] in, This is a set of vectors constructed from the recursive sequences of actual vehicle speed, actual acceleration, expected acceleration, vehicle speed error integral, and acceleration error integral. , The first time after the current moment For each predicted timestamp, the recursive values ​​of actual vehicle speed, actual acceleration, and expected acceleration, the vehicle speed error integral, and the acceleration error integral, 1 ≤ i ≤ , To predict the step size, , , The vehicle's current speed. The vehicle's current acceleration. The expected acceleration of the previous moment in the current moment. Integral of the vehicle's current speed error Let C be the integral of the vehicle's current acceleration error, and C be the first coefficient matrix. , For the next prediction timestamp at the current moment, the actual vehicle speed, actual acceleration, expected acceleration at the previous moment, vehicle speed error integral, and acceleration error integral are given. , The expected acceleration increment at the current moment, For reference only. , This is the reference speed at the current moment. Let F be the reference acceleration at the current moment, and let F be the second coefficient matrix. This is the third coefficient matrix. A is the fourth coefficient matrix, A is the first kinematic matrix, and B is the second kinematic matrix. Let U be the third kinematic matrix, and U be the desired acceleration increment sequence. , To control the period, k is the preset gain. As a preset time constant, For the reference value sequence, .

[0131] Based on the above example, optionally, the device further includes: a parameter identification module, used to acquire multiple acceleration segments and multiple deceleration segments of the vehicle; adjust the parameters of the driving model to be adjusted according to a preset longitudinal driving model and the sample expected acceleration sequence, sample vehicle speed sequence, and sample longitudinal acceleration sequence corresponding to the acceleration segments to obtain preset driving model parameters; wherein, the driving model parameters to be adjusted include the driving model time constant and the driving model gain to be adjusted, and the preset driving model parameters include a preset driving model time constant and a preset driving model gain; adjust the parameters of the braking model to be adjusted according to a preset longitudinal braking model and the sample expected deceleration sequence, sample vehicle speed sequence, and sample longitudinal deceleration sequence corresponding to the deceleration segments to obtain preset braking model parameters; wherein, the braking model parameters to be adjusted include the braking model time constant and the braking model gain to be adjusted, and the preset braking model parameters include a preset braking model time constant and a preset braking model gain; wherein, the preset time constant includes a preset driving model time constant and a preset braking model time constant, and the preset gain includes a preset driving model gain and a preset braking model gain.

[0132] Based on the above example, optionally, the parameter identification module is further configured to determine the driving model parameters to be adjusted corresponding to minimizing the pre-established driving model loss function based on the sample expected acceleration sequence, the sample vehicle speed sequence, the sample longitudinal acceleration sequence, and the preset longitudinal driving model minimizing the pre-established driving model loss function, and to determine the preset driving model parameters; wherein, the preset longitudinal driving model is used to determine the output vehicle speed sequence and output acceleration sequence corresponding to the acceleration segment based on the driving model parameters to be adjusted, the first item of the sample vehicle speed sequence, and the sample expected acceleration sequence corresponding to the acceleration segment; the pre-established driving model loss function is a loss function constructed based on the output vehicle speed sequence, the output acceleration sequence, the sample vehicle speed sequence corresponding to the acceleration segment, and the sample longitudinal acceleration sequence.

[0133] Based on the above example, optionally, the desired jerk constraint conditions include:

[0134]

[0135] in, This represents the expected acceleration increment corresponding to the m-th step size after the current time. To control the cycle, To control the step size, To preset the minimum desired jerk, The preset maximum expected jerk;

[0136] The actual acceleration constraints include:

[0137]

[0138] in, The first time after the current moment The actual acceleration recursive value corresponding to each predicted timestamp To predict the step size, To preset the minimum acceleration, The preset maximum acceleration;

[0139] The desired acceleration constraint conditions include:

[0140]

[0141] in, The first time after the current moment The expected acceleration recursive value corresponding to each predicted timestamp To predict the step size, To preset the minimum acceleration, The preset maximum acceleration;

[0142] The objective function includes:

[0143]

[0144] in, Let be the objective function. The first time after the current moment The actual vehicle speed recursive value corresponding to each predicted timestamp The first time after the current moment The reference vehicle speed corresponding to each predicted timestamp The first time after the current moment The actual acceleration recursive value corresponding to each predicted timestamp The first time after the current moment Reference acceleration corresponding to each predicted timestamp The first time after the current moment -1 expected acceleration recursive value corresponding to the predicted timestamp The first time after the current moment The vehicle speed error integral corresponding to each predicted timestamp The first time after the current moment The acceleration error integral corresponding to each predicted timestamp The first time after the current moment The expected acceleration increment corresponding to each predicted timestamp To predict the step size, To control the step size, , , , as well as All of these are preset weight parameters.

[0145] The longitudinal control device provided in this disclosure can execute the steps in the longitudinal control method provided in this disclosure, and has the execution steps and beneficial effects, which will not be elaborated here.

[0146] Figure 4 This is a schematic diagram of the structure of an electronic device according to an embodiment of this disclosure. See below for details. Figure 4 It shows a schematic diagram of a structure suitable for implementing the electronic device 300 in the embodiments of this disclosure. Figure 4 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0147] like Figure 4As shown, the electronic device 300 may include a processing device 301, a read-only memory (ROM) 302, a random access memory (RAM) 303, a bus 304, an input / output (I / O) interface 305, an input device 306, an output device 307, a storage device 308, and a communication device 309. The processing device (e.g., a central processing unit, a graphics processor, etc.) 301 can perform various appropriate actions and processes to implement the methods of the embodiments described in this disclosure, based on a program in the ROM 302 or a program loaded from the storage device 308 into the RAM 303. The RAM 303 also stores various programs and data required for the operation of the electronic device 300. The processing device 301, the ROM 302, and the RAM 303 are interconnected via the bus 304. The input / output (I / O) interface 305 is also connected to the bus 304.

[0148] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts, thereby implementing the longitudinal control method as described above. In such embodiments, the computer program can be downloaded and installed from a network via communication device 309, or installed from storage device 308, or installed from ROM 302. When the computer program is executed by processing device 301, it performs the functions defined in the methods of embodiments of this disclosure.

[0149] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0150] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to:

[0151] Based on the planned speed of the reference trajectory points, a reference vehicle speed sequence is determined, and based on the planned acceleration of the reference trajectory points, a reference acceleration sequence is determined; wherein, the reference trajectory points are the planned trajectory points for the subsequent tracking of the vehicle, the planned speed of the reference trajectory points is the planned speed of the vehicle when it reaches the reference trajectory points, and the planned acceleration of the reference trajectory points is the planned acceleration of the vehicle when it reaches the reference trajectory points.

[0152] Based on the expected acceleration increment correlation, the vehicle's current speed, vehicle's current acceleration, the vehicle's expected acceleration at the previous moment, the vehicle's current speed error integral, the vehicle's current acceleration error integral, the reference speed sequence, and the reference acceleration sequence, the pre-established objective function is minimized under preset constraints to determine the expected acceleration increment sequence; wherein, the expected acceleration increment correlation is the correlation between the actual speed recursive sequence, the actual acceleration recursive sequence, the expected acceleration recursive sequence, the speed error integral sequence, the acceleration error integral sequence, the reference speed sequence, the reference acceleration sequence, and the expected acceleration increment sequence; the actual speed recursive sequence is the correlation between the vehicle's current speed, current acceleration, current acceleration, the vehicle's current acceleration at the previous moment, the vehicle's current speed error integral, the vehicle's current acceleration error integral, the reference speed sequence, the reference acceleration sequence, and the expected acceleration increment sequence; the actual speed recursive sequence is the correlation between the vehicle's current speed, current acceleration, current acceleration at the previous moment, the vehicle's current speed error integral, the vehicle's current acceleration error integral, the reference speed sequence, the reference acceleration sequence, and the expected acceleration increment sequence; The objective function is a sequence composed of the actual vehicle speed recursive values ​​at each moment, the actual acceleration recursive sequence being a sequence composed of the actual acceleration recursive values ​​of the vehicle at each moment, and the desired acceleration recursive sequence being a sequence composed of the desired acceleration recursive values ​​of the vehicle at each moment from the previous moment; the objective function is a function constructed based on the actual vehicle speed recursive sequence, the reference vehicle speed sequence, the actual acceleration recursive sequence, the reference acceleration sequence, the desired acceleration recursive sequence, the vehicle speed error integral sequence, the acceleration error integral sequence, and the desired acceleration increment sequence; the preset constraints include desired acceleration constraints, desired acceleration constraints, and actual acceleration constraints.

[0153] The sum of the first term in the expected acceleration increment sequence and the expected acceleration of the vehicle at the previous time is determined as the expected acceleration at the current time.

[0154] Optionally, when one or more of the above-described procedures are executed by the electronic device, the electronic device may also perform other steps described in the above embodiments.

[0155] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0156] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

Claims

1. A longitudinal control method, characterized in that, The method includes: Based on the planned speed of the reference trajectory points, a reference vehicle speed sequence is determined, and based on the planned acceleration of the reference trajectory points, a reference acceleration sequence is determined; wherein, the reference trajectory points are the planned trajectory points for the subsequent tracking of the vehicle, the planned speed of the reference trajectory points is the planned speed of the vehicle when it reaches the reference trajectory points, and the planned acceleration of the reference trajectory points is the planned acceleration of the vehicle when it reaches the reference trajectory points. Based on the expected acceleration increment correlation, the vehicle's current speed, vehicle's current acceleration, the vehicle's expected acceleration at the previous moment, the vehicle's current speed error integral, the vehicle's current acceleration error integral, the reference speed sequence, and the reference acceleration sequence, a pre-established objective function is minimized under preset constraints to determine the expected acceleration increment sequence. The expected acceleration increment correlation is the correlation between the actual speed recursive sequence, the actual acceleration recursive sequence, the expected acceleration recursive sequence, the speed error integral sequence, the acceleration error integral sequence, the reference speed sequence, the reference acceleration sequence, and the expected acceleration increment sequence. The actual speed recursive sequence is the correlation between the vehicle's current speed at each... The objective function is a sequence of actual vehicle speed recursive values ​​at each moment, where the actual acceleration recursive sequence is the sequence of actual acceleration recursive values ​​of the vehicle at each moment, and the desired acceleration recursive sequence is the sequence of desired acceleration recursive values ​​of the vehicle at each moment from the previous moment. The objective function is a function constructed based on the actual vehicle speed recursive sequence, the reference vehicle speed sequence, the actual acceleration recursive sequence, the reference acceleration sequence, the desired acceleration recursive sequence, the vehicle speed error integral sequence, the acceleration error integral sequence, and the desired acceleration increment sequence. The preset constraints include desired acceleration constraints, desired acceleration constraints, and actual acceleration constraints. The sum of the first term in the expected acceleration increment sequence and the expected acceleration of the vehicle at the previous time is determined as the expected acceleration at the current time. The correlation of the expected acceleration increment includes: in, This is a set of vectors constructed from the recursive sequences of actual vehicle speed, actual acceleration, expected acceleration, vehicle speed error integral, and acceleration error integral. , The first time after the current moment For each predicted timestamp, the recursive values ​​of actual vehicle speed, actual acceleration, and expected acceleration, the vehicle speed error integral, and the acceleration error integral, 1 ≤ i ≤ , To predict the step size, , , The vehicle's current speed. The vehicle's current acceleration. The expected acceleration of the previous moment in the current moment. Integral of the vehicle's current speed error Let C be the integral of the vehicle's current acceleration error, and C be the first coefficient matrix. , For the next prediction timestamp at the current moment, the actual vehicle speed, actual acceleration, expected acceleration at the previous moment, vehicle speed error integral, and acceleration error integral are given. , The expected acceleration increment at the current moment, For reference only. , This is the reference speed at the current moment. Let F be the reference acceleration at the current moment, and let F be the second coefficient matrix. This is the third coefficient matrix. A is the fourth coefficient matrix, A is the first kinematic matrix, and B is the second kinematic matrix. Let U be the third kinematic matrix, and U be the desired acceleration increment sequence. , To control the period, k is the preset gain. As a preset time constant, For the reference value sequence, .

2. The method according to claim 1, characterized in that, The step of determining a reference vehicle speed sequence based on the planned velocity of the reference trajectory points, and determining a reference acceleration sequence based on the planned acceleration of the reference trajectory points, includes: Based on the current time, the aiming time, the control period, and the maximum and minimum timestamps among the planned timestamps corresponding to the reference trajectory points, determine each predicted timestamp; For each predicted timestamp, determine the adjacent planning timestamps corresponding to the predicted timestamps, as well as the adjacent planning velocities and adjacent planning accelerations corresponding to each of the adjacent planning timestamps; Based on the predicted timestamp, the adjacent planning timestamp, and the adjacent planning speed, a reference vehicle speed corresponding to the predicted timestamp is determined, and based on the predicted timestamp, the adjacent planning timestamp, and the adjacent planning acceleration, a reference acceleration corresponding to the predicted timestamp is determined. Based on the reference vehicle speed corresponding to each predicted timestamp, a reference vehicle speed sequence is constructed, and based on the reference acceleration corresponding to each predicted timestamp, a reference acceleration sequence is constructed.

3. The method according to claim 2, characterized in that, The process of determining each predicted timestamp based on the maximum and minimum timestamps among the planned timestamps corresponding to the current time, the pre-aiming time, the control cycle, and the reference trajectory point includes: Each predicted timestamp is determined using the following formula: in, For the current moment, The first time after the current moment A predicted timestamp, For the aiming time, To control the cycle, The minimum timestamp among all the planned timestamps corresponding to the reference trajectory point. The maximum timestamp among all the planned timestamps corresponding to the reference trajectory point; The step of determining the reference vehicle speed corresponding to the predicted timestamp based on the predicted timestamp, the adjacent planning timestamps, and the adjacent planning speeds includes: The reference vehicle speed corresponding to the predicted timestamp is determined using the following formula: in, The first time after the current moment The reference vehicle speed corresponding to each predicted timestamp The first time after the current moment A predicted timestamp, For the position located at the The prediction timestamp before and the one before the first prediction timestamp The planned timestamps are adjacent to the predicted timestamps. For the position located at the After the predicted timestamp and related to the first The planned timestamps are adjacent to the predicted timestamps. It is located at the The adjacent planning velocities corresponding to adjacent planning timestamps prior to the predicted timestamp. It is located at the The adjacent planning velocity corresponding to the adjacent planning timestamp after the prediction timestamp; The step of determining the reference acceleration corresponding to the predicted timestamp based on the predicted timestamp, the adjacent planning timestamps, and the adjacent planning acceleration includes: The reference acceleration corresponding to the predicted timestamp is determined using the following formula: in, The first time after the current moment Reference acceleration corresponding to each predicted timestamp The first time after the current moment A predicted timestamp, For the position located at the The prediction timestamp before and the one before the first prediction timestamp The planned timestamps are adjacent to the predicted timestamps. For the position located at the After the predicted timestamp and related to the first The planned timestamps are adjacent to the predicted timestamps. It is located at the The adjacent planning acceleration corresponding to the planning timestamp before the predicted timestamp. It is located at the The adjacent planning acceleration corresponding to the adjacent planning timestamp after the prediction timestamp.

4. The method according to claim 1, characterized in that, Also includes: Acquire multiple acceleration and deceleration segments of the vehicle; Based on the preset longitudinal driving model and the sample expected acceleration sequence, sample vehicle speed sequence and sample longitudinal acceleration sequence corresponding to the acceleration segment, the driving model parameters to be adjusted are adjusted to obtain the preset driving model parameters; wherein, the driving model parameters to be adjusted include the driving model time constant and the driving model gain to be adjusted, and the preset driving model parameters include the preset driving model time constant and the preset driving model gain. Based on the preset longitudinal braking model and the sample expected deceleration sequence, sample vehicle speed sequence and sample longitudinal deceleration sequence corresponding to the deceleration segment, the braking model parameters to be adjusted are adjusted to obtain the preset braking model parameters; wherein, the braking model parameters to be adjusted include the braking model time constant and the braking model gain to be adjusted, and the preset braking model parameters include the preset braking model time constant and the preset braking model gain. The preset time constant includes a preset driving model time constant and a preset braking model time constant, and the preset gain includes a preset driving model gain and a preset braking model gain.

5. The method according to claim 4, characterized in that, The step involves adjusting the parameters of the driving model to be adjusted based on the preset longitudinal driving model and the sample expected acceleration sequence, sample vehicle speed sequence, and sample longitudinal acceleration sequence corresponding to the acceleration segment, to obtain the preset driving model parameters, including: Based on the sample expected acceleration sequence, the sample vehicle speed sequence, the sample longitudinal acceleration sequence, and the preset longitudinal driving model minimizing the pre-established driving model loss function, the driving model parameters to be adjusted corresponding to minimizing the pre-established driving model loss function are determined as the preset driving model parameters. The preset longitudinal driving model is used to determine the output vehicle speed sequence and output acceleration sequence corresponding to the acceleration segment based on the driving model parameters to be adjusted, the first item of the sample vehicle speed sequence, and the sample expected acceleration sequence corresponding to the acceleration segment; the pre-established driving model loss function is a loss function constructed based on the output vehicle speed sequence, the output acceleration sequence, the sample vehicle speed sequence corresponding to the acceleration segment, and the sample longitudinal acceleration sequence.

6. The method according to claim 1, characterized in that, The desired acceleration constraint conditions include: in, This represents the expected acceleration increment corresponding to the m-th step size after the current time. To control the cycle, To control the step size, To preset the minimum desired jerk, The preset maximum expected jerk; The actual acceleration constraints include: in, The first time after the current moment The actual acceleration recursive value corresponding to each predicted timestamp To predict the step size, To preset the minimum acceleration, The preset maximum acceleration; The desired acceleration constraint conditions include: in, The first time after the current moment The expected acceleration recursive value corresponding to each predicted timestamp To predict the step size, To preset the minimum acceleration, The preset maximum acceleration; The objective function includes: in, Let be the objective function. The first time after the current moment The actual vehicle speed recursive value corresponding to each predicted timestamp The first time after the current moment The reference vehicle speed corresponding to each predicted timestamp The first time after the current moment The actual acceleration recursive value corresponding to each predicted timestamp The first time after the current moment Reference acceleration corresponding to each predicted timestamp The first time after the current moment -1 expected acceleration recursive value corresponding to the predicted timestamp The first time after the current moment The vehicle speed error integral corresponding to each predicted timestamp The first time after the current moment The acceleration error integral corresponding to each predicted timestamp The first time after the current moment The expected acceleration increment corresponding to each predicted timestamp To predict the step size, To control the step size, , , , as well as All of these are preset weight parameters.

7. A longitudinal control device, characterized in that, include: The reference value sequence determination module is used to determine a reference vehicle speed sequence based on the planned speed of the reference trajectory points, and to determine a reference acceleration sequence based on the planned acceleration of the reference trajectory points; wherein, the reference trajectory points are the trajectory points for the planned subsequent tracking of the vehicle, the planned speed of the reference trajectory points is the planned speed of the vehicle when it reaches the reference trajectory points, and the planned acceleration of the reference trajectory points is the planned acceleration of the vehicle when it reaches the reference trajectory points; The objective function solving module is used to minimize a pre-established objective function under preset constraints, based on the expected acceleration increment correlation, the vehicle's current speed, the vehicle's current acceleration, the vehicle's expected acceleration at the previous time step, the vehicle's current speed error integral, the vehicle's current acceleration error integral, the reference speed sequence, and the reference acceleration sequence, to determine the expected acceleration increment sequence. The expected acceleration increment correlation is the correlation between the actual speed recursive sequence, the actual acceleration recursive sequence, the expected acceleration recursive sequence, the speed error integral sequence, the acceleration error integral sequence, the reference speed sequence, the reference acceleration sequence, and the expected acceleration increment sequence. The actual speed recursive sequence is... The sequence of actual vehicle speed recursive values ​​at each moment, the actual acceleration recursive sequence being the sequence of actual acceleration recursive values ​​at each moment, and the expected acceleration recursive sequence being the sequence of expected acceleration recursive values ​​of the vehicle at each moment from the previous moment; the objective function is a function constructed based on the actual vehicle speed recursive sequence, the reference vehicle speed sequence, the actual acceleration recursive sequence, the reference acceleration sequence, the expected acceleration recursive sequence, the vehicle speed error integral sequence, the acceleration error integral sequence, and the expected acceleration increment sequence; the preset constraints include expected acceleration constraints, expected acceleration constraints, and actual acceleration constraints. The desired acceleration determination module is used to determine the desired acceleration at the current moment as the sum of the first term in the desired acceleration increment sequence and the desired acceleration of the vehicle at the previous moment; the desired acceleration increment correlation includes: in, This is a set of vectors constructed from the recursive sequences of actual vehicle speed, actual acceleration, expected acceleration, vehicle speed error integral, and acceleration error integral. , The first time after the current moment For each predicted timestamp, the recursive values ​​of actual vehicle speed, actual acceleration, and expected acceleration, the vehicle speed error integral, and the acceleration error integral, 1 ≤ i ≤ , To predict the step size, , , The vehicle's current speed. The vehicle's current acceleration. The expected acceleration of the previous moment in the current moment. Integral of the vehicle's current speed error Let C be the integral of the vehicle's current acceleration error, and C be the first coefficient matrix. , For the next prediction timestamp at the current moment, the actual vehicle speed, actual acceleration, expected acceleration at the previous moment, vehicle speed error integral, and acceleration error integral are given. , The expected acceleration increment at the current moment, For reference only. , This is the reference speed at the current moment. Let F be the reference acceleration at the current moment, and let F be the second coefficient matrix. This is the third coefficient matrix. A is the fourth coefficient matrix, A is the first kinematic matrix, and B is the second kinematic matrix. Let U be the third kinematic matrix, and U be the desired acceleration increment sequence. , To control the period, k is the preset gain. As a preset time constant, For the reference value sequence, .

8. An electronic device, characterized in that, The electronic device includes: One or more processors; Storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the longitudinal control method as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the longitudinal control method as described in any one of claims 1-6.

Images