Control method and system of two-wheeled electric vehicle

Abstract

The invention relates to the technical field of electric vehicle control, in particular to a control method and system for a two-wheeled electric vehicle, and the method comprises the steps: collecting vehicle operation data and driver operation data, obtaining point cloud data through a radar arranged on the two-wheeled electric vehicle, and obtaining image data through a camera arranged on the two-wheeled electric vehicle; fusing the data to estimate the driving danger; when it is determined that the driving danger exists, whether the driver makes danger response or not is judged based on the driver operation data; when the driver does not make the danger response, determining a target control quantity based on the fused data; and distributing the target control quantity to a power system or a braking system of the two-wheeled electric vehicle based on a preset distribution strategy so as to control the two-wheeled electric vehicle through the power system or the braking system. Therefore, the driving danger is sensed by actively collecting various data for safety control, and the safety of the two-wheeled electric vehicle can be greatly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electric vehicle control, in particular to a control method and system of a two-wheeled electric vehicle. BACKGROUND

[0002] Currently, the safety system of a two-wheeled electric motorcycle mainly relies on an electronic auxiliary system with relatively single function, such as an anti-lock braking system or a linked braking system. Such systems usually only optimize the operation of the driver, such as distributing brake pressure, but lack the ability of active perception and avoidance of external environmental risks of the vehicle, belonging to the passive safety category.

[0003] To improve active safety, some existing technical solutions attempt to introduce sensors such as millimeter wave radars or cameras for environmental perception. However, a single type of sensor has inherent technical limitations, for example, millimeter wave radars have limited accuracy in target classification and lateral position measurement, while cameras are easily affected by light, rain, snow and other harsh weather conditions, resulting in insufficient reliability and all-weather working ability of the perception system. More importantly, the perception system of many solutions lacks deep and efficient closed-loop linkage with the dynamic stability control system of the vehicle, and the risk information perceived is mostly only used as a warning to the driver, and cannot automatically trigger the vehicle actuator to intervene, and cannot realize active risk avoidance, so the safety of the two-wheeled electric vehicle has not been greatly improved. Therefore, how to actively improve the safety of the two-wheeled electric vehicle has become a technical problem to be solved. SUMMARY

[0004] Therefore, the purpose of the present application is to provide a control method and system of a two-wheeled electric vehicle to actively improve the safety of the two-wheeled electric vehicle and thus maximize the safety of the two-wheeled electric vehicle.

[0005] To achieve the above purpose, the present application adopts the following technical solutions: In a first aspect, the present application provides a control method of a two-wheeled electric vehicle, comprising: collecting basic data, the basic data including environmental perception data, vehicle operation data and driver operation data, the environmental perception data including point cloud data obtained through a radar arranged on the two-wheeled electric vehicle and image data obtained through a camera arranged on the two-wheeled electric vehicle; fusing the basic data and estimating driving risk based on the fused data; when it is determined that there is driving risk, judging whether the driver responds to the risk based on the driver operation data; if the driver does not respond to the risk, determining a target control amount based on the fused data; The target control quantity is allocated to a power system or a braking system of the two-wheeled electric vehicle based on a preset allocation strategy, so as to control the two-wheeled electric vehicle through the power system or the braking system.

[0006] Further, in some embodiments of the present application, the basic data is fused, and driving danger is estimated based on the fused data, including: A coordinate conversion relationship between the point cloud data and the image data is established, and data collection frequencies of the radar and the camera are unified, so that the data collected by the radar and the camera are spatiotemporally aligned; The image data is subjected to feature extraction through a neural network to obtain image features; The point cloud data is converted into bird's-eye view grid data, and motion trajectory features are extracted through convolution technology; The image features and the motion trajectory features are subjected to channel dimension splicing, and fused features are generated through convolution compression dimension; Driving danger is estimated based on the fused features and a preset fuzzy rule.

[0007] Further, in some embodiments of the present application, the determination of the target control quantity based on the fused data includes: determining the target control quantity based on the fused features and a reinforcement learning principle.

[0008] Further, in some embodiments of the present application, the basic data is fused, and driving danger is estimated based on the fused data, including: A coordinate conversion relationship between the point cloud data and the image data is established, and data collection frequencies of the radar and the camera are unified, so that the data collected by the radar and the camera are spatiotemporally aligned; A target state vector is predicted based on the vehicle operation data and the environment data through a preset vehicle dynamics model and a Kalman filtering principle; A collision time and a vehicle stability limit in a current state are estimated based on the environment data and the target state vector; Driving danger is estimated based on the collision time and the vehicle stability limit in the current state.

[0009] Further, in some embodiments of the present application, the vehicle dynamics model includes a multi-body dynamics model and a simplified linear model, the multi-body dynamics model includes a four-body rigid body model and a multi-body tire coupling model, and the simplified linear model includes a single-track model and a double-track model; The four-body rigid body model is a motion equation established based on the Jourdain principle by connecting a frame, a front fork, a rear wheel and a front wheel of the two-wheeled electric vehicle as rigid bodies through joint constraints; The multi-body tire coupling model is a model based on a Pacejka magic formula tire model and including a nonlinear relationship between a longitudinal force and a lateral force of the tire. The single-track model is a model simplifying the two-wheeled electric vehicle as a rigid body. The double-track model is a roll equation considering load transfer of the two wheels.

[0010] Further, in some embodiments of the present application, the determining the target control quantity based on the fused data comprises: determining the target control quantity based on the target state vector and a stability control target.

[0011] Further, in some embodiments of the present application, the elements of the stability control target include vehicle body driving stability and comfort and driving physical constraints.

[0012] Further, in some embodiments of the present application, the target control quantity includes a yaw moment.

[0013] Further, in some embodiments of the present application, the distributing the target control quantity to the power system or the braking system of the two-wheeled electric vehicle based on a preset distribution strategy to control the two-wheeled electric vehicle through the power system or the braking system comprises: when the power system can meet the demand of the target control quantity, reducing motor output torque or entering a kinetic energy recovery mode through the power system; when the power system cannot meet the demand of the target control quantity, providing a supplementary braking force through the braking system while controlling the power system to reduce the motor output torque or enter the kinetic energy recovery mode.

[0014] In a second aspect, the present application provides a control system of a two-wheeled electric vehicle, comprising: a collection module configured to collect basic data, the basic data including environment perception data, vehicle operation data and driver operation data, the environment perception data including point cloud data acquired through a radar arranged on the two-wheeled electric vehicle and image data acquired through a camera arranged on the two-wheeled electric vehicle; a fusion estimation module configured to fuse the basic data and estimate driving danger based on fused data; a judgment module configured to judge whether a driver responds to danger based on the driver operation data when it is determined that there is driving danger; a decision module configured to determine a target control quantity based on the fused data if the driver does not respond to danger; an execution module configured to distribute the target control quantity to a power system or a braking system of the two-wheeled electric vehicle based on a preset distribution strategy to control the two-wheeled electric vehicle through the power system or the braking system.

[0015] The present application relates to electric vehicle control technical field, specifically relates to a two-wheeled electric vehicle control method and system, the method comprises: collecting vehicle operation data and driver operation data, and obtaining point cloud data through the radar arranged on the two-wheeled electric vehicle, and obtaining image data through the camera arranged on the two-wheeled electric vehicle;The data is fused to estimate the driving risk;When it is determined that there is a driving risk, it is judged whether the driver responds to the risk based on the driver operation data;When the driver does not respond to the risk, the target control quantity is determined based on the fused data;And based on the preset allocation strategy, the target control quantity is distributed to the power system or the braking system of the two-wheeled electric vehicle, so as to control the two-wheeled electric vehicle through the power system or the braking system. In this way, the safety of the two-wheeled electric vehicle can be greatly improved by actively collecting various data to perceive the driving risk and control safety. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, below will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0017] Figure 1 is the flowchart of the control method of the two-wheeled electric vehicle provided by the embodiment of the present application.

[0018] Figure 2 is the principle diagram of the control method of the two-wheeled electric vehicle provided by the embodiment of the present application.

[0019] Figure 3 is the structure diagram of the control system of the two-wheeled electric vehicle provided by the embodiment of the present application. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical scheme and advantages of the present application more clear, the technical scheme of the present application will be described in detail below. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of the present application.

[0021] Figure 1 is the flowchart of the control method of the two-wheeled electric vehicle provided by the embodiment of the present application, please refer to Figure 1 , the embodiment can include the following steps: S101, collecting basic data.

[0022] Specifically, the basic data includes environment perception data, vehicle operation data and driver operation data, wherein the environment perception data includes point cloud data acquired by a radar arranged on the two-wheeled electric vehicle and image data acquired by a camera arranged on the two-wheeled electric vehicle.

[0023] S102, fusing the basic data and estimating driving danger based on the fused data.

[0024] Specifically, in the present application, the environment data in the basic data can be fused, for example, the data can be fused based on an observer of Kalman filtering principle to obtain fused data in the form of a target state vector; or feature extraction can be performed through a preset neural network or model and then the driving danger can be estimated based on the fused features.

[0025] S103, when it is determined that there is danger, judging whether the driver responds to the danger based on the driver operation data.

[0026] S104, if the driver does not respond to the danger, determining a target control quantity based on the fused data.

[0027] For example, the target control quantity can be a yaw moment, which can be converted into a driving force or a braking force of a wheel and distributed through, for example, a Direct Yaw-Moment Control (DYC) technology in actual application.

[0028] S105, distributing the target control quantity to a power system or a braking system of the two-wheeled electric vehicle based on a preset distribution strategy to control the two-wheeled electric vehicle through the power system or the braking system.

[0029] Specifically, in the present application, when the target control quantity is determined, the power system can be used to respond preferentially, and the braking system can be used to supplement when the power system cannot meet the demand of the target control quantity.

[0030] The control method of the two-wheeled electric vehicle provided in the present application can greatly improve the safety of the two-wheeled electric vehicle by actively collecting various data to perceive driving danger and perform safety control.

[0031] Further, in some embodiments of the present application, when the point cloud data acquired by the radar and the image data acquired by the camera are fused, the two kinds of data need to be spatially aligned first, including: establishing a coordinate conversion relationship of the point cloud data and the image data, and unifying the data acquisition frequency of the radar and the camera, so that the data acquired by the radar and the camera are spatially and temporally aligned.

[0032] Specifically, the conversion relationship between the radar coordinate system and the camera pixel coordinate system can be established through a chessboard calibration board or laser radar point cloud matching, or the conversion relationship between a unified coordinate system other than the two is determined, so as to convert the two into the same coordinate system. For example, the Zhang calibration method is used to solve the camera internal and external parameters, combined with the polar coordinate data of the millimeter wave radar, a unified Cartesian coordinate system is constructed, the point cloud data and the image data are converted into the Cartesian coordinate system, and the synchronous alignment in space is realized. For time synchronization alignment, the sampling frequencies of the radar and the camera can be aligned by using hardware triggering or software interpolation (such as cubic spline interpolation).

[0033] In addition, in some embodiments of the present application, the object contour and the category (such as vehicle, pedestrian) semantic information recognized by the image data of the camera can also be used to guide the clustering analysis of the radar raw point cloud in reverse, so as to effectively distinguish multiple targets in close proximity (such as distinguishing the front vehicle and the roadside fence), so as to improve the perception accuracy and robustness in complex scenes.

[0034] On this basis, the environmental data is fused. For example, in some embodiments, the image data can be feature-extracted by a neural network to obtain image features; the point cloud data is converted into bird's eye view grid data, and the motion trajectory features are extracted by convolution technology; the image features and the motion trajectory features are spliced in the channel dimension, and the fusion features are generated by convolution compression dimension; finally, based on the fusion features and the preset fuzzy rules, the driving danger is estimated.

[0035] Specifically, for feature extraction of image data collected by the camera, YOLOv5-CSPNet can be used, such as using CSPDarknet53 as the backbone network, and reducing the amount of calculation by cross-stage local connection while retaining the integrity of the gradient flow. In addition, in some embodiments of the present application, during the feature extraction process, a spatial attention mechanism can also be applied to focus on the region of interest, for example, to enhance the key attention of the lane line and the obstacle in the curve scene.

[0036] For feature extraction of point cloud data obtained by the radar, the point cloud data can be first converted into bird's eye view grid, at this time, each grid unit contains motion features (such as speed variance, point density); on this basis, convolution technology can be used to extract motion trajectory features, such as using ConvLSTM to capture the acceleration trend.

[0037] On this basis, the image features extracted from the image data of the camera and the motion trajectory features extracted from the point cloud data of the radar are spliced in the channel dimension, and the fusion features are generated by convolution compression dimension, that is, the fusion data obtained by the feature extraction fusion method. For example, the P3, P4, P5 corresponding feature maps of YOLOv5 are fused with the motion trajectory features of the radar in the same spatial scale.

[0038] It should be noted that in some embodiments of the present application, when the features are extracted by the model to realize the fusion of point cloud data and image data, the fusion weight can be dynamically adjusted according to the confidence of the camera and the radar, for example, in a low light environment, the radar weight is increased to 70%, and the camera weight is reduced to 30%; in clear road conditions, the weights of both are balanced. At the same time, the number of the above-mentioned cameras and radars can be set to multiple, and in actual application, the device failure can be specifically identified by methods such as Mahalanobis distance to trigger redundant device switching, such as enabling the surround view camera when the main camera fails.

[0039] On this basis, the driving danger can be directly estimated based on the fused features and the preset fuzzy rules.

[0040] Specifically, the fuzzy rule library can be designed in advance, such as defining fuzzy variables such as "road surface wet slip degree" and "driving danger level", and judging by fusing the features and the fuzzy rule library. At the same time, defuzzification processing can also be performed, such as using the barycentric method to make the final output result an accurate value, so as to determine whether there is driving danger, for example, when the output driving danger is greater than or less than 85, it is determined that there is danger, so as to continue subsequent analysis (i.e. judging whether the driver has made a dangerous response and determining the target control quantity and specific execution).

[0041] On this basis, the target control quantity can also be directly determined by the reinforcement learning principle based on the above-mentioned fused features. For example, the PPO algorithm can be used to train the policy network, the input is the above-mentioned fused features, and the output is the target control quantity. In actual application, the reward function can also be designed in combination with the vehicle running data, so as to improve the final control accuracy.

[0042] In some other embodiments of the present application, the fusion of the basic data can also be based on the Kalman filtering principle, including predicting the target state vector based on the vehicle running data and the environmental data, through the preset vehicle dynamics model and the Kalman filtering principle; and then estimating the collision time and determining the vehicle stability limit under the current state based on the environmental data and the target state vector; and estimating the driving danger based on the collision time and the vehicle stability limit under the current state.

[0043] Specifically, first, in the present application, the vehicle dynamics model specifically includes a multi-body dynamics model and a simplified linear model. The multi-body dynamics model includes a four-body rigid body model and a multi-body tire coupling model, and the simplified linear model includes a single-track model and a double-track model. Different models are used to analyze different scenes to cooperate with the observer based on the Kalman filtering principle to predict the target state vector, and then analyze based on the target state vector to actively intervene in time when there is driving danger, so as to ensure the driving safety of the two-wheeled electric vehicle.

[0044] In this application, the four-body rigid model is a model that takes the frame, front fork, rear wheel and front wheel of the two-wheeled electric vehicle as rigid bodies, and connects them through joint constraints to establish a motion equation based on the Jourdain principle, which is used to predict the risk of side slipping in a lane changing scenario. The target state vector involved in this model can include the center of mass velocity, yaw rate, roll angle, steering angle and tire slip angle, etc., and can also be used to determine the vehicle stability limit.

[0045] The multi-body tire coupling model is a model that includes the nonlinear relationship between the longitudinal force and the lateral force of the tire based on the Pacejka magic formula tire model, which is used to analyze the influence of tire adhesion on vehicle body stability, so as to accurately predict the change of vehicle body stability when the tire adhesion decreases (such as due to wet road surface). The model can be specifically expressed as:

[0046] wherein Fx is the tire longitudinal force; Fy is the tire lateral force; represents the tire longitudinal force or the tire lateral force; λ is the side slip rate; BCDE respectively represent the stiffness factor, the shape factor, the peak factor and the curvature factor, which are all parameters calibrated through bench test. The model can be used to predict the target state vector or the side slip rate λ, and can also be used to determine the vehicle stability limit.

[0047] The single-track model is a model that simplifies the two-wheeled electric vehicle to a rigid body by ignoring the difference between the two wheels, which is used for high-speed straight driving analysis (which can be regarded as a simplified version of the above four-body rigid model in this scenario, thereby improving the estimation efficiency). The specific dynamics equation is:

[0048] wherein, is the total mass of the vehicle; is the center of mass lateral velocity of the vehicle, i.e. the movement speed of the vehicle in the left-right direction; is the yaw rate, i.e. the rotational angular velocity of the vehicle around the vertical axis (i.e. z-axis); is the center of mass longitudinal velocity of the vehicle, i.e. the forward and backward speed of the vehicle; is the front wheel lateral force, i.e. the lateral interaction force between the front wheel and the road surface; is the rear wheel lateral force, i.e. the lateral interaction force between the rear wheel and the road surface; is the rotational inertia of the vehicle around the z-axis; is the yaw angular acceleration; is the distance from the center of mass to the front axle; is the distance from the center of mass to the rear axle.

[0049] The double-track model is a roll equation considering two-wheel load transfer, which is used to analyze the influence of vehicle body roll on tire force distribution when driving on a curve. The roll equation is specifically expressed as:

[0050] wherein, is the moment of inertia of the vehicle around the roll axis; is the vehicle body roll angle acceleration; is the gravitational acceleration; is the vertical distance from the center of mass of the vehicle to the roll axis; is the vehicle body roll angle; is the roll stiffness of the suspension; is the roll damping of the suspension; is the vehicle body roll angular velocity. The model can be used to estimate the target state vector such as the vehicle body roll angle acceleration in the formula, and can be used to determine the vehicle stability limit.

[0051] In practical applications, based on the above model, the collected vehicle operation data such as accelerometer and gyroscope data, wheel speed sensor linear velocity, etc. can be used as observation vectors, and the observer based on Kalman filtering principle can predict the target state vector such as the center of mass velocity, yaw rate, roll angle and rate, etc. (which can be flexibly selected according to actual needs).

[0052] In practical applications, adaptive unscented Kalman filtering principle can also be used to dynamically adjust the process noise covariance matrix Q, for example, the road roughness (such as the degree of bumping) identified by the camera in the environmental data is used to increase the diagonal elements of Q in real time to suppress high-frequency vibration interference. In addition, neural networks such as deep learning networks can be used to extract features such as lane curvature and obstacle position from environmental data such as camera image data as state update quantities for Kalman filtering to improve the prediction efficiency of the target state vector.

[0053] In this way, based on the environmental data and the target state vector and using the above model, the possible collision and collision time can be estimated, and the vehicle stability limit under the current state can be analyzed, and then the driving risk can be comprehensively estimated based on the collision time and the vehicle stability limit under the current state, and when the driving risk is estimated and the driver is determined not to respond to the danger, the target control quantity is determined.

[0054] At the same time, based on the target state vector and the stability control target, the target control quantity is determined. The elements of the stability control target include: vehicle body driving stability and comfort, driving physical constraints and roll stability.

[0055] For example, the vehicle body driving stability and comfort target can be expressed as:

[0056] where J is the representation of the objective function; k is the time step index; N is the prediction horizon; w1 is the yaw rate weight; w2 is the roll angle weight; w3 is the steering angle weight; w4 is the control increment weight; is the steering angle; represents the control increment (i.e. the difference between the target control quantity at the current time and the target control quantity at the previous time ). In practical applications, the minimum J can be taken as the goal to determine the specific target control quantity.

[0057] In some embodiments of the present application, the target control quantity can also be determined based on other objectives, for example, any one or more of the following: maximum tracking of the target yaw rate, minimization of the vehicle body roll angle, maximization of the driver's acceleration intention, and reduction of braking energy consumption.

[0058] In addition, in the present application, when determining the target control quantity, the driving physical constraints are also considered, so as to ensure that the actually determined target control quantity cannot exceed the actual physical limit. The driving physical constraints can specifically include, for example, tire force saturation constraints, maximum torque of the control motor, battery output power, brake pressure upper limit, etc. For example, the tire force saturation constraint is represented as:

[0059] by controlling between the maximum value of the tire longitudinal force or the tire lateral force and the minimum value of the tire longitudinal force or the tire lateral force , so as to prevent the braking force or the driving force from exceeding the adhesion limit, to avoid side slipping.

[0060] In this way, by predicting the future state of the vehicle (i.e. the target state vector) based on the model, advanced control can be achieved, rather than merely correcting after instability; at the same time, when generating the control parameters, multiple control objectives are considered simultaneously, achieving global optimization rather than local optimization; and the driving physical constraints in control can also be processed to keep the control instructions within a safe execution range. Thus, the vehicle can be ensured to travel safely while the user's riding experience can also be guaranteed.

[0061] In practical applications, the above process can be based on a heterogeneous computing based on an edge computing architecture, such as using an NVIDIA Jetson AGX Orin, deploying the related calculations of the above dynamic model on a central processing unit (CPU), deploying the related calculations of the neural network on a graphics processing unit (GPU), and deploying the related calculations of the Kalman filter on a digital signal processor (DSP).

[0062] And the priority of the above various task operations can be set using a real-time scheduler, such as setting the priority of data acquisition to 100, setting the priority of inference calculation to 90, and setting the priority of the generation of actuator instructions, i.e., the distribution of target control quantities, to 80, so as to perform thread processing to ensure that the control cycle is stable within a preset time range.

[0063] Further, in some embodiments of the present application, based on a preset distribution strategy, the target control quantity is distributed to the power system or the braking system of the two-wheeled electric vehicle to control the two-wheeled electric vehicle through the power system or the braking system, including: when the power system can meet the demand of the target control quantity, reducing the motor output torque or entering the kinetic energy recovery mode through the power system; when the power system cannot meet the demand of the target control quantity, while controlling the power system to reduce the motor output torque or enter the kinetic energy recovery mode, providing supplemental braking force through the braking system.

[0064] Specifically, in practical applications, when the target control quantity is obtained for actual execution, power priority adjustment and hydraulic braking assistance can be based on, such as first reducing the motor output torque or entering the kinetic energy recovery mode through the MCU to generate a gentle deceleration; when the required deceleration exceeds the electric braking capability, the required hydraulic braking force is accurately calculated through the ECU, and the ESC system is instructed to intervene to provide supplemental braking force under the premise of ensuring anti-locking.

[0065] Among them, in the braking process, if the vehicle body posture is monitored to be abnormal (such as the rear wheel off the ground or the tendency of side slipping, etc.), the front and rear wheel braking force distribution ratio can be dynamically adjusted through an electronic control unit (ECU), so as to generate a stabilizing torque through slight braking of a single wheel to correct the vehicle body posture and ensure the safety of control execution.

[0066] In some other embodiments of the present application, dynamic distribution can also be performed, for example, the distribution mode of the target control quantity can be dynamically adjusted based on real-time working conditions such as battery power, motor temperature, current vehicle speed, etc.

[0067] In addition, in some embodiments of the present application, some basic data can be fused for specific tasks in specific scenarios and then corresponding individual control can be performed.

[0068] For example, when it is confirmed that target vehicle tracking is needed (such as by receiving a user instruction or by automatically identifying the need for target vehicle tracking through environmental data and vehicle operation data), an observer based on Kalman filtering principle is used to predict the target state vector at a future time based on the vehicle operation data or environmental data at the current time, so as to perform corresponding control. For example, the speed, acceleration, and position at the next time are predicted based on the angular velocity and acceleration at the current time. At the same time, in this process, the Kalman gain can be adjusted based on the distance observation value of the radar and the visual frame center coordinates of the camera, and the process noise covariance can be adjusted, so as to improve the prediction accuracy and ensure the effect of target vehicle tracking.

[0069] In addition, in some embodiments of the present application, individual and rapid control can also be directly performed based on environmental data. For example, when a vehicle that is rapidly approaching the driver's blind area is detected based on environmental data, if it is detected that the driver has a steering intention to the same side as the rapidly approaching vehicle (such as by detecting through a handle sensor), a warning can be directly and rapidly issued, and a small pulse braking force can be applied to the wheels through the ESC system of the two-wheeled electric vehicle, so as to generate a slight yaw moment through the operation, prompt or assist the driver to pull the vehicle back to the original lane, thereby realizing the function upgrade from early warning to auxiliary collision avoidance.

[0070] Based on the same inventive concept, the present application also provides a control system of a two-wheeled electric vehicle, Figure 3 is a structural schematic diagram of the control system of the two-wheeled electric vehicle provided by the embodiments of the present application, as Figure 3 shown, the system comprises: a collection module 11 for collecting basic data, the basic data including environmental perception data, vehicle operation data, and driver operation data, the environmental perception data including point cloud data obtained through a radar arranged on the two-wheeled electric vehicle, and image data obtained through a camera arranged on the two-wheeled electric vehicle; a fusion and estimation module 12 for fusing the basic data and estimating driving danger based on the fused data; a judgment module 13 for judging whether the driver responds to the danger based on the driver operation data when it is determined that there is driving danger; a decision module 14 for determining a target control amount based on the fused data if the driver does not respond to the danger; The execution module 15 is configured to distribute the target control variable to the power system or the braking system of the two-wheeled electric vehicle based on the preset distribution strategy, so as to control the two-wheeled electric vehicle through the power system or the braking system.

[0071] As to the system in the above-mentioned embodiments, the specific manner in which the various modules perform operations has been described in detail in the embodiments related to the method, and will not be described in detail here.

[0072] It can be understood that the same or similar parts in the above-mentioned embodiments can be mutually referred to, and the content not described in detail in some embodiments can be referred to the same or similar content in other embodiments.

[0073] It should be noted that, in the description of the present application, the terms "first", "second", etc. are only for the purpose of description, and cannot be understood as indicating or implying relative importance. In addition, in the description of the present application, unless otherwise specified, the meaning of "a plurality of" is at least two.

[0074] Any process or method descriptions in flow charts or otherwise described herein can be understood as representing code modules, segments, or portions of code which include one or more executable instructions for performing specific logic functions or steps in the process, and that the various embodiments of the application include the use of alternative or additional code to implement the described processes, methods, or functions, and that the alternative or additional code can be implemented using a computer program product, such as a computer program tangibly embodied in a machine-readable storage medium for execution by, or to control the operation of, a data processing apparatus, and that the program can be implemented in a plurality of programming languages, including an object-oriented programming language such as Java, Smalltalk, C++, or the like, and procedural programming languages such as C, or the like, and that the scope of the preferred embodiments of the application encompasses the use of programming languages over many of which the terms "module", "segment", or "portion" can be broadened in scope.

[0075] It should be understood that the various parts of the present application can be realized in hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be realized by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if realized in hardware, and as in another embodiment, it can be realized by any one or a combination of the following technologies known in the art: discrete logic circuit with logic gate circuit for implementing logic function on data signal, application specific integrated circuit with suitable combination logic gate circuit, programmable gate array (PGA), field programmable gate array (FPGA), etc.

[0076] Those skilled in the art of the present technology can understand that all or part of the steps carried out by the above-mentioned embodiment method can be completed by a program instructing the relevant hardware, and the program can be stored in a computer readable storage medium, and when executed, includes one or a combination of the steps of the method embodiment.

[0077] In addition, each function unit in each embodiment of the present application can be integrated in one processing module, or each unit can be physically present separately, or two or more units can be integrated in one module. The integrated module can be realized in the form of hardware, or in the form of a software function module. When the integrated module is realized in the form of a software function module and sold or used as an independent product, it can also be stored in a computer readable storage medium.

[0078] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0079] In the description of the present specification, the description referring to the terms "one embodiment", "some embodiments", "an example", "a specific example" or "some examples" etc. means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily mean the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0080] Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A control method of a two-wheeled electric vehicle, characterized by, The method comprises the following steps: collecting basic data, the basic data comprising environment perception data, vehicle operation data and driver operation data, the environment perception data comprising point cloud data acquired by a radar arranged on the two-wheeled electric vehicle and image data acquired by a camera arranged on the two-wheeled electric vehicle; fusing the basic data and estimating driving risk based on the fused data; when it is determined that there is driving risk, judging whether the driver responds to the risk based on the driver operation data; if the driver does not respond to the risk, determining a target control quantity based on the fused data; distributing the target control quantity to the power system or the braking system of the two-wheeled electric vehicle based on a preset distribution strategy to control the two-wheeled electric vehicle through the power system or the braking system.

2. The control method of a two-wheeled motor vehicle according to claim 1, characterized by, The method of fusing the basic data and estimating driving risk based on the fused data comprises the following steps: establishing a coordinate conversion relationship between the point cloud data and the image data and unifying the data acquisition frequencies of the radar and the camera to make the data acquired by the radar and the camera time-space aligned; extracting image features from the image data through a neural network; converting the point cloud data into bird's-eye view grid data and extracting motion trajectory features through convolution technology; concatenating the image features and the motion trajectory features in channel dimension and generating fused features through convolution compression dimension; estimating driving risk based on the fused features and a preset fuzzy rule.

3. The control method of a two-wheeled motor vehicle according to claim 2, characterized by, The method of determining the target control quantity based on the fused data comprises the following steps:

4. The control method of a two-wheeled motor vehicle according to claim 1, characterized by, determining the target control quantity based on the fused features and a reinforcement learning principle. The method of fusing the basic data and estimating driving risk based on the fused data comprises the following steps: establishing a coordinate conversion relationship between the point cloud data and the image data and unifying the data acquisition frequencies of the radar and the camera to make the data acquired by the radar and the camera time-space aligned; predicting a target state vector based on the vehicle operation data and the environment data through a preset vehicle dynamics model and a Kalman filtering principle; estimating a collision time and determining a vehicle stability limit in a current state based on the environment data and the target state vector; 5. The control method of a two-wheeled motor vehicle according to claim 4, characterized by, estimating driving risk based on the collision time and the vehicle stability limit in the current state. The vehicle dynamics model comprises a multi-body dynamics model and a simplified linear model, the multi-body dynamics model comprising a four-body rigid body model and a multi-body tire coupling model, and the simplified linear model comprising a single-track model and a double-track model; the four-body rigid body model is a motion equation established based on the Jourdain principle by connecting the frame, the front fork, the rear wheel and the front wheel of the two-wheeled electric vehicle as rigid bodies through joint constraints; the multi-body tire coupling model is a model comprising a nonlinear relationship between tire longitudinal force and lateral force based on a Pacejka magic formula tire model; the single-track model is a model simplifying the two-wheeled electric vehicle into a rigid body; 6. The control method of a two-wheeled motor vehicle according to claim 5, characterized by, the double-track model is a roll equation considering two-wheeled load transfer.

7. The control method of a two-wheeled motor vehicle according to claim 6, characterized by, The method of determining the target control quantity based on the fused data comprises the following steps: determining the target control quantity based on the target state vector and a stability control target. The elements of the stability control target comprise vehicle body driving stability and comfort and driving physical constraints.

8. The control method of a two-wheeled motor vehicle according to claim 1, characterized by, The target control quantity comprises a yaw moment.

9. The control method of a two-wheeled motor vehicle according to claim 1, characterized by, The target control quantity is distributed to a power system or a braking system of the two-wheeled electric vehicle based on a preset distribution strategy, so as to control the two-wheeled electric vehicle through the power system or the braking system, comprising: When the power system can meet the demand of the target control quantity, the motor output torque is reduced or the kinetic energy recovery mode is switched on through the power system; When the power system cannot meet the demand of the target control quantity, the braking system provides supplementary braking force while the power system reduces the motor output torque or switches on the kinetic energy recovery mode.

10. A control system for a two-wheeled electric vehicle, characterized by Comprise: The acquisition module is used for acquiring basic data, the basic data comprising environmental perception data, vehicle operation data and driver operation data, the environmental perception data comprising point cloud data acquired through a radar arranged on the two-wheeled electric vehicle and image data acquired through a camera arranged on the two-wheeled electric vehicle; The fusion estimation module is used for fusing the basic data and estimating driving danger based on the fused data; The judgment module is used for judging whether the driver responds to danger based on the driver operation data when it is determined that there is driving danger; The decision module is used for determining a target control quantity based on the fused data if the driver does not respond to danger; The execution module is used for distributing the target control quantity to the power system or the braking system of the two-wheeled electric vehicle based on a preset distribution strategy, so as to control the two-wheeled electric vehicle through the power system or the braking system.

Images