Vehicle simulation method and system

Abstract

A method for simulating an electric vehicle (EV) includes creating a simulation model (402) that associates a plurality of target behaviors of the target vehicle with the EV; acquiring a plurality of vehicle parameters (412) of the EV to generate a set of EV control parameters (430); acquiring a plurality of configuration parameters (414) of the target vehicle based on the set of EV control parameters (430) and the plurality of configuration parameters (414) of the target vehicle; providing a set of simulated target vehicle controls using the simulation model (402); and outputting the set of simulated target vehicle controls to the EV such that the EV is controlled to achieve the plurality of target behaviors of the target vehicle based on the set of simulated target vehicle controls, wherein the simulation model (402) is a neural network trained to reflect the relationship between the set of EV control parameters (430) and the set of simulated target vehicle controls.

Description

[Technical Field]

[0001] The present disclosure relates to the technical field of electric vehicles (EVs), and more particularly to a vehicle simulation method and system for simulating target behavior of a target vehicle in an electric vehicle. [Background technology]

[0002] Electric vehicles behave differently compared to internal combustion engine (ICE) vehicles. For example, unlike ICE vehicles, EVs do not need to maintain engine speed within a limited range. As a result, EVs do not have the sawtooth torque-speed curve of ICE vehicles. In another example, while EV drivers can enjoy the quiet and fast acceleration of an electric motor, they may miss the enjoyment of hearing the gas engine rumble and feeling the vibrations as it revs. Therefore, drivers of EVs and ICE vehicles have different driving experiences, and these differences may deter drivers of more traditional vehicles from purchasing EVs. Summary of the Invention [Problem to be solved by the invention]

[0003] The disclosed method and system is directed to overcoming one or more of the problems set forth above, as well as other problems. [Means for solving the problem]

[0004] One aspect of the present disclosure provides a method for simulating an electric vehicle (EV), which may include: creating a simulation model that associates a plurality of target behaviors of the subject vehicle with the EV; acquiring a plurality of vehicle parameters of the EV to generate an additional set of EV control parameters; acquiring a plurality of configuration parameters of the subject vehicle; providing a set of simulated subject vehicle controls using the simulation model based on the set of EV control parameters and the plurality of configuration parameters of the subject vehicle, where the simulation model is a neural network trained to reflect relationships between the set of EV control parameters and the set of simulated subject vehicle controls; and outputting the set of simulated subject vehicle controls to the EV such that the EV is controlled to achieve the plurality of target behaviors of the subject vehicle based on the set of simulated subject vehicle controls.

[0005] Another aspect of the present disclosure provides a simulation system for an electric vehicle (EV). The simulation system may include a plurality of input devices that provide a plurality of vehicle parameters, a memory containing program instructions, and a processor coupled to the memory and the plurality of input devices. When executing the program instructions, the processor is configured to: create a simulation model that associates a plurality of target behaviors of the subject vehicle with the EV; acquire a plurality of vehicle parameters of the EV to generate an additional set of EV control parameters; acquire a plurality of configuration parameters of the subject vehicle; provide a set of simulated subject vehicle controls using the simulation model based on the set of EV control parameters and the plurality of configuration parameters of the subject vehicle; the simulation model is a neural network trained to reflect a relationship between the set of EV control parameters and the set of simulated subject vehicle controls; and output the set of simulated subject vehicle controls to the EV such that the EV is controlled to achieve the plurality of target behaviors of the subject vehicle based on the set of simulated subject vehicle controls.

[0006] Another aspect of the present disclosure provides an electric vehicle (EV), which may include a battery pack for providing power to the EV, a set of wheels, at least one electric motor coupled to the battery pack for providing drive force to the set of wheels to drive the EV, and an onboard computer system, the onboard computer system being configured to perform the following processes: creating a simulation model associating a plurality of target behaviors of a subject vehicle with the EV, acquiring a plurality of vehicle parameters of the EV to generate an additional set of EV control parameters, acquiring a plurality of configuration parameters of the subject vehicle, providing a set of simulated subject vehicle controls using the simulation model based on the set of EV control parameters and the plurality of configuration parameters of the subject vehicle, where the simulation model is a neural network trained to reflect a relationship between the set of EV control parameters and the set of simulated subject vehicle controls, and outputting the set of simulated subject vehicle controls to the EV such that the EV is controlled to achieve the plurality of target behaviors of the subject vehicle based on the set of simulated subject vehicle controls.

[0007] Other aspects of the present disclosure will be apparent to those skilled in the art in light of the description, claims, and figures of the present disclosure. [Brief explanation of the drawings]

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the disclosed embodiments are briefly described below. Other drawings can be derived from such drawings by those skilled in the art without creative efforts and can be included in the present disclosure. [Figure 1] 1 illustrates an exemplary operating environment incorporating certain embodiments of the present disclosure. [Figure 2] 1 illustrates a block diagram of an exemplary electric vehicle (EV), according to an embodiment of the present disclosure. [Figure 3A] FIG. 1 illustrates a block diagram of an exemplary in-vehicle computer system according to an embodiment of the present disclosure. [Figure 3B] FIG. 1 illustrates a block diagram of an exemplary computer system according to an embodiment of the present disclosure. [Figure 4] FIG. 1 illustrates a block diagram of an exemplary simulation system for an EV, according to an embodiment of the present disclosure. [Figure 5A] FIG. 1 illustrates a block diagram of an exemplary simulator model, according to an embodiment of the present disclosure. [Figure 5B] FIG. 1 illustrates a block diagram of an exemplary target profile according to an embodiment of the present disclosure. [Figure 5C] FIG. 1 illustrates a block diagram of an example driver profile according to an embodiment of the present disclosure. [Figure 6] 1 illustrates a flowchart of an exemplary method for simulating an EV, according to an embodiment of the present disclosure. [Figure 7] 1 illustrates a flowchart of an exemplary method for training a simulation model, according to an embodiment of the present disclosure. [Figure 8] 10 illustrates a flowchart of another exemplary method for training a simulation model, according to an embodiment of the present disclosure. [Figure 9A] 1 illustrates the training of an exemplary simulator model according to an embodiment of the present disclosure. [Figure 9B] 10 illustrates a calculation of forward cycle consistency loss according to an embodiment of the present disclosure. [Figure 9C] 10 illustrates a calculation of reverse cycle integrity loss according to an embodiment of the present disclosure. [Figure 10] 1 illustrates a flowchart of an exemplary method for mining driving data for a training dataset, according to an embodiment of the present disclosure. [Figure 11] 1 illustrates an exemplary heat map according to an embodiment of the present disclosure. [Figure 12] 10 illustrates a flowchart for discovering new data regions according to an embodiment of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0009] The technical solutions of the present disclosure are described below with reference to the accompanying drawings. These embodiments are provided to enable a more complete understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art. It should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein.

[0010] In the embodiments of the present disclosure, a phrase such as "A and B are connected" may include a situation in which A and B are connected to each other and in contact with each other, or a situation in which A and B are connected via another component and are not in direct contact with each other. Furthermore, terms such as "first" and "second" are used to distinguish between similar objects and are not necessarily used to describe a specific order or sequence.

[0011] For most consumers, electric vehicles (EVs) are new and often offer EV drivers a new driving experience. Some drivers may find the EV driving experience interesting and exciting, while others may miss the driving experience of more traditional vehicles, particularly high-performance internal combustion engine (ICE) vehicles such as Mercedes-AMG series vehicles. Therefore, for EVs with sufficient capabilities, an on-board simulation method may be provided for the EV to simulate certain ICE vehicle characteristics to enhance the EV driving experience. That is, according to the present disclosure, an electric vehicle (EV) may be equipped with an on-board simulation system for simulating certain characteristics of the target vehicle. FIG. 1 illustrates an exemplary operating environment incorporating certain embodiments of the present disclosure.

[0012] As shown in FIG. 1 , operating environment 100 may include EV 110, driver / user 120, user terminal 130, cloud server 140, and communication network 150. Any number of EVs, users, user terminals, servers, and / or communication networks may be included, as well as other components. EV 110 may include any vehicle that operates on battery power, such as a pure electric vehicle or a hybrid electric vehicle, including an automobile, aircraft, or water vehicle. For example, EV 110 may include a battery pack for providing electrical power to EV 110, a set of wheels, at least one electric motor coupled to the battery pack for providing driving force to the set of wheels to drive EV 110, a wireless communication device for connecting to a cloud server and / or a mobile device carried by the driver of EV 110, and an on-board computer system for simulating a target vehicle on EV 110. The wireless communication device also provides wireless connectivity to the Internet for profile synchronization and error and log reporting. Driver 120 may be driving EV 110, which may be owned by driver 120 or by someone else and driven solely by driver 120. User terminal 130 may include any portable user device, such as a smartphone, a personal digital assistant (PDA), a notebook, a laptop, or a combination vehicle equipment and user device such as Apple CarPlay®. Any portable device may be included as user terminal 130. Furthermore, user terminal 130 may be carried or owned by driver 120, or may be located within or as part of EV 110.

[0013] Furthermore, communication network 150 may include any type of communication network, such as a wired and / or wireless network, for connecting EV 110 and / or user terminal 130 to the Internet or cloud server 140. Cloud server 140 may be provided by a commercial entity for managing, monitoring, maintaining, or servicing EV 110, such as a dealership or vehicle manufacturer, or a service provider of EV-related services. Cloud server 140 may store certain data required for the simulation and may perform computations offloaded from EV 110.

[0014] FIG. 2 illustrates a block diagram of an exemplary electric vehicle (EV) according to an embodiment of the present disclosure. As shown in FIG. 2, the electric vehicle 110 may include various subsystems or components. Specifically, the EV 110 may include a plurality of wheels 240, an electric motor 222, an accelerator pedal 208, a brake pedal 206, a differential 238, a battery pack 216, an onboard computer system 230, a motor controller 218, a steering wheel 228, a driver's seat 226, a dashboard 232, a sound system 234, and actuators 236. The EV 110 may also include a charger 202, a converter 214, a 12V battery 204, sensors 210, a wireless transceiver 212, an antenna 242, a transmission 224, an inverter 220, and the like. Any number of these subsystems or components may be included, certain components may be omitted, or other components may be included.

[0015] Onboard computer system 230 may control various components of EV 110. Charger 202 may charge battery pack 216 via converter 214 for converting alternating current (AC) or direct current (DC) input to a suitable charging source. Charger 202 may be the EV 110's onboard charger. The onboard charger may be a Level 1 charger that receives 120 VAC output from a wall outlet. EV 110 / battery pack 216 can also be charged by a Level 2 or Level 3 external charger that uses a high-voltage AC power source and can charge battery pack 216 more quickly than the onboard charger.

[0016] In some embodiments, the battery pack 216 may include a battery thermal management system for heating the battery pack 216 when the temperature of the battery pack 216 is below a predetermined low temperature threshold or for cooling the battery pack 216 when the temperature of the battery pack 216 is above a predetermined high temperature threshold. The battery pack 216 operates more efficiently when the temperature of the battery pack 216 is within a range between the predetermined low temperature threshold and the predetermined high temperature threshold.

[0017] The output of the battery pack 216 may be provided to a motor controller 218 to control an electric motor 222. The output of the battery pack 216 may pass through an inverter 220. The inverter 220 may adjust the voltage of the battery pack 216 to a voltage suitable for driving the electric motor 222. The output of the electric motor 222 may pass through a transmission 224 and a differential 238 to drive wheels 240.

[0018] Additionally, electric motor 222 may include a stator and a rotor (not shown in FIG. 2). The stator is the fixed outer shell of electric motor 222 attached to the chassis of EV 110. The rotor is the rotating element that supplies torque to transmission 224 of EV 110. Transmission 224 of EV 110 adjusts the rotor's rotational speed before using the rotor's torque to drive differential 238 of EV 110. Differential 238 of EV 110 distributes torque to wheels 240 according to a specific ratio appropriate for driving conditions.

[0019] The EV 110 may include two or more electric motors 222. For example, the EV 110 may include two electric motors 222. One electric motor 222 drives the two front wheels 240, and another electric motor 222 drives the two rear wheels 240. In another example, the EV 110 may include four electric motors 222. Each of the four electric motors 222 drives each of the four wheels 240. If each wheel 240 is directly driven by one electric motor 222, the differential 238 may be eliminated. Each wheel 240 may be equipped with a tire (not shown).

[0020] An accelerator pedal 208 and a brake pedal 206 may also be provided to accelerate and decelerate EV 110. Sensors 210 and actuators 236 may be provided to facilitate acceleration and deceleration. For example, sensors 210 may detect the positions of accelerator pedal 208 and brake pedal 206 and make those positions available to onboard computer system 230. Onboard computer system 230 controls electric motor 222 via motor controller 218 and inverter 220 in response to the positions of accelerator pedal 208 and brake pedal 206. Actuators 236 may be controlled by onboard computer system 230 to dynamically adjust the suspension of EV 110 based on the state of EV 110. In some embodiments, actuators 236 may be controlled by onboard computer system 230 to adjust the stiffness of accelerator pedal 208 and brake pedal 206.

[0021] EV 110 may also use sound system 234 and dashboard 232 to provide certain vehicle interactions to driver 120. For example, in addition to playing music and radio channels, sound system 234 may be controlled by on-board computer system 230 to simulate engine sounds. Dashboard 232 may be controlled by on-board computer system 230 to display certain information to driver 120, including an image of the dashboard of another vehicle.

[0022] The 12V battery 204 may be used to provide auxiliary power to various components of the EV 110, such as an onboard computer system 230, a dashboard 232, a sound system 234, sensors 210, actuators 236, a wireless transceiver 212, and other control circuits.

[0023] The wireless transceiver 212 may be connected to the antenna 242. The wireless transceiver 212 may facilitate communication between the onboard computer system 230, the cloud server 140, and the user terminal 130 shown in FIG. 1 . The wireless transceiver 212 may include a cellular communication transceiver supporting 3G / 4G / 5G cellular communication for communicating with the cloud server 140, and a Bluetooth (registered trademark; the same applies hereinafter) / Wi-Fi transceiver for communicating with the user terminal 130. Other wireless communication formats may also be used.

[0024] EV 110 may provide an individual (e.g., driver 120 or another passenger in the EV) with a steering wheel 228 and a driver's seat 226 for driving EV 110. For example, driver 120 of EV 110 sits in driver's seat 226 and uses steering wheel 228 to steer the driving direction of EV 110. In one embodiment, a speaker (not shown) may be positioned below driver's seat 226 to play simulated engine sounds to the individual or to play low-frequency sounds that simulate vibrations caused by a gas engine. EV 110 may also include a Global Positioning System (GPS) device (not shown) for detecting the current location of EV 110.

[0025] The current location may be used to determine local traffic / safety rules and regulations, which are used in performing safety checks on controlling the EV 110.

[0026] During operation, onboard computer system 230 may obtain multiple configuration parameters from one or more of the onboard computer system's memory storage device, a cloud server, and a mobile device carried by the driver of EV 110. Onboard computer system 230 may perform various control functions for EV 110 and may also execute a simulation process to simulate specific vehicle behaviors of another vehicle. When executing the simulation process, onboard computer system 230 may offload some or all of the simulation process to a cloud server to limit energy consumption in EV 110. FIG. 3A shows a block diagram of an exemplary onboard computer system according to an embodiment of the present disclosure. As shown in FIG. 3A, computer system 300 / onboard computer system 230 may include a processor 304, memory 302, a display screen 306, a microphone / speaker 308, an interface 310, sensors 312, actuators 314, a camera 316, and the like. Computer system 300 may be onboard computer system 230 shown in FIG. 2. Certain devices may be omitted, and other devices may be included.

[0027] The memory 302 may store program instructions that, when executed by the processor 304, perform an in-vehicle simulation method for the EV 110. In some embodiments, the memory 302 may include dynamic random access memory (DRAM), an embedded multimedia controller (e.MMC), low-power DRAM (LPDRAM), NOR flash memory, single-level cell (SLC) NAND flash memory, a solid-state drive (SSD), a universal flash storage (UFS) device, or a combination thereof.

[0028] In some embodiments, processor 304 may be one or more hardware processors, microprocessors, and microcontrollers distributed throughout various parts of the EV. For example, processor 304 may include a vehicle network processor dedicated to in-vehicle networking, a vision processor dedicated to vision processing, a radar processor dedicated to radar processing, a processor for engine control, a graphics processing unit (GPU) for dashboard rendering, an audio digital signal processor (DSP) for audio processing, a communications processor to support wireless communications such as 5G mobile, an artificial intelligence processor to implement neural networks, or a combination thereof. In some embodiments, certain computing tasks described above may be offloaded to a cloud server to preserve computing power available for EV operation. Accordingly, the cloud server returns results to the EV upon completion of the computing tasks offloaded from EV 110.

[0029] The computer system 300 may also include a display screen 306 and a microphone / speaker 308 for interacting with a user 120 of the computer system 300. The display screen 306 is part of a human-machine interface (HMI) for facilitating interaction between the driver 120 and the computer system 300. The HMI may include an infotainment system and an instrument cluster. The display screen 306 may include any suitable type of computer display device or electronic device display. For example, the display screen 306 may include an LCD display device, an OLED display device, or a combination thereof. The display screen 306 may be a touch-controlled display screen. The display screen 306 may include gesture sensors for hand gestures of an individual (e.g., the driver 120, a passenger) in front of the display screen 306. The display screen 306 may also include haptic drivers for providing haptic feedback. The display screen 306 may also include a head-up display (HUD) that presents information at eye level to the driver 120. The display screen 306 may include a transparent window display that uses a projector mounted inside the EV 110 to project an image onto a transparent film sandwiched or laminated in the window of the EV 110. The microphone / speaker 308 allows an individual to interact with the computer system 300 using voice commands. The microphone / speaker 308 may include active noise cancellation capabilities. The computer system 300 may also include other peripherals for interacting with an individual.

[0030] The computer system 300 can connect various accessories, such as sensors 312, actuators 314, and cameras 316, using an interface 310. The interface 310 may include a vehicle interface processor (VIP). The sensors 312 and actuators 314 may be the sensors 210 and actuators 236 shown in FIG. 2. The computer system 300 may include a memory 302, a processor 304, a display screen 306, a microphone / speaker 308, and an internal bus for connecting the interface 310 together. The interface 310 may be connected to the sensors 312, actuators 314, and cameras 316 via the internal bus. For example, the internal bus may be a controller area network (CAN) bus, a FlexRay bus, a media-oriented systems transport (MOST) bus, an automotive Ethernet bus, a local interconnect network (LIN) bus, or a combination thereof. The internal bus may also be used to connect other accessories not shown in FIG. 3A.

[0031] 1, in operation, user terminal 130 and / or cloud server 140 may interact with EV 110 or on-board computer system 230 via communication network 150 to execute certain user or server processes. Figure 3B shows a block diagram of an exemplary computer system according to an embodiment of the present disclosure. Computer system 350 may be cloud server 140 shown in FIG. 1.

[0032] As shown in FIG. 3B , computer system 350 may include memory 352, processor 354, communication interface 358, input / output devices 360, and data storage device 362. Other devices may also be included. Processor 354 may include any suitable hardware processor(s). Additionally, processor 354 may include multiple cores for multithreading or parallel processing and may include graphics capabilities for processing a human-machine interface (HMI) (i.e., an example of input / output device 360). Memory 352 may include any suitable memory module, such as ROM, RAM, flash memory module, erasable and rewritable memory, as well as mass storage devices, such as CD-ROM, DVD, U-disk, and hard disk. Memory 352 may store computer program instructions or program modules that, when executed by processor 354, perform various processes to interact with onboard computer system 230 on EV 110.

[0033] Furthermore, the computer system 350 may further include a display. The display may be any suitable display technology suitable for displaying images or videos. For example, the display may include a liquid crystal display (LCD) screen, an organic light-emitting diode (OLED) screen, or the like, or may be a touchscreen. The communication interface 358 may include a specific network interface device for establishing a connection over a communication network. The input / output device 360 ​​may include any suitable input device for inputting information to the processor 354 and / or output device for outputting information from the processor 354, such as a keypad, keyboard, and mouse device, a camera, a microphone, and other sensors. Furthermore, the data storage device 362 may include one or more data storage devices for storing specific data and performing specific operations on the stored data, such as database lookups and model training. Because the memory size of the onboard computer system 230 is limited, data required for the simulation may be stored in the data storage device 362. When necessary, data for the simulation may be downloaded from the data storage device 362 by the onboard computer system 230. Additionally, local regulations and / or rules may prevent data from being transmitted (exited) from the EV 110 without specific restrictions. Data may be tokenized, encrypted, and purged before being transmitted to the cloud server 140. Personally identifiable information must be protected whether it is stored in the onboard computer system 230 or transmitted to the cloud server 140. In either case, handling of the data must be consistent with the driver's consent or a service contract signed by the driver. In some embodiments, local regulations and / or rules may limit the data collected by and transmitted from the EV. Data collected by and egressing from the EV may be tokenized, encrypted, and purged to ensure that personally identifiable information (PII) data is handled appropriately.Data collected by and transmitted from the EV is processed, used, and stored in accordance with the driver's consent and preferences.

[0034] Returning to FIG. 1 , in operation, EV 110 may be driven by driver 120. After driver 120 enters EV 110, driver 120 may interact with onboard computer system 230 of EV 110 to provide input to onboard computer system 230, which may then execute a simulation process to simulate specific vehicle behaviors, actions, and / or characteristics of a target vehicle on EV 110. The input may include user information and / or configuration information. As used herein, the term “simulating” may refer to the process of obtaining static and / or dynamic parameters of EV 110 to realize specific vehicle behaviors, actions, and / or characteristics of the target vehicle on EV 110, and safely controlling EV 110 to realize the specific vehicle behaviors, actions, and / or characteristics of the target vehicle.

[0035] For example, the target behavior of the target vehicle may include vehicle handling characteristics, gear shifting, dynamic engine sounds and vibrations to the driver's seat, a simulated dashboard of the target vehicle, and combinations thereof. Vehicle handling characteristics reflect how the vehicle responds and reacts to vehicle driver inputs and may include at least the vehicle's weight distribution and the vehicle's tire cornering stiffness. The vehicle's weight distribution may include the center of mass height, center of mass, roll angle inertia, and yaw and pitch angle inertia. Other factors that contribute to the vehicle's handling characteristics include the stiffness of the vehicle frame, electronic stability control, steering precision, power delivery to the wheels, braking effectiveness, vehicle body aerodynamics, and vehicle suspension spring rate.

[0036] In an internal combustion engine (ICE) vehicle, gasoline is burned to create mechanical motion to propel the ICE vehicle, and gear shifting is used to adapt the vehicle transmission to various vehicle speeds. Gear shifting in an ICE vehicle causes abrupt changes in torque. Unlike ICE vehicles, an EV uses one or more electric motors to drive the wheels of the EV. The one or more electric motors can be electronically controlled to drive the wheels of the EV at various vehicle speeds without the need for gear shifting. Reproducing gear shifts that cause abrupt changes in torque is one aspect of simulating the behavior of a target vehicle.

[0037] Furthermore, the gasoline engine of an ICE vehicle generates a lot of noise and causes the ICE vehicle to vibrate when in operation, whereas the electric motor(s) of an EV do not generate a lot of noise and cause the EV to vibrate. To simulate the behavior of the target vehicle, the EV may need to replicate dynamic engine sounds and vibrations, which may be limited to the space surrounding the driver's seat.

[0038] The target vehicle does not necessarily have to be an ICE vehicle; a different EV or other type of vehicle can also be the target vehicle simulated on the EV 110. In another aspect of the simulation, different vehicles often include different dashboards. Simulating the behavior of the target vehicle may include recreating the target vehicle's dashboard. If the EV can include one or more display screens in the dashboard position, the EV controls the one or more display screens to render the target vehicle's dashboard. For example, if the target vehicle is another EV, a simulated dashboard of the other EV can be displayed on the EV so that the driver 120 can have the sensation of the other EV when looking at the dashboard. In some embodiments, important information specific to the EV (i.e., remaining battery charge) needs to be shown. When the simulation is running, this important information can be shown in the EV's original format or in a different format. Additionally, dashboard customization can be facilitated to allow the driver to view specific information (i.e., navigation information) that may not have been on the target vehicle's display.

[0039] Various other behaviors, actions, and / or characteristics of the subject vehicle may be simulated on EV 110. The simulation process may be implemented as simulation software running on onboard computer system 230, or as a combination of software and hardware executed by onboard computer system 230 or by both onboard computer system 230 and cloud server 140, or as a combination of software and hardware executed by onboard computer system 230 and / or cloud server 140. That is, onboard computer system 230 and / or cloud server 140 may implement a simulation system on EV 110 to execute the simulation process based on a request from driver 120. FIG. 4 shows a block diagram of an exemplary simulation system for an EV according to an embodiment of the present disclosure.

[0040] 4, the simulation system 400 may include a simulator 402, multiple input modules 410, multiple output modules 420, and multiple action modules 430. Other modules may also be included. The simulator 402 may include any suitable mathematical model or algorithm for generating specific simulation output parameters based on input parameters. FIG. 5A illustrates a block diagram of an example simulator model according to an embodiment of the present disclosure.

[0041] As shown in FIG. 5A , the simulator 402 may include a neural network model 500. In this specification, the neural network model 500 is also referred to as a vehicle simulator model. The vehicle simulator model (EV simulator) is trained to reflect the relationship between multiple vehicle parameters and multiple configuration parameters as inputs and one or more control parameters as outputs. Other types of artificial intelligence / machine learning models may also be used. The neural network model 500 may be a deep learning network model or a combination of multiple machine learning models and may include an input layer 504, intermediate layers 506 and 508 (hidden layers), and an output layer 510. In the case of a convolutional neural network, the hidden layer may include a convolutional layer. Furthermore, the neural network model 500 may also include a generative adversarial network for desired performance improvement. Furthermore, the input 502 may be provided to the input layer 504, and the output 512 may be provided by the output layer 510. Each layer may include one or more neural network nodes. The number of neural network layers is used for illustrative purposes, and any number of neural network layers may be used. The parameters of the neural network model 500 may be obtained by the on-board computer system 230 or stored / transferred from the cloud server 140.

[0042] The neural network model 500 may be initially trained by the cloud server 140, for example, to establish the simulator 402. For example, the cloud server 140 may obtain historical data of the target vehicle (e.g., a vehicle of the same type as the target vehicle) and the EV 110 as input 502 and output 512 values ​​to train the neural network model 500. The cloud server 140 may also obtain operational vehicle data and driver 120 data to train the neural network model 500. For example, the vehicle data may include usage statistics, such as the simulation duration, location, and startup frequency, as well as an error log, such as when the simulator 402 fails to initialize or is stopped by a safety module shown in FIG. 4 during operation. In some embodiments, the cloud server 140 may extract only a portion of the vehicle data that can be used to train the vehicle simulator model. In this way, the amount of data collected can be reduced, conserving energy in the EV and minimizing communication network usage. In some embodiments, local regulations and rules may limit what data is allowed to leave the EV, so data extraction by cloud server 140 may be restricted or may require prior consent from someone with authority over the EV (e.g., the owner).

[0043] For example, in some embodiments, input 502 may include a plurality of vehicle parameters of the EV, including runtime parameters of the EV. The runtime parameters of the EV may be parameters that describe or affect the runtime behavior of the vehicle, and the runtime behavior may refer to the conditions under which the vehicle is traveling. For example, the runtime parameters of the EV may include at least a steering wheel position, a vehicle weight distribution, a road condition, an impact distribution, a brake position, a torque vector, and battery availability, and output 512 may include one or more control parameters, including a target turn angle, a target steering ratio, a target weight distribution, and a target wheelbase. Input 502 may further include driving conditions of the EV. For example, the driving conditions of the EV may include at least a location, data from a rain / fog sensor, and a camera feed for understanding the terrain. Input 502 may further include local regulations and rules.

[0044] In another embodiment, input 502 may include multiple vehicle parameters of the EV, including accelerator pedal position, steering wheel position, road conditions, observed traction, available power, motor temperature, battery temperature, gyro measurements, accelerator measurements, suspension position, throttle position, brake position, and regeneration settings, and output 512 may include one or more control parameters, including target acceleration or deceleration, target drive wheels, target traction settings, and target handling characteristics.

[0045] In another embodiment, input 502 may include multiple vehicle parameters of the EV, including accelerator pedal position and window position, and output 512 may include one or more control parameters, including at least a target engine sound on the driver's side of EV 110. Any suitable input and output parameters may be used.

[0046] After the neural network model 500 has been trained, the neural network model 500 may be loaded into the on-board computer system 230 of the EV 110, for example, by retrieving it from the cloud server 140 via the communication network 150 or by storing the model data locally on the EV 110.

[0047] 4, multiple input modules 410 may include several modules configured to provide input parameters based on a particular type of input parameter. Specifically, multiple input modules 410 may include a vehicle input module 412, a configuration module 414, and a safety module 416. Other modules may also be included.

[0048] Vehicle input module 412 may provide input parameters related to the vehicle itself, i.e., EV 110 or the host vehicle. That is, vehicle input module 412 may provide values ​​for multiple vehicle parameters of EV 110. The multiple vehicle parameters may include torque parameters, weight distribution, steering parameters, acceleration and deceleration parameters, and suspension parameters.

[0049] Torque (pound-feet or newton-meters) is the amount of tractive force that the electric motor 222 generates while the driver 120 of the EV 110 is pressing the accelerator pedal 208. Horsepower refers to the power that the electric motor 222 generates. Weight distribution is the amount of total vehicle weight that is placed on the ground at an axle, axle group, or individual wheel. Weight distribution affects how quickly the vehicle accelerates and decelerates, as well as how well the vehicle handles when cornering, because the weight transfer that occurs when the vehicle is moving affects the level of tire grip.

[0050] Steering parameters include steering wheel position, brake pedal position, shock position, vehicle weight distribution, torque vector, road surface conditions, battery availability, or a combination thereof. Acceleration and deceleration parameters include steering position, road surface conditions, observed traction, available power, motor temperature, battery pack temperature, gyro measurements, accelerator measurements, suspension position, throttle position, brake pedal position, regeneration settings, or a combination thereof. Some parameters may appear in two or more of steering parameters, acceleration and deceleration parameters, and suspension parameters. Other parameters may also be included. Furthermore, vehicle input module 412 may provide vehicle parameters in real time during operation. Alternatively, vehicle input module 412 may provide stored vehicle parameters. Some vehicle parameters change dynamically and may be obtained from sensors on EV 110. Some other vehicle parameters are static and may be obtained from an EV data storage device on a cloud server. Thus, onboard computer system 230 may obtain multiple vehicle parameters of EV 110 locally and / or remotely from a cloud server.

[0051] Additionally, configuration module 414 may provide input parameters related to the configuration of the simulation, i.e., the behavior of the subject vehicle. That is, configuration module 414 may provide values ​​of multiple subject vehicle parameters and / or information about the driver, i.e., configuration parameters. The multiple configuration parameters include at least information about the subject vehicle and settings of the subject vehicle for the driver. For example, configuration module 414 may obtain a subject profile of the subject vehicle and / or a driver profile of the driver of EV 110 to determine the configuration parameters for the simulation. Other information may also be included.

[0052] A subject profile can include information for configuring a simulation. Figure 5B illustrates a block diagram of an example subject profile according to an embodiment of the present disclosure. As shown in Figure 5B, subject profile 550 can include vehicle manufacturer 552, vehicle model 554, vehicle behavior list 556, host vehicle requirements 558, simulator information 560, driver information 562, and other information 564. Some information items may be omitted and others may be added.

[0053] Vehicle make 552 may indicate the manufacturer (brand) of the target vehicle, and vehicle model 554 may indicate the model of the target vehicle. Vehicle behavior list 556 may indicate one or more vehicle behaviors or parameters to be simulated. For example, a target profile for a target vehicle may include a horsepower / torque curve, suspension rate / type / programming, steering ratio, wheelbase, vehicle type, speed-fuel consumption curve, dashboard data, steering control weight, accelerator pedal feel, brake pedal feel, and braking values, or a combination thereof.

[0054] For an ICE vehicle, torque is equal to horsepower multiplied by a constant (e.g., 5,252) divided by the rotational speed (revolutions per minute, or RPM). Gear shifting causes the torque versus time curve of an ICE vehicle to appear as a sawtooth curve. When the driver of the ICE vehicle presses the gas pedal of the ICE vehicle, torque increases over time and then drops sharply when the gear shift occurs. Unlike ICE vehicles, EVs do not require gear shifting, and the torque versus time curve initially rises and continues to rise as long as the EV driver 120 continues to press the accelerator pedal 208. EVs and ICE vehicles exhibit substantially different behaviors in response to a driver pressing the accelerator or gas pedal. The gear shift behavior of an ICE vehicle is simulated by the onboard computer system 230 of the EV 110 to give the driver 120 of the EV 110 the feel of an ICE vehicle's gear shift behavior.

[0055] The suspension may include an active suspension of the target vehicle, which controls the vertical motion of the target vehicle's wheels relative to the target vehicle chassis or body, or a passive suspension, where vertical motion is entirely determined by the road surface and is provided by large springs. An active suspension may vary the stiffness of shock absorbers to adapt to changing road or dynamic conditions, or may use actuators to raise and lower the chassis independently at each wheel. The suspension may be spring-based, and the suspension rate may be referred to as the spring rate. The spring rate is a factor in setting the vehicle's ride height. When a spring is compressed or extended, the force it exerts is proportional to the change in its length. The spring rate is the change in force exerted by the spring divided by the change in spring deflection. The spring rate may be programmed to adapt to the vehicle's weight. Therefore, the behavior of the target vehicle's suspension must be simulated on the EV110.

[0056] The steering ratio refers to the ratio between the rotation angle of the steering wheel and the rotation angle of the wheels. A high steering ratio means that the steering wheel must be rotated more to turn the wheels, but it is easier to turn the steering wheel. A low steering ratio means that the steering wheel must be rotated less to turn the wheels, but it is more difficult to turn the steering wheel. Therefore, the behavior of the steering wheel of the target vehicle needs to be simulated on the EV110.

[0057] Wheelbase is the horizontal distance between the centers of the front and rear wheels. When a vehicle accelerates, the suspension often causes the rear of the vehicle to sink and the front of the vehicle to lift. When a vehicle decelerates, the suspension often causes the rear of the vehicle to lift and the front of the vehicle to sink. The relative rise and fall of the front and rear of the vehicle and the wheelbase together affect the weight distribution of the vehicle and the feeling of driving the vehicle. Therefore, the wheelbase behavior of a target vehicle needs to be simulated on the EV110.

[0058] The vehicle type may include a sedan type, a grand touring (GT) type, and a sport utility vehicle (SUV) type. The vehicle type affects the feeling of driving the vehicle and serves to simulate the behavior of the target vehicle on the EV 110. The speed-fuel consumption curve is used to estimate the fuel consumption of the target vehicle. When the driver 120 drives the EV 110 operating in simulation mode, the estimated fuel consumption of the target vehicle may be displayed on the dashboard.

[0059] Dashboard data refers to data displayed on the dashboard of the target vehicle. While certain dashboard data for the target vehicle may no longer be applicable to the EV 110, it is still estimated and presented to the driver 120 on the dashboard of the EV 110, resulting in the driver 120 of the EV 110 feeling like they are driving the target vehicle. The steering control weights are used in four-wheel steering (FWD) to improve turning agility and stability at various vehicle speeds. When the driver 120 turns the steering wheel at low speeds, the front wheels rotate in the direction of travel and the rear wheels rotate in the opposite direction, effectively reducing the vehicle's turning radius and making low-speed maneuvers quicker and easier. High-speed steering rotates both the front and rear wheels in the same direction to enhance high-speed stability. This steering behavior of the target vehicle may be simulated on the EV 110. For certain ICE profiles, four-wheel steering behavior at high speeds may be disabled. For example, four-wheel steering may be disabled for safety reasons when the EV's speed exceeds a predetermined speed threshold.

[0060] The accelerator pedal feel and brake pedal feel vary from vehicle to vehicle. The driver 120 often memorizes the accelerator pedal feel and brake pedal feel of the target vehicle. The accelerator pedal feel and brake pedal feel of the target vehicle need to be simulated on the EV 110. Braking values ​​include braking distance or stopping distance. Braking distance is the distance a vehicle travels from fully depressing the brake pedal until the vehicle comes to a complete stop. Braking distance is primarily affected by the vehicle speed and the coefficient of friction between the tires and the road surface. The braking values ​​of the target vehicle need to be simulated on the EV 110. Furthermore, for safety reasons, the EV's maximum braking capability is always available in the event of an emergency. Emergencies may include, but are not limited to, a flat tire, a headlight failure, a stuck throttle / accelerator, an engine stall, an imminent collision, wildlife on the road, and running off the road.

[0061] 5B, host vehicle requirements 558 may indicate one or more requirements for a host vehicle for running a simulation, such as horsepower, powertrain configuration, etc. Simulator information 560 may indicate specific information about the simulator for simulating the subject vehicle, and driver information 562 may be used to identify a subject profile when driver identification information is used to search for and identify the subject profile. Other information 564 may be used for other user or vehicle-specific information.

[0062] Additionally, the driver profile may include information specifically about the driver 120 to facilitate the simulation process. Figure 5C illustrates a block diagram of an example driver profile according to an embodiment of the present disclosure. As shown in Figure 5C, the driver profile 580 may include driver identification information 582, driver personal information 584, driver vehicle information 586, driver account information 588, driver social media information 590, a target profile list 592, and other information 594. Some information items may be omitted and others may be added.

[0063] Driver identification information 582 may indicate the identity of driver 120, which may be used to search a database. Driver personal information 584 may include personal information about driver 120, such as weight, gender, age, address, location, and occupation. Driver vehicle information 586 may include the driver's vehicle-specific information, such as vehicle registration, vehicle garage information, and vehicle usage information. Driver account information 588 may include login information for accessing cloud server 140, and driver social media information 590 may include information about the driver's social network presence, such as the driver's social media access information for sharing recorded driving data. The recorded driving data may include simulation data, including information about a selected ICE profile. Simulation data within the recorded driving data shared by the driver on social media may be used as a training data set for training a simulation model. Subject profile list 592 may include one or more subject profiles that the driver may use or select to run a simulation, and each subject profile may be individually selected by driver 120 to simulate the behavior of the subject vehicle on EV 110. Other information 594 may include other application-specific information.

[0064] Returning to FIG. 4 , the input module 410 may also include a safety module 416. The safety module 416 may provide information for performing safety checks during the simulation process to ensure that the simulation is safe and complies with certain rules and regulations. For example, the safety module 416 may include range information for configuration parameters so that the range information can be used for safety checks on the configuration parameters to ensure that their values ​​are within safe ranges. The safety module 416 may also include range information for output parameters and / or action parameters so that the range information can be used for safety checks on the output parameters and / or action parameters to ensure that their values ​​are within safe ranges.

[0065] Additionally, the safety module 416 may include regulatory information, which may be used for safety checks on output and / or action parameters to ensure that their values ​​verify regulations, with or without location information. The safety module 416 may also include specific system patch or update information, which may be used to update or patch the simulator 402 and other modules. The safety module 416 may also include host vehicle requirements, such as engine fault codes, overdue maintenance, and low tire pressure. The safety module 416 may also include other requirements or information, such as road topography, local restrictions and / or regulations, weather conditions, etc.

[0066] 4, the plurality of output modules 420 may include a look and feel module 422, a target dynamics module 424, and a data module 426. Other modules may also be included. The look and feel module 422 may be static and / or receive simulator 402 output parameters related to the look and feel of the target vehicle, such as dashboard displays, light displays, and driver and steering position and attitude. That is, all data related to how the target vehicle should look and feel, i.e., how the vehicle should sound, how the vehicle should look, and what the displays should be like.

[0067] Target dynamics module 424 may receive simulator 402 output parameters that are dynamic and required to perform specific sequential actions on EV 110 to change the driving characteristics of EV 110, i.e., all data related to runtime vehicle behavior, such as acceleration or deceleration. When an action is required to achieve the target behavior, target dynamics module 424 may provide information to playback module 432 to cause a change in vehicle behavior to satisfy the target profile of driver 120.

[0068] Furthermore, the data module 426 may record simulation data from the simulator 402 and store the data or upload the data to the cloud server 140 for further training of a simulation model in the simulator 402. For example, the data module 426 may record driving data and share the driving data via social media or other networks under a simulated scenario. That is, the driver 120 may use the data module 426 to share driving data of the virtual vehicle (i.e., the simulated target vehicle) on social media. In some embodiments, a heat map may be introduced to analyze the driving data from the EV to identify / remove noise. After noise is identified / removed from the driving data from the EV, the driving data may be clustered into one or more regions. Patterns and anomalies may be discovered based on one or more regions of the driving data from the EV. For example, a vehicle simulation model may be created and trained to correspond to specific patterns and anomalies. Therefore, additional revenue may be generated by providing pattern-specific and / or anomaly-specific simulation services to customers.

[0069] 4, the plurality of action modules 430 may include a playback module 432, a vehicle control, display, and sound module 434, a vehicle powertrain module 436, and a vehicle data analysis module 438. Other modules may also be included. The playback module 432 may be provided to perform actions necessary to achieve a target behavior of the target vehicle. That is, the playback module 432 may receive information from the target dynamics module 424 to determine one or more actions necessary to achieve the target behavior and may further instruct each module to perform the associated action.

[0070] For example, vehicle control, display, and sound module 434 may perform actions that fall under the categories of control, display, and sound. Vehicle powertrain module 436 may perform actions related to the powertrain, and vehicle data analytics 438 may collect all action data and further anonymize the collected data so that the anonymized driving data can be uploaded to cloud server 140 for analysis and / or simulation model training. For example, collected data may include usage statistics such as simulation duration, location, and activation frequency, as well as error logs such as when simulator 402 fails to initialize or is stopped by safety module 416 during operation.

[0071] During operation, the simulation system 400 or on-board computer system 230 interacts with the driver 120 to execute various simulation processes provided by the simulation system 400. Figure 6 shows a flowchart of an exemplary simulation method for an EV according to an embodiment of the present disclosure. The EV may be the EV 110 shown in Figures 1 and 2. The processor 304 of the on-board computer system 230 may execute program instructions stored in the memory 302 to implement the simulation method.

[0072] In S602, a simulation model is created that associates a plurality of target vehicle behaviors with the EV.

[0073] Recreating or replaying the experience of driving a target vehicle for a driver of an EV can bring a sense of nostalgia to the driver. The EV can operate in a normal mode or a simulation mode. When the EV operates in the normal mode, the EV exhibits EV behavior. When the EV operates in the simulation mode, the EV exhibits multiple target behaviors of the target vehicle. In other words, to replay multiple target behaviors of the target vehicle on the EV, the driver of the EV needs to operate the EV in the simulation mode.

[0074] When the EV is operated in a normal mode by a driver, the driver of the EV controls the EV via a plurality of input devices of the EV to generate a plurality of vehicle parameters of the EV. An on-board computer system (EV simulator) of the EV acquires the plurality of vehicle parameters of the EV to generate a set of EV control parameters and controls the EV based on the set of EV control parameters.

[0075] When an EV is operated in a simulation mode by a driver, the EV driver controls the EV via multiple input devices of the EV to generate multiple vehicle parameters for the EV. An on-board computer system of the EV acquires the multiple vehicle parameters of the EV and generates a set of EV control parameters. However, the on-board computer system does not directly control the EV based on the set of EV control parameters. Instead, the on-board computer system uses a simulation model to convert the set of EV control parameters into a set of simulated target vehicle controls and controls the EV based on the set of simulated target vehicle controls. The simulated target vehicle controls may be output settings or other output parameters that control the operation of the EV. Examples of simulated target vehicle controls include one or more of a target turning angle, a target acceleration or deceleration, a target drive wheel, a target traction setting, a target handling characteristic, and a target engine sound.

[0076] Both the set of EV control parameters and the set of simulated target vehicle controls are used to control the EV. The set of EV control parameters are used to control the EV when the EV operates in a normal mode. The set of simulated target vehicle controls are used to control the EV when the EV operates in a simulation mode. When the EV operates in the simulation mode, the EV exhibits a target behavior of the target vehicle. The conversion of the set of EV control parameters to the set of simulated target vehicle controls via the simulation model facilitates operation of the EV in the simulation mode.

[0077] In some embodiments, the goal of converting a set of EV control parameters into a set of simulated target vehicle controls is to reproduce multiple target behaviors of the target vehicle on the EV. However, due to differences between the EV and the target vehicle, the accurate conversion of a set of EV control parameters into a set of simulated target vehicle controls is determined not by optimizing individual parameters but by the combined effect of all parameters. In other words, accurate conversion is not only an objective engineering issue, but also a subjective look and feel issue. Furthermore, accurate conversion focuses on the subjective look and feel perceived by the driver of the EV. For the purposes of this disclosure, the perceptions of other passengers in the EV are irrelevant.

[0078] In some embodiments, because an accurate transformation is related to the look and feel perceived by an EV driver, it is difficult to obtain a mathematical function for a simulation model that reflects the relationship between a set of EV control parameters and a set of simulated target vehicle controls. To obtain the simulation model, a machine learning approach is considered. In this case, the simulation model is created and trained on a high-performance computer system separate from the EV. After the simulation model is trained using a training dataset and achieves certain performance goals, it can be loaded into the EV's on-board computer system. The simulation model can transform the set of EV control parameters into a set of simulated target vehicle controls in real time, allowing the EV driver to feel as if they are driving the target vehicle.

[0079] In some embodiments, the simulation model may be loaded onto a cloud server, and the simulation process may also be executed on the cloud server. In this case, data required for the simulation process may be exchanged between the EV and the cloud server, thereby maintaining battery charge for driving the EV.

[0080] In some embodiments, the simulation model is a neural network, and the set of EV control parameters is input to the neural network to obtain a set of simulated target vehicle controls.

[0081] 6, in S604, a plurality of vehicle parameters of the EV are obtained to generate a set of EV control parameters, the plurality of vehicle parameters of the EV including at least a runtime parameter of the EV and a driving state of the EV.

[0082] To simulate multiple target behaviors of a target vehicle on an EV, the on-board computer system needs to acquire information about both the EV and the target vehicle. The multiple vehicle parameters of the EV are real-time data collected by various sensors and input devices of the EV. Based on the multiple vehicle parameters of the EV, the on-board computer system generates a set of EV control parameters. In a normal mode, the set of EV control parameters can be used to directly control the EV. In a simulation mode, the set of EV control parameters is input into a simulation model (target vehicle simulator) to obtain a set of simulated target vehicle controls. The set of simulated target vehicle controls is used to control the EV to simulate multiple target behaviors of the target vehicle.

[0083] For example, the vehicle parameters may include steering parameters, acceleration and deceleration parameters, and suspension parameters. The steering parameters may include steering wheel position, brake pedal position, shock position, vehicle weight distribution, torque vector, road condition, battery availability, or a combination thereof. The acceleration and deceleration parameters may include steering wheel position, road condition, observed traction, available power, motor temperature, battery pack temperature, gyro reading, accelerator reading, suspension position, throttle position, brake pedal position, regeneration setting, or a combination thereof. Some parameters may appear in more than one of steering parameters, acceleration and deceleration parameters, and suspension parameters. Onboard computer system 230 may obtain vehicle parameters of EV 110 in real time from various subsystems or components of EV 110.

[0084] Referring back to FIG. 6 , at S606, multiple configuration parameters of the target vehicle are obtained. The configuration parameters of the target vehicle may be settings or other parameters that affect how the vehicle operates. In one embodiment, the multiple configuration parameters include at least information about the target vehicle and settings of the target vehicle for the driver. For example, after the onboard computer system 230 receives a simulation request from the driver 120, the onboard computer system 230 may enter a simulation mode and initiate various simulation processes. The onboard computer system 230 may then obtain configuration parameters for the simulation process. Specifically, the onboard computer system 230 may obtain driver-related information, as well as target behavior and target vehicle information. Additionally or optionally, safety information related to the simulation process may be obtained locally or remotely from the cloud server 140 by the onboard computer system 230. For example, the safety information may include EV health information, local traffic / safety rules and regulations, and ambient information from a driver assistance system.

[0085] A driver profile may be obtained for driver-related information, and a target profile may be obtained for target behavior and target vehicle information. The target profile may indicate the target behavior to be simulated and / or the target vehicle to be simulated. The target profile may include various types of information, such as horsepower / torque curves, weight distribution, suspension rate / type / programming, steering ratio, wheelbase, vehicle type, fuel consumption curve, dashboard data, control weights for steering, accelerator pedal feel, and brake pedal feel, braking values, or combinations thereof. The target behavior is not initially configured on the EV 110.

[0086] In certain embodiments, the target profile may be included in a driver profile. Thus, on-board computer system 230 may obtain the target vehicle from the driver profile of driver 120. In certain circumstances, obtaining the target profile of the target vehicle may include using an internal camera to obtain the identity of the driver of the EV and retrieving the driver profile and / or target profile associated with the identity of the driver of the EV. In some embodiments, driver 120 may also input the target vehicle into on-board computer system 230 via an input interface of on-board computer system 230.

[0087] Optionally or additionally, after obtaining the configuration parameters, the onboard computer system 230 may determine, based on the safety information, whether it is safe to launch the simulation before launching the simulation. For example, the onboard computer system 230 may evaluate the health of the EV based on engine fault codes, overdue maintenance, and low tire pressure. After launching the simulation, the onboard computer system 230 may perform a safety check to determine whether it is safe to continue the simulation. Specifically, the onboard computer system 230 may perform a first safety check on the configuration parameters to determine whether the configuration parameters are within a safe range (first safety range) for the EV 110. If a particular configuration parameter is outside the safe range, the onboard computer system 230 may prompt an error message and exit the simulation mode. If the first safety check is passed, the onboard computer system 230 may enter the simulation mode and continue the simulation process.

[0088] In some embodiments, before the EV begins operating in simulation mode, onboard computer system 230 may obtain a request from the driver of the EV to simulate a target behavior of a target vehicle on the EV. For example, driver 120 may interact with onboard computer system 230 to input a simulation request to onboard computer system 230 and also provide a driver profile to onboard computer system 230. Interaction between driver 120 and onboard computer system 230 may occur in various ways. In one embodiment, driver 120 may manually input the request and / or driver profile via a human-machine interface (HMI) of onboard computer system 230. In some other embodiments, driver 120 may input the request and / or driver profile via another device.

[0089] For example, driver 120 may carry user terminal 130. Driver 120 may request a simulation and manage on-board computer system 230 and the simulation via user terminal 130. Driver 120 may configure a driver profile on user terminal 130 and load the driver profile into on-board computer system 230. The driver profile may include, among other driver-specific information items, information about target vehicles in a target profile list. Driver 120 may also manage the connection between EV 110 and cloud server 140. For example, driver 120 may use a mobile application on user terminal 130 to configure on-board computer system 230 of EV 110 to establish a connection to cloud server 140.

[0090] In some embodiments, after the driver 120 enters the EV 110, the onboard computer system 230 may recognize the driver 120 through a camera of the EV 110 and may make a simulation request via the camera. In some other embodiments, the onboard computer system 230 may recognize the driver 120 through a user terminal 130 wirelessly connected to the onboard computer system 230. For example, the wireless connection is a Bluetooth connection. After recognizing the driver 120, the onboard computer system 230 retrieves a driver profile associated with the driver 120. The onboard computer system 230 may obtain the driver profile from a memory device of the onboard computer system 230, from the wirelessly connected user terminal 130 of the driver 120, or from the cloud server 140.

[0091] Referring back to FIG. 6 , in S608, the simulation model (target vehicle simulator) is used to provide a set of simulated target vehicle controls based on the set of EV control parameters and the plurality of configuration parameters of the target vehicle. More specifically, the set of EV control parameters and the plurality of configuration parameters of the target vehicle are provided to simulator 402 as input parameters, and simulator 402 (target vehicle simulator) may generate a set of simulated target vehicle controls that reflect the target behavior of the target vehicle, i.e., are controlled to achieve the target behavior. In some embodiments, to conserve battery charge for EV 110, on-board computer system 230 may offload execution of the target vehicle simulation process to cloud server 140. After completing the target vehicle simulation process, cloud server 140 returns the plurality of output parameters to on-board computer system 230. In this case, data exchange between on-board computer system 230 and cloud server 140 via wireless communication may cause a slight delay.

[0092] In some embodiments, to accurately reflect the relationship between a set of EV control parameters and a set of simulated target vehicle controls, a simulation model needs to be trained using a training data set and tested using a test data set. The training data set and the test data set often include a sample set of EV control parameters and a corresponding set of simulated target vehicle controls. However, the training data set and the test data set are not easily obtained.

[0093] In the process of designing an EV or target vehicle, a designer often creates an engineering model of the EV or target vehicle. For example, an engineering model of the EV (EV simulator) can be used to generate multiple sets of EV control parameters, and an engineering model of the target vehicle can be used to generate multiple sets of target vehicle controls. In another example, a manufacturer of the EV can collect multiple sets of EV control parameters, and another manufacturer of the target vehicle can collect multiple sets of target vehicle controls. In another example, a driver of the EV can collect multiple sets of EV control parameters for the EV driven by the driver, and a driver of the target vehicle can collect multiple sets of target vehicle controls for the target vehicle driven by the driver.

[0094] In some embodiments, the target vehicle is a vehicle different from the EV, the EV is configured with an adjustable seat, and the behavior of the target vehicle includes at least a seat profile whereby the lumbar support reinforcement and seat depth of the adjustable seat of the EV are adjusted to simulate that of the seat of the target vehicle.

[0095] In some embodiments, the adjustable seat of the EV is controlled by the driver of the EV to gradually return to the original setting after a preconfigured time period has expired.

[0096] In some embodiments, the subject vehicle is an ICE vehicle, the EV is configured with a massage actuator in the driver's seat, and the subject vehicle behavior includes controlling the massage actuator in the driver's seat to vibrate to simulate vibrations of the engine of the ICE.

[0097] In some embodiments, the target vehicle is a vehicle different from the EV, the EV is equipped with an olfactory scent dispenser, and the behavior of the target vehicle includes a scent profile such that the olfactory scent dispenser is controlled to emit a fragrance to simulate a scent found within the target vehicle, the scent being a genuine leather or vegan leather scent.

[0098] FIG. 7 shows a flowchart of an exemplary method for training a simulation model according to an embodiment of the present disclosure. As shown in FIG. 7, in S702, multiple sets of EV control parameters are collected, and in S704, multiple sets of target vehicle controls are collected. Regardless of how the multiple sets of EV control parameters and the multiple sets of target vehicle controls are collected, the multiple sets of EV control parameters and the multiple sets of target vehicle controls can be used as training data sets and test data sets for training and testing the simulation model, respectively. As shown in FIG. 7, in S706, a neural network is trained using the multiple sets of EV control parameters and the multiple sets of target vehicle controls to obtain a simulation model corresponding to the target vehicle.

[0099] Because the training dataset does not include multiple sets of simulated target vehicle controls, the neural network of the simulation model must have a structure that can utilize a readily available training dataset. In some embodiments, the neural network is a Cycle Generative Adversarial Network (CycleGAN). The CycleGAN includes a first GAN and a second GAN. The first GAN includes a first generator model and a first classifier. The second GAN includes a second generator model and a second classifier.

[0100] CycleGAN technology was originally developed for unpaired image-to-image transformation. In the present disclosure, the sets of EV control parameters and the sets of simulated target vehicle controls are not image data. However, like the image data transformed by CycleGAN, the sets of EV control parameters and the sets of simulated target vehicle controls are unpaired. CycleGAN can be trained using unpaired training datasets from two different domains (e.g., the EV domain and the target vehicle domain). In an embodiment of the present disclosure, the unpaired training datasets are the sets of EV control parameters and the sets of target vehicle controls. Therefore, a simulation model having a CycleGAN structure can be trained to accurately transform a set of EV control parameters into a corresponding set of simulated target vehicle controls that provides an EV driver with the authentic look and feel of the target vehicle's multiple target behaviors.

[0101] The following describes how to use the training dataset to train CycleGAN. Figure 8 shows a flowchart of another exemplary method for training a simulation model according to an embodiment of the present disclosure. As shown in Figure 8, the method for training a simulation model includes the following processes:

[0102] In S802, multiple sets of EV control parameters are input to a first generator model to generate multiple sets of simulated target vehicle controls.

[0103] In this case, the set of EV control parameters are runtime EV control parameters resulting from the driver's operation of the EV, and the set of simulated target vehicle controls are runtime EV control parameters for simulating the target vehicle. The purpose of the simulation model is to convert the set of EV control parameters into a corresponding set of simulated target vehicle controls for the simulated target vehicle controls. If multiple sets of EV control parameters and multiple sets of simulated target vehicle controls corresponding to the multiple sets of EV control parameters were readily available, only the first generator model could be trained to accurately convert from the EV control parameter set to the corresponding simulated target vehicle control set. However, such a paired training dataset is not readily available. Instead, only the multiple sets of EV control parameters and the multiple sets of target vehicle controls can be collected as the training dataset. Here, the set of target vehicle controls are runtime target vehicle control parameters resulting from the driver's operation of the target vehicle.

[0104] The CycleGAN technique is utilized to overcome the problem of lack of paired training datasets.

[0105] In S804, the multiple sets of target vehicle controls and the multiple sets of simulated target vehicle controls are input to a first classifier to calculate the likelihood that the multiple sets of simulated target vehicle controls are the multiple sets of target vehicle controls, and update the first generator model in each training iteration.

[0106] The first classifier is used to make the sets of simulated target vehicle controls (EV control parameters) generated by the first generator model resemble the sets of target vehicle controls (target vehicle control parameters). However, because the training dataset, i.e., the sets of EV control parameters and the sets of target vehicle controls, are not paired, even if the sets of simulated target vehicle controls (EV control parameters) resemble the sets of target vehicle controls, the sets of simulated target vehicle controls are not accurate transformations of the sets of EV control parameters. For example, the sets of simulated target vehicle controls can generally simulate the gear shift of the target vehicle, but may not accurately reflect the gear shift of the target vehicle specific to the accelerator pedal position. The accelerator pedal position is indirectly captured in the set of EV control parameters. This issue is addressed by the cycle consistency loss function described below.

[0107] At S806, the plurality of sets of target vehicle controls are input to a second generator model to generate a plurality of sets of simulated EV control parameters.

[0108] To form a CycleGAN structure, the second GAN performs the inverse transformation of the first GAN. In this case, the set of target vehicle controls are runtime target vehicle control parameters resulting from the driver's operation of the target vehicle, and the set of simulated EV control parameters are runtime target vehicle control parameters for simulating an EV. Because the simulation model is only intended to simulate the target vehicle on an EV, the set of simulated EV control parameters has no practical use but is essential for CycleGAN training purposes.

[0109] In S808, the multiple sets of EV control parameters and the multiple sets of simulated EV control parameters are input to a second classifier, a likelihood that the multiple sets of simulated EV control parameters are the multiple sets of EV control parameters is calculated, and at each training iteration, a second generator model is updated.

[0110] Similarly, the second classifier improves the performance of the second generator model.

[0111] 9A illustrates training of an exemplary simulator model, according to an embodiment of the present disclosure. In some embodiments, as shown in FIG. 9A , training the neural network includes inputting the sets of EV control parameters generated by EV engineering model 902 to a first generator model 912 to generate a plurality of sets of simulated target vehicle controls, inputting the plurality of sets of target vehicle controls and the plurality of sets of simulated target vehicle controls to a first classifier 922 to calculate a likelihood that the plurality of sets of simulated target vehicle controls are the plurality of sets of target vehicle controls, and updating first generator model 912 at each training iteration. inputting the plurality of sets of target vehicle controls generated by the above formula into a second generator model 914 to generate a plurality of sets of simulated EV control parameters; inputting the plurality of sets of EV control parameters and the plurality of sets of simulated EV control parameters into a second classifier 924 to calculate a likelihood that the plurality of sets of simulated EV control parameters are the plurality of sets of EV control parameters, and updating the second generator model 914 at each training iteration; and calculating a cycle consistency loss and updating the first generator model 912 and the second generator model 914 at each training iteration.

[0112] Training the simulation model includes two aspects. One aspect is to maximize the likelihood through any classifier. Another aspect is to minimize the cycle consistency loss. As shown in Figure 8, in S810, the cycle consistency loss is calculated to update the first generator model and the second generator model in each training iteration.

[0113] At each training iteration, the cycle consistency loss is calculated and there are two ways it can be used to update the first and second generator models: the cycle consistency loss can be calculated as a forward cycle consistency loss or a backward cycle consistency loss, both of which are described below.

[0114] 9B illustrates a calculation of forward cycle integrity loss according to an embodiment of the present disclosure. In some embodiments, as shown in FIG. 9B , calculating the cycle integrity loss includes inputting a plurality of sets of EV control parameters generated by EV engineering model 902 into a first generator model 912 to generate a plurality of sets of simulated subject vehicle controls, inputting the plurality of sets of simulated subject vehicle controls generated by first generator model 912 into a second generator model 914 to generate a plurality of sets of simulated EV control parameters, and comparing the plurality of sets of EV control parameters with the plurality of sets of simulated EV control parameters generated by second generator model 914.

[0115] 9C illustrates calculating reverse cycle consistency loss according to an embodiment of the present disclosure. In some embodiments, as shown in FIG. 9C , calculating the cycle consistency loss includes inputting a plurality of sets of subject vehicle controls generated by subject vehicle engineering model 904 into a second generator model 914 to generate a plurality of sets of simulated EV control parameters, inputting the plurality of sets of simulated EV control parameters generated by second generator model 914 into a first generator model 912 to generate a plurality of sets of simulated subject vehicle controls, and comparing the plurality of sets of subject vehicle controls with the plurality of sets of simulated subject vehicle controls generated by first generator model 912.

[0116] Referring to FIG. 6, at S610, the set of simulated target vehicle controls is output to the EV such that the EV is controlled to achieve multiple target behaviors of the target vehicle based on the set of simulated target vehicle controls.

[0117] In some embodiments, a simulation model is trained on a readily available data set. The simulation model accurately translates a set of EV control parameters into a set of simulated target vehicle controls. The set of simulated target vehicle controls is used to control the EV in a manner that provides the driver of the EV with the authentic look and feel of multiple target vehicle behaviors.

[0118] In some embodiments, a safety check is performed on the set of simulated target vehicle controls before outputting the set of simulated target vehicle controls to the EV. The safety check ensures that the EV will operate safely when the set of simulated target vehicle controls is used to control the EV.

[0119] In some embodiments, creating the simulation model includes retrieving the simulation model from a cloud server that includes multiple pre-trained simulation models for each of multiple target vehicles. The multiple simulation models may be trained respectively for the multiple target vehicles. Different target vehicles correspond to different simulation models. The EV retrieves the simulation model corresponding to the target vehicle selected by the driver of the EV.

[0120] In some embodiments, the plurality of target behaviors of the target vehicle include vehicle handling characteristics, gear shifting, dynamic engine sounds and vibrations to the driver's seat, a simulated dashboard of the target vehicle, or a combination thereof. The plurality of target behaviors of the target vehicle are not pre-configured on the EV.

[0121] In some embodiments, the plurality of vehicle parameters of the EV include steering wheel position, vehicle weight distribution, road surface condition, shock distribution, brake position, torque vector, battery availability, accelerator pedal position, steering wheel position, observed traction, available power, motor temperature, battery temperature, gyro measurements, accelerator measurements, suspension position, throttle position, brake position, regeneration setting, window position, or a combination thereof.

[0122] In some embodiments, each set of simulated target vehicle controls includes a target turning angle, a target steering ratio, a target weight distribution, a target wheelbase, a target acceleration or deceleration, a target drive wheel, a target traction setting, a target handling characteristic, a target engine sound on at least the driver's side of the EV, or a combination thereof.

[0123] In some embodiments, the target vehicle includes an internal combustion engine (ICE) vehicle or another electric vehicle.

[0124] In some embodiments, onboard computer system 230 may obtain output parameters (i.e., a set of simulated target vehicle controls) from simulator 402 to achieve a target behavior of the target vehicle. After obtaining the output parameters, based on the safety information, onboard computer system 230 may perform a second safety check on the parameters (control parameters) to determine whether the particular parameter is within a safe range (second safety range) for EV 110. If any parameter is outside the safe range, onboard computer system 230 may generate an error message and may stop using the out-of-range parameter in any further processes or actions.

[0125] In some embodiments, output parameters from simulator 402 may be fed into another machine learning process to determine safety risks. The other machine learning process may include a reinforcement learning algorithm that identifies immediate (imminent) safety risks and avoids operating the EV in the presence of the immediate safety risks, as well as identifies future safety risks and avoids operating the EV in the presence of future safety risks.

[0126] In some embodiments, a reinforcement learning algorithm is used to dynamically adjust the safety margin of the second safety check to minimize safety risks.

[0127] Further, on-board computer system 230 may perform one or more actions to achieve the target behavior of the target vehicle based on the output parameters. For example, on-board computer system 230 may generate control parameters for relevant subsystems or components of EV 110 to control the subsystems / components to achieve the simulated target behavior of the target vehicle. For example, the one or more control parameters may include steering control, powertrain control, suspension control, dashboard display, engine sound, haptic control, driver's seat control, or a combination thereof. The simulated behavior of the target vehicle may include vehicle handling characteristics, gear shifting, dynamic engine sound and vibration to the driver's seat, a simulated dashboard of the target vehicle, and a combination thereof.

[0128] Certain actions, such as display and sound-related actions, may be performed statically by onboard computer system 230, while certain other actions, such as powertrain and driving-related actions, may be performed dynamically while EV 110 is in operation. For example, onboard computer system 230 may render different dashboard displays on the EV as target dashboards for target vehicles. In some embodiments, onboard computer system 230 may perform automatic dashboard adjustments on EV 110 as target dashboards for target vehicles. For example, onboard computer system 230 may divide the electric dashboard of EV 110 into multiple zones, measure the ambient light intensity of each zone with one or more light sensors, and adjust the contrast, brightness, and content of each zone based on the ambient light intensity of each zone.

[0129] In one embodiment, based on the output parameters, onboard computer system 230 may adjust certain subsystems to simulate a target turning angle, a target steering ratio, a target weight distribution, and a target wheelbase. In another embodiment, based on the output parameters, onboard computer system 230 may adjust certain subsystems to simulate a target acceleration or deceleration, a target drive wheel, a target traction setting, and a target handling characteristic. In another embodiment, based on the output parameters, onboard computer system 230 may adjust an onboard speaker subsystem to simulate a target engine sound. In another embodiment, based on the output parameters, onboard computer system 230 may perform driver seat transformations, including tightening bolts, changing stiffness, changing lumbar support, or a combination thereof, within a preconfigured time period and resume the driver seat configuration after the preconfigured time period has expired.

[0130] Additionally, onboard computer system 230 may perform one or more data operations based on the simulated target vehicle on the EV. That is, specific vehicle-related data may be processed and / or analyzed in various ways based on the virtual vehicle, i.e., the simulated target vehicle. For example, onboard computer system 230 may obtain the location of EV 110 and perform further location-based actions, such as determining local driving restrictions corresponding to the location and applying local driving restrictions restrictions on control parameters to EV 110. In one embodiment, onboard computer system 230 may collect simulation process data, as well as vehicle data of the simulated target vehicle, and anonymize and upload the collected data to cloud server 140 for training a vehicle simulator model. In another embodiment, onboard computer system 230 may collect simulated driving data of driver 120 and upload the driving data to social media sites identified by driver 120. Other data operations may also be performed.

[0131] Therefore, embodiments of the present disclosure provide an EV simulation method and system that simulates the behavior of a target vehicle on an EV. While driving the EV, a driver can request a simulation of the target vehicle's behavior and enjoy the look and feel of the target vehicle and the fun of driving the target vehicle. The in-vehicle simulation system also allows the driver to share driving data collected when the EV is operating in simulation mode on social media. When the EV is operating in simulation mode, vehicle data generated by the in-vehicle simulation system can also be collected and uploaded to a server as training data for training a new simulator model or for retraining and updating a current simulator model.

[0132] The present disclosure also provides a method for mining driving data to obtain a training dataset. When an EV operates in a simulation mode, various parameters of the EV can be collected in real time. After the EV completes a trip, if the driver consents to the collection of driving data, the driving data is uploaded to a cloud server. The driving data may be mined to discover data domains. The data domains can be used to sort the driving data into data clusters. The data clusters can be used as a training dataset for training a simulation model.

[0133] 10 shows a flowchart of an exemplary method for mining running data of a training dataset according to an embodiment of the present disclosure. As shown in FIG. 10, the running data mining method includes the following steps:

[0134] In S1002, driving data is acquired from the EV.

[0135] The trip data is collected in real time while the EV is operating in simulation mode. The trip data may be stored in the EV's on-board computer system. The trip data may be uploaded to a cloud server after the EV completes a trip. The trip data may be retrieved directly from the EV's on-board computer system storage.

[0136] In S1004, the driving data is pre-processed to identify data that is not relevant to the simulation model training.

[0137] Because driving data is collected to train a simulation model, data that is not relevant to simulation model training may be identified.

[0138] In S1006, a heat map is created and the pre-processed data is sorted into a plurality of data clusters corresponding to a plurality of data regions.

[0139] For example, the heat map shown in FIG. 11 can be used to reduce noise from a large amount of driving data for each of multiple data regions. The multiple data regions can be used to classify (cluster) the driving data into multiple data clusters. The heat map can be used to classify and / or generalize the driving data, which can then be used to further purge the driving data to identify noise within the driving data. The purged driving data can then be used as a training data set to improve the performance of a simulation model.

[0140] In S1008, the plurality of data clusters is output as a training data set for training a simulation model.

[0141] 12 shows a flowchart for discovering new data regions according to an embodiment of the present disclosure. Data that cannot be classified into multiple data clusters can be processed to discover new data regions.

[0142] In S1202, pre-processed driving data that does not belong to any data cluster is obtained. For example, the pre-processed driving data that does not belong to any existing data cluster can be used to form a new data cluster that identifies new correlations. The new data cluster can be used to extract valuable insights.

[0143] In S1204, a pattern recognition process is performed to determine new data regions.

[0144] The pattern recognition process may include a random forest algorithm to gain more valuable insights from the pre-processed driving data that does not belong to any data cluster. The pattern recognition process may be used to extract training data sets for training simulation models for specific / extreme scenarios.

[0145] In S1206, the new data region is output for use in processing the driving data to obtain a training data set for training the simulation model.

[0146] For example, driving data based on weather and road topography may be noise for certain data areas, such as frequent braking by drivers in congested areas such as India or China. However, for other data areas, such as acceleration / torque during rain / snow, frequent braking data can be used to improve the simulation model. In some embodiments, the frequent braking data can be used to determine a safety margin for the second safety check.

[0147] In embodiments of the present disclosure, driving data may be collected with driver consent to train a simulation model to improve simulation performance. Driving data may also be used to determine a safety margin for safety checks.

[0148] The above embodiments have described in detail the objectives, technical solutions, and beneficial effects of the present disclosure. The above disclosed embodiments are only some of the embodiments of the present invention, not all of the embodiments of the present invention, and should not be used to limit the scope of the present invention. Any other embodiments obtained based on the above embodiments without the need for creative work by those skilled in the art fall within the scope of protection of the present disclosure. Furthermore, under non-contradictory circumstances, the embodiments and features in the embodiments may be combined with each other. Therefore, any changes, equivalent substitutions, and modifications made according to the present disclosure remain within the scope of the present disclosure.

Claims

1. 1. A method for simulating an electric vehicle (EV), comprising: a process for creating a simulation model associating a plurality of target behaviors of a target vehicle with the electric vehicle, the simulation model being a neural network trained to reflect relationships between EV control parameters and simulated target vehicle controls, the process comprising: (a) collecting a plurality of sets of EV control parameters; (b) collecting a plurality of sets of simulated target vehicle controls; (c) processing the plurality of sets of EV control parameters and the plurality of sets of target vehicle controls using a heat map to identify noise through data clustering; and (d) after processing the plurality of sets of EV control parameters and the plurality of sets of target vehicle controls, training the neural network with the plurality of sets of EV control parameters and the plurality of sets of target vehicle controls to obtain the simulation model corresponding to the target vehicle; obtaining a plurality of vehicle parameters of the electric vehicle to generate an additional set of EV control parameters, the plurality of vehicle parameters of the electric vehicle including at least runtime parameters of the electric vehicle and operating conditions of the electric vehicle; obtaining a plurality of configuration parameters of the target vehicle, the configuration parameters including at least information about the target vehicle and settings of the target vehicle for a driver; providing a set of simulated subject vehicle controls using the simulation model based on the additional set of EV control parameters and the plurality of configuration parameters of the subject vehicle; outputting the set of simulated target vehicle controls to the electric vehicle such that the electric vehicle is controlled to achieve the plurality of target behaviors of the target vehicle based on the set of simulated target vehicle controls; Including, A simulation method comprising:

2. performing a safety check on the set of simulated target vehicle controls before outputting the set of simulated target vehicle controls to the electric vehicle; Further comprising:

2. The simulation method according to claim 1.

3. The process of creating the simulation model includes: and obtaining, for each of a plurality of target vehicles, a simulation model from a cloud server including a plurality of trained simulation models via a neural network.

2. The simulation method according to claim 1.

4. An electric vehicle (EV), a wireless communication device for connecting to a cloud server and / or a mobile device carried by a driver of the electric vehicle; an on-board computer system; The in-vehicle computer system a process for creating a simulation model associating a plurality of target behaviors of a target vehicle with the electric vehicle, the simulation model being a neural network trained to reflect relationships between EV control parameters and simulated target vehicle controls, the process comprising: (a) collecting a plurality of sets of EV control parameters; (b) collecting a plurality of sets of simulated target vehicle controls; (c) processing the plurality of sets of EV control parameters and the plurality of sets of target vehicle controls using a heat map to identify noise through data clustering; and (d) after processing the plurality of sets of EV control parameters and the plurality of sets of target vehicle controls, training the neural network with the plurality of sets of EV control parameters and the plurality of sets of target vehicle controls to obtain the simulation model corresponding to the target vehicle; obtaining a plurality of vehicle parameters of the electric vehicle to generate an additional set of EV control parameters, the plurality of vehicle parameters of the electric vehicle including at least runtime parameters of the electric vehicle and operating conditions of the electric vehicle; obtaining a plurality of configuration parameters of the target vehicle, the configuration parameters including at least information about the target vehicle and settings of the target vehicle for a driver; providing a set of simulated subject vehicle controls using the simulation model based on the additional set of EV control parameters and the plurality of configuration parameters of the subject vehicle; outputting the set of simulated target vehicle controls to the electric vehicle such that the electric vehicle is controlled to achieve the plurality of target behaviors of the target vehicle based on the set of simulated target vehicle controls. An electric vehicle characterized by:

5. Collecting the plurality of sets of EV control parameters includes: collecting a plurality of sets of the EV control parameters generated by inputting the plurality of vehicle parameters of the electric vehicle into an EV simulator that simulates a behavior of the electric vehicle; collecting a plurality of sets of the EV control parameters generated by a plurality of anonymized electric vehicles; and / or collecting a plurality of sets of the EV control parameters generated by the electric vehicles.

2. The simulation method according to claim 1.

6. collecting a plurality of sets of target vehicle controls collecting a plurality of sets of control of the subject vehicle generated by inputting the plurality of vehicle parameters of the electric vehicle into a subject vehicle simulator that simulates a behavior of the subject vehicle; collecting a plurality of sets of the target vehicle controls generated by a plurality of anonymized target vehicles; and / or and collecting a plurality of sets of the subject vehicle controls generated by the subject vehicle.

7. the neural network is a cycle generative adversarial network (CycleGAN) including a first generator model, a first classifier, a second generator model, and a second classifier; Training the neural network comprises: inputting the plurality of sets of EV control parameters into the first generator model to generate a plurality of sets of simulated target vehicle controls; inputting the plurality of sets of target vehicle controls and the plurality of sets of simulated target vehicle controls into the first classifier to calculate likelihoods of the plurality of sets of simulated target vehicle controls as the plurality of sets of target vehicle controls, and updating the first generator model at each training iteration; inputting the plurality of sets of target vehicle controls into the second generator model to generate a plurality of sets of simulated EV control parameters; inputting the plurality of sets of EV control parameters and the plurality of sets of simulated EV control parameters to the second classifier to calculate likelihoods of the plurality of sets of simulated EV control parameters as the plurality of sets of EV control parameters, and updating the second generator model at each training iteration; and at each training iteration, calculating a cycle consistency loss to update the first generator model and the second generator model. The simulation method according to claim 1 .

8. Calculating the cycle consistency loss comprises: inputting the plurality of sets of EV control parameters into the first generator model to generate the plurality of sets of simulated target vehicle controls; inputting the plurality of sets of simulated target vehicle controls generated by the first generator model into the second generator model to generate the plurality of sets of simulated EV control parameters; and comparing the plurality of sets of EV control parameters with the plurality of sets of simulated EV control parameters generated by the second generator model.

9. Calculating the cycle consistency loss comprises: inputting the plurality of sets of target vehicle controls into the second generator model to generate the plurality of sets of simulated EV control parameters; 8. The simulation method of claim 7, comprising: inputting a plurality of sets of the simulated EV control parameters generated by the second generator model into the first generator model to generate a plurality of sets of the simulated target vehicle controls; and comparing the plurality of sets of target vehicle controls with the plurality of sets of simulated target vehicle controls generated by the first generator model.

10. 1. A simulation system for an electric vehicle (EV), comprising: a plurality of input devices for providing a plurality of vehicle parameters; a memory containing program instructions; a processor coupled to the memory and the plurality of input devices, the processor, when the program instructions are executed, creating a simulation model associating a plurality of target behaviors of a target vehicle with the electric vehicle, the simulation model being a neural network trained to reflect relationships between EV control parameters and simulated target vehicle controls, by: (a) collecting a plurality of sets of EV control parameters; (b) collecting a plurality of sets of simulated target vehicle controls; (c) processing the plurality of sets of EV control parameters and the plurality of sets of target vehicle controls using a heat map to identify noise through data clustering; and (d) after processing the plurality of sets of EV control parameters and the plurality of sets of target vehicle controls, training the neural network with the plurality of sets of EV control parameters and the plurality of sets of target vehicle controls to obtain the simulation model corresponding to the target vehicle; obtaining the plurality of vehicle parameters of the electric vehicle to generate an additional set of EV control parameters, the plurality of vehicle parameters of the electric vehicle including at least a runtime parameter of the electric vehicle and an operating state of the electric vehicle; obtaining a plurality of configuration parameters of the target vehicle, the configuration parameters including at least information about the target vehicle and settings of the target vehicle for a driver; providing a set of simulated subject vehicle controls using the simulation model based on the additional set of EV control parameters and the plurality of configuration parameters of the subject vehicle; and outputting the set of simulated target vehicle controls to the electric vehicle such that the electric vehicle is controlled to achieve the plurality of target behaviors of the target vehicle based on the set of simulated target vehicle controls.

Images