Identifying AI personas that optimally influence driving behavior

Abstract

This invention provides a system and method for identifying artificial intelligence (AI) personas that optimally influence driving behavior. [Solution] For example, the method of the technology of the present disclosure may include (1) determining a target driving behavior for the driver of a vehicle based on driving conditions, (2) identifying a persona for an AI assistant that has the highest predicted probability of influencing the driver to perform the target driving behavior, and (3) using the AI ​​assistant having the identified persona to present the driver with information to influence the driver to perform the target driving behavior. In a particular embodiment, identifying a persona for an AI assistant that has the highest predicted probability of influencing the driver to perform the target driving behavior may include determining that the identified persona reduces the objective function the most among a plurality of personas.

Description

Technical Field

[0001] The present disclosure generally relates to automotive systems and technologies. More particularly, some embodiments relate to identifying an artificial intelligence (AI) persona that optimally affects driving behavior.

Background Art

[0002] An artificial intelligence (AI) assistant can refer to computer technology that utilizes generative AI to perform tasks, answer questions, execute commands, etc.

[0003] Some existing vehicle technologies utilize an AI assistant to present information to a driver. In some cases, such information can be presented to influence the driver to perform a target / desired driving behavior (e.g., changing lanes or refraining from changing lanes, preparing for a right or left turn, decelerating, driving with less aggression, etc.).

Summary of the Invention

[0004] Systems according to various embodiments of the disclosed technology are provided. The systems according to embodiments of the technology disclosed herein include.

[0005] In various embodiments, methods are provided. The methods according to embodiments of the technology disclosed herein include.

[0006] Other features and aspects of the disclosed technology will become apparent from the following detailed description, taken in conjunction with the accompanying drawings that exemplify features of embodiments of the disclosed technology. This summary is not intended to limit the scope of the invention described herein, which is defined only by the claims appended hereto.

[0007] This disclosure is described in detail by reference to the following figures in one or more different embodiments. The figures are provided for illustrative purposes only and merely illustrate typical or exemplary embodiments. [Brief explanation of the drawing]

[0008] [Figure 1] Figure 1 shows examples of vehicles according to various embodiments of the technology of this disclosure. [Figure 2] Figure 2 shows examples of processes that can be performed to identify and deploy artificial intelligence (AI) personas that best influence driving behavior, according to various embodiments of the technology of this disclosure. [Figure 3] Figure 3 shows other examples of processes that can be performed to identify and deploy AI personas that best influence driving behavior, according to various embodiments of the technology of this disclosure. [Figure 4] Figure 4 shows examples of AI personas according to various embodiments of the technology of this disclosure. [Figure 5] Figure 5 shows further examples of AI personas according to various embodiments of the technology of this disclosure. [Figure 6] Figure 6 shows other examples of AI personas according to various embodiments of the technology of this disclosure. [Figure 7] Figure 7 shows an example of a computational component that can be used to implement various features of the embodiments described herein. [Modes for carrying out the invention]

[0009] These figures are not exhaustive and do not limit this disclosure to the exact form disclosed.

[0010] As described above, some existing vehicle technologies utilize artificial intelligence (AI) assistants to present information to the driver. In some cases, such information (e.g., verbal instructions, audiovisual notifications) may be presented to influence the driver to perform a target / desired driving action (e.g., changing lanes or refraining from changing lanes, preparing to turn left or right, slowing down, driving with less aggression, etc.).

[0011] The systems and methods of the technologies disclosed herein improve upon existing technologies by identifying and deploying specific / optimal personas (e.g., strict personas, friendly personas, personas based on the driver's family composition, personas based on occupants of nearby vehicles) for AI assistants that are predicted to be most likely to influence drivers and perform target driving behaviors. In other words, the systems and methods can predict that the identified personas will exert the greatest social influence on the driver (compared to other available personas) and influence target driving behaviors. In this way, the systems and methods can improve upon existing / alternative technologies that utilize a single / general persona for the AI ​​assistant—or that cannot tailor / select a persona for the AI ​​assistant based on desired driving behaviors or driving situations—in terms of their potential to influence target driving behaviors.

[0012] For example, the system of the technology of this disclosure may be configured to (1) determine a target driving behavior for the vehicle driver based on driving conditions, (2) identify a persona for an AI assistant that has the highest predicted probability of influencing the driver to perform the target driving behavior, and (3) use the AI ​​assistant having the identified persona to present the driver with information to influence the driver to perform the target driving behavior.

[0013] In various embodiments, the system and method can identify the optimal persona for an AI assistant by minimizing the objective cost function. For example, the system and method can identify the persona for an AI assistant that has the highest predicted probability of influencing the driver to perform a target driving behavior by determining that the identified persona reduces the objective cost function the most among several possible personas. Here, the terms in the objective cost function may reflect the impact (in terms of indicators such as safety and performance) that using the AI ​​assistant with each persona has on the predicted probability of the driver performing the target driving behavior. Through this analysis and subsequent deployment of the identified / optimal personas, the system and method can improve the potential to influence target driving behavior compared to existing / alternative technologies that utilize a single / generic persona for the AI ​​assistant. Furthermore, the system and method can improve / fine-tune the prediction of driver behavior based on the cost function by adding additional cost function terms—i.e., terms that reflect the impact of using a specific persona for the AI ​​assistant to inform the driver.

[0014] In specific implementations, the system and method can also identify time intervals for presenting information to the driver that have the highest predicted probability of influencing the driver to perform a target driving behavior. For example, the identified / optimal time interval could be a predetermined time before the target driving behavior (e.g., making a right or left turn) is expected to occur, or a predetermined time after the driver engages in undesirable driving behavior (e.g., aggressive driving, running one or more red lights, speeding, etc.). In this way, the system and method can improve the potential to influence target driving behavior compared to existing / alternative technologies that cannot adjust (or consider) the timing of information presented by the AI ​​assistant. That is, the system and method are designed so that the timing of information presentation can be as important / impactful as the content of the information itself.

[0015] Similar to identifying the optimal persona for an AI assistant, systems and methods may use an objective cost function to identify the optimal time interval for providing information to a driver. For example, a system and method may determine that a specific time interval (and persona) among several time intervals (and personas) reduces the objective cost function the most. As discussed above, the first term of the objective cost function may reflect the impact of using an AI assistant with each persona on the predicted probability that a driver will engage in a target driving behavior. Similarly, the second term of the objective cost function may reflect the impact of presenting information within each time interval on the predicted probability that a driver will engage in a target driving behavior. Through this analysis and the subsequent deployment of the identified / optimal persona at the identified / optimal time interval, systems and methods can improve the potential impact on target driving behavior compared to existing / alternative technologies that utilize a single / general persona for the AI ​​assistant and / or cannot adjust (or consider) the timing of information delivered by the AI ​​assistant. Furthermore, the system and method can improve / fine-tune the prediction of driver behavior based on the cost function by adding additional cost function terms—namely, a first term reflecting the impact of using a specific persona about the AI ​​assistant to present information to the driver, and a second term reflecting the impact of presenting information to the driver within a specific time interval.

[0016] In various implementations, the system and method may utilize augmented reality (AR) technology to present the driver with an identified / optimal persona. For example, the system and method may use an AR device (e.g., AR glasses worn by the driver) to display a visual avatar of a specific persona to the driver. For example, if the identified persona is the driver's grandparents, the system and method may use an AR device to display a visual avatar of the driver's grandparents appearing in the passenger seat of the vehicle. In connection with this, the system and method may have the visual avatar present linguistic information / instructions to the driver in the voice of the driver's grandparents.

[0017] In certain implementations, systems and methods may leverage machine learning / AI to learn which personas of the AI ​​assistant are most likely to influence different driving behaviors. For example, systems and methods may use machine learning to analyze a particular driver's driving behavior history when different personas of the AI ​​assistant are used to present instructions / information to the driver. Thus, systems and methods can learn which personas are most likely to influence a particular driver's different driving behaviors. In a related context, in some implementations, systems and methods may leverage the driving data history of multiple drivers to learn which personas are most likely to influence different driving behaviors of the average driver.

[0018] In various implementations, the system and method can create an AI persona and then transfer the AI ​​persona from one vehicle to another. For example, the driver of a first vehicle may create an AI persona based on themselves or another high-frequency occupant of the first vehicle (e.g., the driver's child) by uploading images / videos and audio recordings of the driver / high-frequency occupant to the system of the technology of this disclosure. In this case, the system (which may be implemented in the first vehicle) may use various techniques (stable diffusion, large-scale language models (LLMs)) to create an AI persona based on the uploaded images / videos and audio recordings of the driver / high-frequency occupant of the first vehicle. For example, the system may apply a generative model to photographs of the driver's family members so that an AI persona based on the family member appears and speaks a generated sentence (e.g., "Your driving is getting scarier and scarier") that minimizes a cost function. Furthermore, to further minimize the cost function, the system may adjust the visual appearance of the AI ​​persona based on the family member to reflect an anxious emotional state.

[0019] Therefore, when the first vehicle is traveling on a road section and detects that the second vehicle is driving unsafely (e.g., speeding, weaving between lanes), the first vehicle can transfer the AI ​​persona of the driver / high-frequency occupant of the first vehicle to the second vehicle. The second vehicle may determine that the AI ​​persona based on a person in a nearby vehicle (e.g., an occupant of the first vehicle) is most likely to influence the driver of the second vehicle to take a target driving action (e.g., slow down, stay in the lane). Therefore, the second vehicle may use the AI ​​persona transferred from the second vehicle to present information to the driver of the first vehicle.

[0020] As another example, a system of the technology of this disclosure may collect audiovisual recordings from within the vehicle that capture conversations between the driver and other occupants of the vehicle (e.g., the driver's friends and family members). In this case, the system can create one or more personas about the AI ​​assistant based on these audiovisual recordings. For example, the system can create a persona based on the voices of the other occupants of the vehicle (e.g., voice quality, speech cadence, language syntax, etc.). In this case, the system can learn which of these AI personas is most likely to influence different target driving behaviors for the driver and deploy the AI ​​personas accordingly.

[0021] It should be understood that the systems and methods described herein provide a specific technical solution in the field of AI assistant technology. Specifically, the systems and methods improve AI assistant technology by dynamically identifying and deploying specific / optimal personas for AI assistants that are predicted to be most likely to influence drivers and perform target driving behaviors. In this way, the systems and methods offer a technical improvement over existing / alternative AI assistant technologies that utilize a single / generic persona or are unable to adjust / select personas based on desired driving behaviors or driving situations.

[0022] The system and method can also provide certain technical improvements to autonomous driving technology and driving assistance technology. That is, the system and method add an additional cost function term - namely, a first term that reflects the impact of using a specific persona for the AI assistant that presents information to the driver, and a second term that reflects the impact of presenting information to the driver within a specific time interval - to improve / fine-tune the prediction of driver behavior based on the cost function. Predicting driver behavior can be an important operation for autonomous driving technology and driving assistance technology. Therefore, by facilitating the prediction of driver behavior based on the improved / fine-tuned cost function, the system and method also provide certain technical improvements to autonomous driving technology and driving assistance technology.

[0023] The systems and methods disclosed herein can be implemented using any of a plurality of different vehicles and vehicle types. For example, the systems and methods disclosed herein can be used with automobiles, trucks, motorcycles, recreational vehicles, and other types of vehicles. Additionally, the principles disclosed herein may be utilized by a system external to the vehicle (e.g., a cloud-based system).

[0024] FIG. 1 shows an example of a vehicle 100 according to various embodiments of the technology of the present disclosure.

[0025] Before describing the components of vehicle 100 in detail individually, it may be useful to describe an overview of the operation at a higher level.

[0026] In certain embodiments, vehicle 100 (or more specifically, digital persona circuit 110) can determine a target driving behavior for the driver of vehicle 100 based on the driving situation. Vehicle 100 can make this determination based on information obtained from at least one of sensor 152, autonomous vehicle (AV) system 174, and semi-autonomous vehicle (SAV) system 176.

[0027] In that case, vehicle 100 (or more specifically, digital persona circuit 110) may identify a (digital) persona for an AI assistant that has the highest predicted probability of influencing the driver to perform a target driving behavior. The identified persona may be one of several available digital personas stored in memory unit 108. As described above, in some implementations, vehicle 100 (or more specifically, digital persona circuit 110) may generate one or more of these digital personas based on audiovisual recordings within vehicle 100 or information obtained from other vehicles 180 (e.g., digital personas based on images and / or audio recordings of occupants of other vehicles 180). In certain implementations, vehicle 100 (or more specifically, digital persona circuit 110) may receive digital personas themselves from other vehicles 180.

[0028] In this case, the vehicle 100 (or more specifically, the digital persona circuit 110) may use an AI assistant with an identified persona to present the driver with information to influence the driver to perform target driving behaviors. In some implementations, this may include presenting / displaying the identified persona via an audiovisual unit 172. In other implementations, this may include having an augmented reality (AR) device 140 (e.g., an AR headset, AR glasses, etc.) present / display the identified persona to the driver of the vehicle 100.

[0029] Next, referring more closely to the vehicle 100 and Figure 1, as shown, the vehicle 100 includes a digital persona circuit 110, a sensor 152, and a vehicle system 170. The sensor 152 and the vehicle system 170 can communicate with the digital persona circuit 110 via a wired or wireless communication interface. Although the sensor 152 and the vehicle system 170 are depicted as communicating with the digital persona circuit 110, they may communicate with each other. The digital persona circuit 110 may be implemented as an electronic control unit (ECU) or as part of an ECU. In other embodiments, the digital persona circuit 110 may be implemented independently of the ECU.

[0030] In the specific example shown in Figure 1, the digital persona circuit 110 includes a communication circuit 101, an identification circuit 103 (including a processor 106 and a memory unit 108), and a power supply 112. The components of the digital persona circuit 110 are shown communicating with each other via a data bus, but other interfaces may be included.

[0031] The processor 106 may include one or more general-purpose processing units (GPUs), central processing units (CPUs), microprocessors, or other suitable processing systems. The processor 106 may include single-core or multi-core processors. The memory unit 108 may include one or more different forms of memory or data storage (e.g., flash, RAM, etc.) used to store digital personas, terms / parameters of cost functions for predicting driver behavior, parameters of machine learning models, calibration parameters, images (for analysis or history), point parameters, instructions and variables for the processor 106, and other suitable information. The memory unit 108 may consist of one or more modules of one or more different types of memory and may be configured to store data and other information as well as operational instructions that may be used by the processor 106.

[0032] While the example in Figure 1 shows the use of a processor and memory circuit, in various embodiments the identification circuit 103 may be implemented using any form of circuitry, including, for example, hardware, software, or a combination thereof. For example, one or more processors, controllers, ASICs, PLAs, PALs, CPLDs, FPGAs, logic components, software routines, or other mechanisms may be implemented to constitute the digital persona circuit 110.

[0033] The communication circuit 101 may utilize a wireless transceiver circuit 102 having an associated antenna 105 for wireless communication. The communication circuit 101 may also utilize a wired I / O interface 104 having an associated hardwired data port (not shown). As this example shows, communication with the digital persona circuit 110 may include either or both wired and wireless communication. The wireless transceiver circuit 102 may include a transmitter and receiver (not shown) and enable wireless communication over multiple communication protocols, such as WiFi, Bluetooth®, Near Field Communication (NFC), Zigbee®, and several other wireless communication protocols that are standardized, proprietary, open, point-to-point, networked, or otherwise. The antenna 105 is coupled to the wireless transceiver circuit 102 and is used by the wireless transceiver circuit 102 to wirelessly transmit and receive signals to and from wireless devices. These radio signals may include virtually any kind of information transmitted to or received by the digital persona circuit 110 to / from other entities such as the sensor 152, the vehicle system 170, the AR device 140, and other vehicles 180.

[0034] The wired I / O interface 104 may include transmitters and receivers (not shown) for hardwired communication with other devices. For example, the wired I / O interface 104 can provide hardwired interfaces to other components, including sensors 152 and vehicle systems 170. The wired I / O interface 104 can communicate with other devices using Ethernet or one of several other wired communication protocols, which may be standardized, proprietary, open, point-to-point, networked, or otherwise.

[0035] The power source 112 may include a battery (e.g., Li-ion, Li-Polymer, NiMH, NiCd, NiZn, NiH2, etc., which may be multiple, and may be either rechargeable or primary batteries), a power connector (e.g., for connecting to the vehicle's power supply), an energy harvester (e.g., a solar cell, a piezoelectric system, etc.), or it may include any other suitable power source.

[0036] Sensor 152 may include, for example, a vehicle acceleration sensor 113, a vehicle speed sensor 114, a wheel spin sensor 116 (e.g., one for each wheel), a tire pressure monitoring system (TPMS) 120, an accelerometer such as a triaxial accelerometer 122 for detecting vehicle roll, pitch, and yaw, a vehicle clearance sensor 124, left / right and front / rear slip ratio sensors 126, an environmental sensor 128 (e.g., for detecting salinity or other environmental conditions), an image sensor 130, and a position sensor 132. Other sensors 135 may also be included, if suitable for a given implementation of vehicle 100. Other sensors 135 may include, for example, a gyroscope, an odometer, etc. Other sensors 135 may also include an acoustic sensor configured to capture the voices of occupants inside vehicle 100.

[0037] In some embodiments, the image sensor 130 may include one or more cameras configured to generate image data of the environment around or inside the vehicle 100. The image data may include images of the environment, including people inside the vehicle 100 and people in nearby vehicles in the road section.

[0038] In certain embodiments, the position sensor 132 may include a global positioning satellite sensor, a global positioning sensor, or other types of vehicle position sensors. The position sensor 132 may be configured to generate position data for the vehicle 100 and / or position data for landmarks in the environment surrounding the vehicle 100. The position data may include the precise coordinates (e.g., latitude, longitude, altitude) of the position of the vehicle 100 or the landmark on the Earth's surface.

[0039] In some embodiments, one or more of the sensors 152 may include their own processing capabilities for calculating results regarding additional information that can be provided to the digital persona circuit 110. In other embodiments, one or more of the sensors 152 may be data acquisition-only sensors that provide only raw data to the digital persona circuit 110. In further embodiments, one or more hybrid sensors may be included that provide the digital persona circuit 110 with a combination of raw and processed data. The sensors 152 may provide analog outputs, digital outputs, or a combination of both.

[0040] The vehicle system 170 may include any of several different vehicle components or subsystems used to control or monitor the vehicle 100 and various aspects of its performance. For example, the vehicle system 170 may include any or a combination thereof of the audiovisual unit 172, the automated vehicle (AV) system 174, the semi-autonomous vehicle (SAV) system 176, and other vehicle systems 178.

[0041] As described above, the audiovisual unit 172 may be used to present an AI assistant persona to the driver of the vehicle 100. In certain examples, the audiovisual unit 172 may be implemented as part of an in-vehicle infotainment (IVI) system (an IVI system can provide entertainment and information to the occupants of a vehicle through control elements such as an audio / video interface, a touchscreen display, a button panel, and voice commands). In certain examples, the audiovisual unit 172 may be a liquid crystal display (LCD) screen. In various examples, the audiovisual unit 172 may include multiple displays / screens, such as a dashboard display and a head-up display.

[0042] In general, AV and SAV systems (e.g., AV system 174 and SAV system 176) can control the driving behavior of a vehicle. AV and SAV systems can interpret sensor information, identify appropriate traffic configurations, determine the vehicle's navigation path, and activate vehicle systems according to the determined vehicle navigation path. Many AV and SAV systems are designed to minimize vehicle collisions.

[0043] As described above, the digital persona circuit 110 may use information from the AV system 174 and SAV system 176 to determine the target driving behavior based on the driving conditions. That is, existing AV and SAV systems are well known to be able to determine the desired / optimal driving behavior based on sensor data, analysis of past driving behavior, etc.

[0044] As described above, the system and method can also provide certain technical improvements to autonomous driving and driver assistance technologies (e.g., AV system 174 and SAV system 176). Specifically, the system and method can improve / fine-tune the prediction of driver behavior based on the cost function by adding additional cost function terms—for example, a first term reflecting the impact of using a specific persona for the AI ​​assistant that delivers information to the driver, and a second term reflecting the impact of presenting information to the driver within a specific time interval. Predicting driver behavior can be a critical operation for autonomous driving and driver assistance technologies. Therefore, by facilitating the prediction of driver behavior based on the improved / fine-tuned cost function, the system and method also provide certain technical improvements to autonomous driving and driver assistance technologies (e.g., AV system 174 and SAV system 176).

[0045] As described above, the AR device 140 may include various types of AR devices, such as AR glasses, AR headsets, and projectors / head-up displays that project AR images onto the vehicle's windshield. In certain implementations, the AR device 140 may be part of the vehicle system 170. In other implementations, the AR device 140 may be implemented independently of the vehicle 100. In such implementations, the AR device 140 can communicate with the vehicle 100 / digital persona circuit 110 via wired or wireless communication as described above.

[0046] Figure 2 shows examples of processes 200 that can be performed by System 230 to identify and deploy AI personas that best influence driving behavior, according to various embodiments of the technology of this disclosure. In some embodiments, System 230 may be implemented in a vehicle 250, which may be the same or similar vehicle as Vehicle 100 described with Figure 1.

[0047] As shown in the figure, system 230 can perform operation 202 to determine a target driving behavior for the driver of vehicle 250 based on driving conditions. To make this determination, system 230 may utilize a combination of sensor data from vehicle 250 and the vehicle 250's onboard AV or SAV system. As discussed above, existing AV and SAV systems are well known for being able to determine desired / optimal driving behavior based on sensor data, analysis of past driving behavior, etc.

[0048] In that case, system 230 can perform action 204 to identify a persona for an AI assistant that has the highest predicted probability of influencing the driver to perform a target driving behavior. In certain cases, the identified persona may be based on a specific person (e.g., an acquaintance of the driver, such as a friend or family member of the driver, or a passenger in a nearby vehicle). In other cases, the identified persona may be based on a fusion of multiple people, or may reflect a more generalized type of person, personality type, or mood (e.g., a strict authority figure, a talkative friend, an anxious child).

[0049] In various implementations, the identified persona may reflect a linguistic persona, a visual persona (e.g., a visual avatar), or a combination of both linguistic and visual personas.

[0050] As discussed above, in some implementations, system 230 may leverage machine learning / AI to learn which persona of the AI ​​assistant is most likely to influence different driving behaviors. For example, system 230 may use machine learning to analyze the driving behavior history of the vehicle 250 driver when different personas of the AI ​​assistant are used to present instructions / information to the driver. Thus, system 230 can learn which persona is most likely to influence different driving behaviors of the driver. In a related context, in some implementations, system 230 may leverage the driving data history of multiple drivers to learn which persona is most likely to influence different driving behaviors of the average driver.

[0051] In various implementations, system 230 can create AI personas and then transfer them to other vehicles, or receive AI personas from other vehicles. For example, the driver of a second vehicle may create an AI persona based on themselves or other high-frequency occupants of the second vehicle (e.g., the driver's children) by uploading images / videos and audio recordings of the driver / high-frequency occupants to system 230 or another system communicating with system 230. In this case, the uploading system (e.g., system 230 or a system communicating with system 230) may use various techniques (e.g., stable diffusion, large-scale language models (LLMs)) to create an AI persona based on the uploaded images / videos and audio recordings of the driver / high-frequency occupants of the second vehicle. Therefore, when the second vehicle is traveling on a road section and detects that vehicle 250 is driving unsafely (e.g., speeding, weaving between lanes, etc.), the second vehicle may transfer the AI ​​persona of the driver / high-frequency occupant of the second vehicle to vehicle 250 / system 230. System 230 may determine that the AI ​​persona based on a person in a nearby vehicle (e.g., an occupant of the second vehicle) is most likely to influence the driver of vehicle 250 to perform a target driving action (e.g., slowing down, staying in the lane, etc.). Therefore, system 230 may identify the AI ​​persona transferred from the second vehicle as the persona most likely to influence the driver to perform a target driving action.

[0052] As another example, system 230 may collect audiovisual recordings from within vehicle 250, capturing conversations between the driver and other occupants of vehicle 250 (e.g., the driver's friends and family). In this case, system 230 can create one or more personas about the AI ​​assistant based on these audiovisual recordings. For example, system 250 could create a persona based on the voices of other occupants of vehicle 250 (e.g., voice quality, speech cadence, linguistic syntax, etc.). In this case, system 230 could learn which of these AI personas is most likely to influence different target driving behaviors for the driver, and, in accordance with this learning, identify the optimal persona for the AI ​​assistant (based on target driving behaviors, driving conditions, etc.).

[0053] As discussed above, in a particular implementation, system 230 can identify a persona for an AI assistant that has the highest predicted probability of influencing the driver to perform a target driving action by determining that the identified persona reduces the objective function the most among multiple personas. Here, the term in the objective cost function may reflect the impact on the predicted probability of the driver performing the target driving action of using an AI assistant with each of the multiple personas.

[0054] As discussed above, in some implementations, system 230 can also identify time intervals for presenting information to the driver about an AI assistant with a identified persona that has the highest predicted probability of influencing the driver to perform a target driving action. Similar to identifying the optimal persona, system 230 can identify the optimal time interval by reducing the objective cost function. For example, system 230 may determine that a identified time interval and persona reduce the objective function the most among multiple time intervals and personas. As discussed above, the first term of the objective cost function may reflect the impact of using an AI assistant with each of the multiple personas on the predicted probability of the driver performing a target driving action. Relatedly, the second term of the objective cost function may reflect the impact of presenting information within each of the multiple time intervals on the predicted probability of the driver performing a target driving action.

[0055] As shown in the figure, the system 230 can perform action 206, which involves using an AI assistant with a specified persona to present the driver with information (e.g., instructions, suggestions, etc.) to influence the driver and lead them to a target driving behavior. In embodiments in which the (optimal) time intervals for the system 230 to present information are also specified, the system 230 may use an AI assistant with a specified persona to present information to the driver at the specified time intervals.

[0056] In a specific implementation, system 230 may use an AR device (e.g., AR device 140 in Figure 1) to display a visual avatar of an identified persona to the driver. For example, if the identified persona is an acquaintance of the driver (e.g., a friend or family member of the driver), system 230 may use the AR device to display the acquaintance's visual avatar appearing in the passenger seat of vehicle 250. In connection with this, system 250 may present information using the visual avatar's gestures, the acquaintance's voice, or a combination of gestures and the acquaintance's voice.

[0057] Figure 3 shows examples of processes 300 that can be performed by System 330 to identify and deploy AI personas that best influence driving behavior, according to various embodiments of the technology of this disclosure. In some embodiments, System 330 may be implemented in a vehicle 350. Vehicle 350 may be the same / similar vehicle to vehicle 100 described with Figure 1.

[0058] As shown in the figure, system 330 can perform action 302 to determine a target driving behavior for the driver of vehicle 350 based on the driving conditions. System 330 can perform this action in the same / similar manner as described together with action 202 in Figure 2.

[0059] Subsequently, system 330 can perform action 304 to identify a persona for a visual avatar that has the highest predicted probability of influencing the driver to perform a target driving behavior. System 330 can perform this action in the same / similar manner as described together with action 204 in Figure 2.

[0060] Subsequently, system 330 can perform action 306, which involves displaying a visual avatar of the identified persona to the driver via an AR device, so as to present information that can influence the driver to perform the target driving behavior. System 330 can perform this action in the same / similar manner as described alongside action 206 in Figure 2.

[0061] Figure 4 shows examples of AI personas according to various embodiments of the technology of this disclosure.

[0062] As shown in the diagram, AI Persona 412 can be implemented in vehicle 410. Similarly, AI Persona 422 can be implemented in vehicle 420.

[0063] The AI ​​persona 412 may be based, for example, on the family members of the driver of vehicle 410. As shown in the figure, the AI ​​persona 412 may have an angry emotional state. As described above, the system of the technology of this disclosure may select the AI ​​persona 412 (including the angry emotional state for the AI ​​persona 412) in order to influence the driver of vehicle 410 to perform a target driving behavior.

[0064] AI persona 422 may be based, for example, on the occupants of vehicle 410 (e.g., a young child passenger) (where vehicle 410 may have transferred AI persona 422 to vehicle 420 when vehicle 410 and vehicle 420 are traveling together on the road). As shown in the figure, AI persona 422 may have a sleepy emotional / cognitive state. As described above, the system of the technology of this disclosure may select AI persona 422 to influence the driver of vehicle 420 to perform a target driving behavior (including the sleepy emotional / cognitive state for AI persona 422).

[0065] Figure 5 shows further examples of AI personas according to various embodiments of the technology of this disclosure.

[0066] As shown in the figure, two AI personas (i.e., Judy and Sue) may be implemented in the vehicle 510. The system of the technology of this disclosure may apply a generative language model to the two AI personas to generate a conversation 512 between the two personas commenting on a specific driving situation. As described above, the system of the technology of this disclosure may have two AI personas and generate a conversation between them to influence the driver of the vehicle 510 to perform a target driving behavior.

[0067] In a specific example, Judy may be the actual driver of vehicle 510 (i.e., a non-AI persona), while Sue may be an AI persona. Therefore, the system may apply a generative language model to participate in a conversation 512 with Judy. Sue's voice quality and vocabulary selection may be chosen by the system to influence Judy and lead her to a target driving behavior.

[0068] Figure 6 shows other examples of AI personas according to various embodiments of the technology of this disclosure.

[0069] As shown in the diagram, AI Persona 612 can be implemented in vehicle 610. In a specific implementation, the visual avatar of AI Persona 612 may appear in the rear seat of vehicle 610.

[0070] The AI ​​persona 612 may be based on a young child in a vehicle near vehicle 610. The system of the technology of this disclosure may also use a generative model to have the AI ​​persona 612 converse with the driver of vehicle 610 via conversation 614.

[0071] As described above, the system of the technology of this disclosure may select an AI persona 612 and a conversation 614 to influence the driver of the vehicle 610 to perform a target driving behavior.

[0072] As used herein, the terms circuit and component may describe a given functional unit that is executable according to one or more embodiments of this application. As used herein, a component may be implemented using any form of hardware, software, or a combination thereof. For example, one or more processors, controllers, ASICs, PLAs, PALs, CPLDs, FPGAs, logic components, software routines, or other mechanisms may be implemented to constitute a component. The various components described herein may be implemented as individual components or as described functions, and features may be shared, in part or in whole, among one or more components. In other words, as will become apparent to those skilled in the art after reading this description, the various features and functionalities described herein can be implemented in any given application. They may be implemented in one or more separate or shared components in various combinations and permutations. While various features or functional elements may be described individually or claimed as separate components, it should be understood that such features / functionalities may be shared among one or more common software and hardware elements. Such descriptions do not require, or imply, that separate hardware or software components be used to implement such features or functionality.

[0073] When components are implemented entirely or partially using software, these software elements can be implemented to operate using computational or processing components capable of performing the functions described relating thereto. An example of such a computational component is shown in Figure 7. Various embodiments are described in relation to this exemplary computational component 700. After reading this description, methods for implementing the application using other computational components or architectures will be apparent to those skilled in the art.

[0074] Referring to Figure 7, the arithmetic component 700 can represent, for example, the arithmetic or processing power found in self-adjusting displays, desktops, laptops, notebooks, and tablet computers. They can also be found in handheld arithmetic devices (tablets, PDAs, smartphones, mobile phones, palmtops, etc.). They can also be found in workstations, or in other devices that have a display, server, or any other type of dedicated or general-purpose arithmetic device, as may be desirable or appropriate for a given application or environment. The arithmetic component 700 can also represent the arithmetic power embedded in a given device, or the arithmetic power available to a given device. For example, the arithmetic component can be found in other electronic devices such as portable arithmetic devices, and other electronic devices that may have some form of processing power.

[0075] The arithmetic component 700 may include, for example, one or more processors, controllers, control components, or other processing devices. This may include a processor and / or any one or more components that constitute a user device, a user system, and an undecoded cloud service. The processor 704 may be implemented using a general-purpose or dedicated processing engine, such as a microprocessor, controller, or other control logic. The processor 704 may be connected to the bus 702. However, any communication medium may be used to facilitate interaction between the arithmetic component 700 and other components, or to communicate with the outside.

[0076] The arithmetic component 700 may also include one or more storage components, which are referred to herein simply as the main memory 708. For example, random access memory (RAM) or other dynamic memory may be used to store information and instructions to be executed by the processor 704. The main memory 708 may also be used to store temporary variables or other intermediate information during the execution of instructions to be executed by the processor 704. The arithmetic component 700 may similarly include read-only memory (ROM) or other static storage device coupled to the bus 702 for storing static information and instructions for the processor 704.

[0077] The arithmetic component 700 may also include one or more different forms of information storage mechanisms 710, which may include, for example, a media drive 712 and a storage unit interface 720. The media drive 712 may include a drive or other mechanism for supporting a fixed or removable storage medium 714. For example, a hard disk drive, solid-state drive, magnetic tape drive, optical drive, compact disc (CD) or digital video disc (DVD) drive (R or RW), or other removable or fixed media drive may be provided. The storage medium 714 may include, for example, a hard disk, integrated circuit assembly, magnetic tape, cartridge, optical disc, CD or DVD. The storage medium 714 may be any other fixed or removable medium that is read, written to, or accessed by the media drive 712. As these examples show, the storage medium 714 may include computer-usable storage medium stored in computer software or data.

[0078] In alternative embodiments, the information storage mechanism 710 may include other similar devices for loading computer programs or other instructions or data into the arithmetic component 700. Such devices may include, for example, fixed or removable storage units 722 and interfaces 720. Examples of such storage units 722 and interfaces 720 may include a program cartridge and cartridge interface, a removable storage medium (e.g., flash memory or other removable storage components), and memory slots. Other examples may include PCMCIA slots and cards, as well as other fixed or removable storage units 722 and interfaces 720 that enable the transfer of software and data from the storage unit 722 to the arithmetic component 700.

[0079] The arithmetic component 700 may also include a communication interface 724. The communication interface 724 may be used to enable the transfer of software and data between the arithmetic component 700 and external devices. Examples of the communication interface 724 may include a modem or softmodem, a network interface (such as Ethernet, a network interface card, IEEE 802.XX, or other interfaces). Other examples include a communication port (e.g., a USB port, an IR port, an RS232 port, a Bluetooth® interface, or other ports), or other communication interfaces. The software / data transferred via the communication interface 724 may be carried by signals that may be electronic signals, electromagnetic (including optical) signals, or other signals that can be exchanged by a given communication interface 724. These signals may be provided to the communication interface 724 via a channel 728. The channel 728 carries the signals and may be implemented using a wired or wireless communication medium. Examples of channels may include telephone lines, cellular links, RF links, optical links, network interfaces, local area networks or wide area networks, and other wired or wireless communication channels.

[0080] In this document, the terms “computer program medium” and “computer-ready medium” are used generally to refer to temporary or non-temporary medium. Such medium may be, for example, the storage unit 708, the storage unit 720, the medium 714, and the channel 728. These and various other forms of computer program mediums or computer-ready mediums may be involved in transporting one or more sequences of one or more instructions to a processing device for execution. Such instructions embodied in a medium are generally referred to as “computer program code” or “computer program product” (which may be grouped in the form of a computer program or otherwise). When executed, such instructions may enable the arithmetic component 700 to perform features or functions of the present invention as discussed herein.

[0081] It should be understood that the various features, aspects, and functions described in one or more individual embodiments are not limited in their applicability to the specific embodiments in which they are described. Rather, they may apply individually or in various combinations to one or more other embodiments, regardless of whether such embodiments are described or whether such features are presented as part of an embodiment in which they are described. Accordingly, the scope and breadth of this application should not be limited by any of the exemplary embodiments described above.

[0082] The words, phrases, and variations thereof used in this document should be interpreted as open and not restrictive unless otherwise explicitly stated. As previously mentioned, the word “includes” should be interpreted as “includes but not restricts” or similar. The word “example” is used to provide illustrative cases of the items under discussion, not an exhaustive or restrictive list. The word “one” should be interpreted as “at least one,” “one or more” or similar. Furthermore, adjectives and similar terms such as “conventional,” “traditional,” “usual,” “standard,” and “known” should not be interpreted as limiting the items described to a given time, or to items available at a given time. Instead, they should be read as encompassing conventional, traditional, usual, or standard techniques that are currently available or may be known at any future time. Where this document refers to techniques that are obvious or known to those skilled in the art, such techniques encompass those that are currently obvious or known to those skilled in the art at any future time.

[0083] The presence of broader vocabulary and phrases such as “one or more,” “at least,” “but not limited to,” or other similar phrases in some instances should not be interpreted as meaning that narrower examples are intended or required when those broader phrases may not be present. The use of the term “component” does not mean that all aspects or functions described or claimed as part of that component are organized into a common package. In fact, any or all of the various aspects of a component may be combined into a single package, independently of control logic or other components, or they may be maintained separately and further distributed across multiple groups or packages or distributed across multiple locations.

[0084] Furthermore, the various embodiments described herein are described in relation to exemplary block diagrams, flowcharts, and other figures. As will become apparent to those skilled in the art after reading this document, the illustrated embodiments and their various alternatives can be implemented without being limited to the illustrated examples. For example, the block diagrams and their accompanying descriptions should not be construed as obligating a particular architecture or configuration.

Claims

1. It is a system, Augmented reality (AR) devices and One or more processing resources, A non-temporary computer-readable medium, which is coupled to one or more processing resources, and when executed by one or more processing resources, the system, To determine the target driving behavior for the vehicle driver based on the driving conditions. Identifying a persona for a visual avatar that has the highest predicted probability of influencing the driver to perform the target driving behavior, and A system comprising a non-temporary computer-readable medium having a stored command to display the visual avatar of the identified persona to the driver via the AR device, so as to present the driver with information to influence the driver to perform the target driving behavior.

2. Identifying the persona for the visual avatar that has the highest predicted probability of influencing the driver to perform the target driving behavior includes determining that the identified persona reduces the objective function the most among multiple personas, and The system according to claim 1, wherein the term of the objective cost function reflects the impact on the predicted probability that the driver will engage in the target driving behavior of displaying the visual avatar of each of the multiple personas.

3. The identified personas are: An acquaintance of the aforementioned driver, or In a traffic zone, a person inside a second vehicle in the vicinity of the aforementioned vehicle, The system according to claim 1, including the following:

4. Displaying the visual avatar of the identified persona in order to present the information to the driver is: I heard the voice of the aforementioned acquaintance, or The voice of the person inside the second vehicle, The system according to claim 3, comprising displaying the visual avatar of the identified persona in order to present the aforementioned information.

5. The system according to claim 1, wherein displaying the visual avatar of the identified persona includes displaying the visual avatar of the identified persona so that it appears in the passenger seat of the vehicle.

6. It is a method, Based on the driving conditions, determine the target driving behavior for the vehicle's driver, Identifying a persona for an artificial intelligence (AI) assistant that has the highest predicted probability of influencing the driver to perform the target driving behavior, A method comprising using the AI ​​assistant having the identified persona to present the driver with information that will influence the driver to perform the target driving behavior.

7. Identifying the persona of the AI ​​assistant that has the highest predicted probability of influencing the driver to perform the target driving behavior includes determining that the identified persona reduces the objective function the most among multiple personas, and The method according to claim 6, wherein the term of the objective cost function reflects the impact on the predicted probability of the driver performing the target driving behavior of using the AI ​​assistant having each of the multiple personas.

8. The method according to claim 6, wherein the identified persona includes at least one of a visual persona and a vocal persona.

9. The aforementioned visual persona is, The visual avatar of the driver's acquaintance, or In a traffic zone, a visual avatar of a person inside a second vehicle near the aforementioned vehicle, The method according to claim 8, including the method described in claim 8.

10. The aforementioned vocal persona is, The voice of an acquaintance of the aforementioned driver, or In a traffic zone, the voice of a person inside a second vehicle near the aforementioned vehicle, The method according to claim 8, including the method described in claim 8.

11. Using the AI ​​assistant having the identified persona to present the information to the driver is, The aforementioned acquaintance's visual avatar speaking in the acquaintance's voice, or The visual avatar of the person in the second vehicle speaking in the voice of the person in the second vehicle, The method according to claim 8, comprising using an augmented reality (AR) device to display to the driver.

12. Using the AR device to display the aforementioned visual avatar to the driver is, The method according to claim 11, comprising using the AR device to display the visual avatar so that it appears in the passenger seat of the vehicle.

13. The aforementioned method, The further includes specifying the time interval at which the AI ​​assistant presents the information to the driver with respect to the identified persona, which has the highest predicted probability of influencing the driver to perform the target driving behavior. The method according to claim 6, wherein using the AI ​​assistant having the identified persona to present the information to the driver includes using the AI ​​assistant having the identified persona to present the information to the driver at the identified time interval.

14. Identifying the time interval at which the AI ​​assistant presents the information to the driver with respect to the identified persona having the highest predicted probability of influencing the driver to perform the target driving behavior includes determining that the identified time interval and identified persona among multiple time intervals and multiple personas reduces the objective function the most. The first term of the objective cost function reflects the impact on the predicted probability of the driver performing the target driving behavior of using the AI ​​assistant, which has each of the multiple personas, The method according to claim 13, wherein the second term of the objective cost function reflects the impact on the predicted probability that the driver will perform the target driving action of presenting the information within each of the plurality of time intervals.

15. It is a vehicle, One or more processing resources, A non-temporary computer-readable medium, which is coupled to one or more processing resources, and when executed by one or more processing resources, in the vehicle, To determine the target driving behavior for the driver of the vehicle based on the driving conditions, Identifying a persona for an artificial intelligence (AI) assistant that has the highest predicted probability of influencing the driver to perform the target driving behavior, and A vehicle comprising a non-temporary computer-readable medium having stored instructions for causing the driver to use the AI ​​assistant having the identified persona to present the driver with information to influence the driver and cause him to perform the target driving behavior.

16. Identifying the persona of the AI ​​assistant that has the highest predicted probability of influencing the driver to perform the target driving behavior includes determining that the identified persona reduces the objective function the most among multiple personas, and The vehicle according to claim 15, wherein the term of the objective cost function reflects the impact on the predicted probability of the driver performing the target driving behavior of using the AI ​​assistant having each of the multiple personas.

17. The vehicle according to claim 15, wherein the identified persona includes at least one of a visual persona and a vocal persona.

18. The aforementioned visual persona is, The visual avatar of the driver's acquaintance, or In a traffic zone, a visual avatar of a person inside a second vehicle near the aforementioned vehicle, The vehicle according to claim 17, including the vehicle described in claim 17.

19. The aforementioned vocal persona is, The voice of an acquaintance of the aforementioned driver, or In a traffic zone, the voice of a person inside a second vehicle near the aforementioned vehicle, The vehicle according to claim 17, including the vehicle described in claim 17.

20. Using the AI ​​assistant having the identified persona to present the information to the driver is, The aforementioned acquaintance's visual avatar speaking in the acquaintance's voice, or The visual avatar of the person in the second vehicle speaking in the voice of the person in the second vehicle, The vehicle according to claim 17, comprising using an augmented reality (AR) device to display to the driver.

Images