Methods or systems that utilize human driving characteristics in autonomous driving methods or systems
By analyzing human driving data through machine learning systems, generating driving characteristic profiles, and adjusting the controllers of autonomous driving systems, the problem of autonomous driving systems being unable to simulate human driving characteristics is solved, thus improving the naturalness and comfort of the driving experience.
Patent Information
- Application Number
- CN202310005947.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2022-11-10
- Filing Date
- 2023-01-03
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2043-01-03
AI Technical Summary
Existing autonomous driving systems struggle to simulate the driving characteristics of human drivers, resulting in unnatural and uncomfortable operation in certain driving scenarios.
By analyzing human driver data through machine learning systems, driving characteristic profiles are generated and compared with the controller outputs of autonomous or semi-autonomous driving systems to adjust and tune the controllers to match human driving styles.
It improves the naturalness and comfort of the autonomous driving system in various driving scenarios, enhancing the driving experience.
Smart Images

Figure CN118004211B_ABST
Abstract
Description
Technical Field
[0001] This disclosure generally relates to the field of methods or systems that utilize human driving characteristics in autonomous vehicle driving methods or systems. Background Technology
[0002] Vehicles may include controllers or systems with autonomous driving capabilities. Summary of the Invention
[0003] Multiple variations may include methods or systems that utilize human driving characteristics in autonomous vehicle driving methods or systems.
[0004] Other illustrative variations within the scope of the invention will become apparent from the detailed description provided below. It should be understood that while the detailed description and specific examples disclose variations of the invention, they are intended for illustrative purposes only and are not intended to limit the scope of the invention. Attached Figure Description
[0005] From the detailed description and accompanying drawings, one will gain a more complete understanding of examples of variations within the scope of the invention, wherein:
[0006] Figure 1 Describe a method for using human driving characteristics in a method or system (autonomous vehicle driving method or system).
[0007] Figure 2 Illustrative variations of block diagrams depicting systems and methods that can be used to implement one or more of the functions and actions described herein; and
[0008] Figure 3 An illustrative variant of a vehicle is depicted, equipped with hardware sufficient to perform at least some of the systems and methods described herein. Detailed Implementation
[0009] The following description of the variations is illustrative in nature and is not intended to limit the scope, application or use of the invention.
[0010] Several variations may include using machine learning systems to adjust preset values or operational thresholds and limits of intelligent vehicle driving systems (such as, but not limited to, autonomous or semi-autonomous driving systems). Machine learning systems can be constructed and configured to analyze patterns in data using algorithms and statistical models and to derive inferences from those patterns, learning and adapting without following explicit instructions. In several variations, machine learning systems may include, but are not limited to, neural networks, artificial intelligence systems, or deep learning systems. In several variations, methods or systems may collect data on driving characteristics of a vehicle driven by a human in manual mode when the autonomous or semi-autonomous driving system is not active. While a human is driving the vehicle, signals or data are collected from one or more sensors located in the vehicle to gather inputs on vehicle speed, acceleration, deceleration, pitch, roll, yaw, and their rates of change. The machine learning system may be included in the vehicle or may be located remotely from the vehicle. Signals or data may be sent directly to the machine learning system in real time, or may be stored onboard in the vehicle and sent to or received by the machine learning system at a later date. Collected signals or data relating to vehicle speed, acceleration, deceleration, pitch, roll, yaw, and their rates of change can be sent to or received by a machine learning system to derive the driving characteristics of a general human driver or a specific human driver. These human driver characteristics can be compared with output commands from various types of lateral and longitudinal controllers of the vehicle. In several variations, scoring or weighting criteria can be applied to adjust, tune, or select vehicle controllers. In several variations, the human driver characteristics derived by the machine learning system can be used to adjust preset values or operational thresholds and limits of the intelligent vehicle driving system within the ultimate safe limits of its operation.
[0011] In several variations, the intelligent learning system takes input from manual driving and output from pre-defined traditional vehicle controllers, along with environmental data such as speed, comfort, driving profile, space, terrain, and so on. The machine learning system then compares the outputs of various lateral and longitudinal controllers with the input from manual driving and determines which controllers best match manual driving. This can be repeated across multiple vehicles for various driving scenarios via V2X (Vehicle-to-Instrument) to generate a database for further evaluation and tuning of controller performance to make the output more human-like. The machine learning system can also generate weighting factors for metrics that influence various requirements such as comfort and performance.
[0012] When the intelligent vehicle driving system is not in operation, preset values, thresholds, and limits of the intelligent vehicle driving system can be adjusted. For example, adjustments can be made when the intelligent driving system is in sleep mode, when it is not in operation, when the vehicle is only in manual driving mode, or when the vehicle is turned off.
[0013] Each of the intelligent vehicle driving system and the machine learning system may include a non-transitory computer medium having instructions thereon, which can be executed by an electronic processor to cause a set of actions or functions to be performed.
[0014] In several variants, the machine learning system receives input from sensors and output from pre-defined controllers while a human driver is manually driving the vehicle, along with environmental data such as speed, comfort, driving profile, space, or terrain. The machine learning system then compares the outputs of various lateral and longitudinal controllers of the vehicle with the input from the human driver's manual driving, determines which controllers best match the input, and uses the best-matched controllers to intelligently control the vehicle without manual input. This can be repeated across multiple vehicles for various driving scenarios via V2X to generate a database for further evaluation and tuning of controller performance to make the output more human-like.
[0015] Figure 1 According to several variations of the method, the method may include the following actions: sending data from at least one vehicle sensor of a vehicle to be driven by a human driver to a machine learning system 10; analyzing patterns in the data and making inferences from the patterns in the data 12; deriving a human driver profile from the data and the derived inferences 14; comparing the human driver profile with output commands from at least one of the vehicle's autonomous or semi-autonomous driving systems, either a lateral or longitudinal controller, to determine a difference 16; and, in response to the determined difference, providing an output that can be used to adjust at least one of the vehicle's lateral or longitudinal controllers, tune at least one of the vehicle's lateral or longitudinal controllers, or select a controller 18 that has an output that most closely matches the human driver profile.
[0016] Figure 2Illustrative variations of block diagrams depicting systems and methods that can be used to implement the functions and actions described herein. A vehicle may include a controller 112, which may include: a non-transitory computer-readable medium having instructions stored thereon, and an electronic processor for executing the instructions to implement the functions and actions described herein. In several variations, the system may include only one controller 112, or the system may include multiple controllers, each controller having: a non-transitory computer-readable medium having instructions stored thereon, and an electronic processor for executing the instructions to implement at least one of the functions and actions described herein. The vehicle may include a controller 112 configured and arranged to receive driver steering input 134 via a steering system 114. The controller 112 may additionally be configured and arranged to provide steering actuator commands 126 to the steering system 114. The steering system 114 may output tire angle changes 118 affecting the steering system health state 132 to the controller 112. The controller 112 may also be configured and arranged to provide a braking command 128 to the electric braking system 116, which in turn may apply braking pressure 120 to individual brake calipers. If the sensors and / or steering system 114 have indicated to the controller 112 that evasive maneuvering is in progress, the controller 112 may send a braking motion request to provide differential braking at all wheels, thereby increasing the vehicle's yaw rate. If the steering system 114 indicates that power steering assistance has failed, the controller 112 may receive driver input 134 via the steering wheel and translate the steering request into a braking pressure request or command 128 to be transmitted to the electric braking system 116. The controller 112 may also receive input 271 from various devices 270 designed to measure vehicle state information, including but not limited to lateral acceleration, yaw rate, and wheel speed. The controller may receive input 281 from various devices 280, which may include, but are not limited to, GPS, cameras, lidar, and radar, which may be used in algorithms to estimate various vehicle states. Estimated vehicle states can be helpful, for example, but not limited to, when steering wheel angle, torque, and speed sensors are unavailable. A longitudinal dynamic controller 260 is provided to send torque requests to accelerate or decelerate the front and / or rear wheels. The longitudinal dynamic controller 260 can receive input 261 from controller 112 and can send output 262 to controller 112. In several variations, the propulsion system may include individually controlled electric motors to provide differential drive force to each wheel. Controller 140 collects driver attributes via interface 141 and sends this data to an external server via link 143 to analyze and update the profile of a specific driver. The updated configuration is then returned to the vehicle via interface 151 with controller 140.At the appropriate time (at night or upon driver request), controller 140 will update the profile in controller 112 via interface 142.
[0017] Figure 3 This illustration depicts a portion of a vehicle equipped with hardware sufficient to perform at least some of the systems and methods described herein. Vehicle 250 may include a controller 212 constructed within vehicle 250. Controller 212 may be operatively communicateable with steering system 214 and electric braking system 216. Steering system 214 and electric braking system 216 may be operatively communicateable with at least one wheel 242. A driver may utilize handwheel 244 to provide driver input 134 for lateral movement and to send steering requests to steering system 214. In some variations, steering assist 246 associated with steering interface 244 may be operatively communicateable with controller 212, steering system 214, or electric braking system 216. The controller may receive input 281 from various devices 280, which may include, but are not limited to, GPS, cameras, lidar, and radar, which may be used in algorithms to estimate various vehicle states. The estimated vehicle states may be helpful, for example, but not limited to, when steering wheel angle, torque, and speed sensors are unavailable. The controller 212 can receive inputs and send outputs to the propulsion system 218.
[0018] Variant 1 may include a method for using human driving characteristics in an autonomous or semi-autonomous vehicle, the method comprising: sending data from at least one vehicle sensor of the vehicle to be driven by a human driver to a machine learning system; wherein the machine learning system includes a non-transitory computer-readable medium having instructions stored thereon, the instructions being executable by an electronic processor, causing the machine learning system to perform the following actions: analyzing patterns in the data and drawing inferences from the patterns in the data; deriving a human driver characteristic profile from the data and the derived inferences; comparing the human driver characteristic profile with output commands from at least one of a vehicle lateral controller or a longitudinal controller of the vehicle's autonomous or semi-autonomous driving system to determine a difference; and, in response to the determined difference, providing an output that can be used to adjust at least one of the vehicle lateral controller or the longitudinal controller, tune at least one of the vehicle lateral controller or the longitudinal controller, or select a controller having an output that most closely matches the human driver profile.
[0019] Variation 2 may include the method as described in Variation 1, wherein deriving a human driver characteristic profile includes scoring or weighting the data.
[0020] Variation 3 may include the method as described in any one of Variations 1-2, wherein the human driver characteristic profile is within a preset threshold and limit of the output command of at least one of the vehicle lateral controllers or longitudinal controllers.
[0021] Variation 4 may include the method as described in any one of Variations 1-3, wherein at least one of the vehicle lateral controller or longitudinal controller is adjusted, or at least one of the vehicle lateral controller or longitudinal controller is tuned within external safety limits for operating an autonomous or semi-autonomous vehicle.
[0022] Variation 5 may include the method as described in any one of Variations 1-4, wherein the data is associated with at least one of the rates of change of vehicle acceleration, deceleration, pitch, roll, yaw, or one or more of these.
[0023] Variation 6 may include the method as described in any one of Variations 1-5, wherein the data is related to at least one of comfort, driving profile, space, or terrain.
[0024] Variation 7 may include the method as described in any one of Variations 1-6, further including using the output to adjust at least one of the vehicle lateral controller or longitudinal controller, tune at least one of the vehicle lateral controller or longitudinal controller, or select a controller having an output that most closely matches a human driver profile.
[0025] Variation 8 may include the method as described in any one of Variations 1-7, and further includes operating the vehicle in autonomous or semi-autonomous mode.
[0026] Variant 9 may include a system comprising: a non-transitory computer-readable medium having instructions stored thereon, and a processor for executing the instructions to perform actions using human driving characteristics in an autonomous or semi-autonomous vehicle, the actions including: receiving data from at least one vehicle sensor of a vehicle driven by a human driver; analyzing patterns in the data and making inferences from the patterns in the data; deriving a human driver characteristic profile from the data and the derived inferences; comparing the human driver characteristic profile with output commands from at least one of a vehicle lateral controller or a longitudinal controller of the vehicle's autonomous or semi-autonomous driving system to determine a difference; and providing an output in response to the determined difference, the output being available for adjusting at least one of the vehicle lateral controller or the longitudinal controller, tuning at least one of the vehicle lateral controller or the longitudinal controller, or selecting a controller having an output that most closely matches the human driver profile.
[0027] Variation 10 may include a system as described in Variation 9, wherein deriving a human driver characteristic profile includes scoring or weighting the data.
[0028] Variation 11 may include the system as described in any one of Variations 9-10, wherein the human driver characteristic profile is within preset thresholds and limits of the output commands of at least one of the vehicle lateral controllers or longitudinal controllers.
[0029] Variation 12 may include the system as described in any one of Variations 9-11, wherein at least one of the vehicle lateral controllers or longitudinal controllers is adjusted, or at least one of the vehicle lateral controllers or longitudinal controllers is tuned, within external safety limits for operating an autonomous or semi-autonomous vehicle.
[0030] Variation 13 may include the system as described in any one of Variations 9-12, wherein the data is associated with at least one of the rates of change of vehicle acceleration, deceleration, pitch, roll, yaw, or one or more of these.
[0031] Variation 14 may include a system as described in any one of Variations 9-13, wherein the data is related to at least one of comfort, driving profile, space, or terrain.
[0032] Variation 15 may include the system as described in any one of Variations 9-14, further comprising generating an output to regulate at least one of the vehicle lateral controller or longitudinal controller, tuning at least one of the vehicle lateral controller or longitudinal controller, or selecting a controller having an output that most closely matches a human driver profile.
[0033] Variant 16 may include a non-transitory computer-readable medium having instructions stored thereon that can be executed by an electronic processor to perform actions using human driving characteristics in an autonomous or semi-autonomous vehicle, the actions including: receiving data from at least one vehicle sensor of a vehicle driven by a human driver; analyzing patterns in the data and making inferences from the patterns in the data; deriving a human driver characteristic profile from the data and the derived inferences; comparing the human driver characteristic profile with output commands from at least one of the vehicle lateral controllers or longitudinal controllers of the vehicle's autonomous or semi-autonomous driving system to determine a difference; and providing an output in response to the determined difference, the output being available for adjusting at least one of the vehicle lateral controllers or longitudinal controllers, tuning at least one of the vehicle lateral controllers or longitudinal controllers, or selecting a controller having an output that most closely matches the human driver profile.
[0034] Variation 17 may include the medium described in Variation 16, wherein deriving a human driver characteristic profile includes scoring or weighting the data.
[0035] Variation 18 may include the medium as described in Variation 16 or 17, wherein the human driver characteristic profile is within preset thresholds and limits of the output commands of at least one of the vehicle's lateral or longitudinal controllers.
[0036] Variation 19 may include a medium as described in any one of Variations 16-18, wherein at least one of the vehicle lateral controllers or longitudinal controllers is adjusted or tuned within an external safety zone.
[0037] Variation 20 may include a medium as described in any one of Variations 16-19, wherein the data is associated with at least one of the rates of change of vehicle acceleration, deceleration, pitch, roll, yaw, or one or more of these.
[0038] The above description of the selected variations within the scope of this invention is illustrative in nature only, and therefore, its variations or variants are not considered to depart from the spirit and scope of the invention.
Claims
1. A method for using human driving characteristics in an autonomous or semi-autonomous vehicle, the method comprising: In the future, data from at least one vehicle sensor in a vehicle driven by a free human driver will be sent to a machine learning system. The machine learning system includes a non-transitory computer-readable medium having instructions stored thereon, the instructions being executable by an electronic processor, causing the machine learning system to perform the following actions: Analyze the patterns in the data and draw inferences from the patterns in the data; A profile of human driver characteristics is derived from the data and the inferences drawn. The human driver profile is compared with the output command of at least one of the vehicle's autonomous or semi-autonomous driving system's lateral or longitudinal controllers to determine the differences. In response to the determined differences, an output is provided that can be used to adjust at least one of the vehicle lateral controller or longitudinal controller, tune at least one of the vehicle lateral controller or longitudinal controller, or select the controller with the output that most closely matches the human driver profile.
2. The method of claim 1, wherein, The process of deriving a human driver profile includes scoring or weighting the data.
3. The method of claim 1, wherein, The human driver characteristic profile is within preset thresholds and limits of the output commands of at least one of the vehicle's lateral or longitudinal controllers.
4. The method of claim 1, wherein, Adjust at least one of the vehicle lateral controllers or longitudinal controllers, or tune at least one of the vehicle lateral controllers or longitudinal controllers, within the external safety limits for operating the autonomous or semi-autonomous vehicle.
5. The method of claim 1, wherein, The data is related to at least one of the rates of change of vehicle acceleration, deceleration, pitch, roll, yaw, or one or more of these.
6. The method of claim 1, wherein, The data is related to at least one of comfort, driving profile, space, or terrain.
7. The method of claim 1, further comprising using the output to adjust at least one of the vehicle lateral controller or longitudinal controller, tuning at least one of the vehicle lateral controller or longitudinal controller, or selecting a controller having an output that most closely matches the human driver profile.
8. The method of claim 7, further comprising operating the vehicle in an autonomous or semi-autonomous mode.
9. A system comprising: A non-transitory computer-readable medium having instructions stored thereon, and a processor for executing the instructions to perform actions using human driving characteristics in an autonomous or semi-autonomous vehicle, the actions including: Receive data from at least one vehicle sensor in a vehicle driven by a human driver; Analyze the patterns in the data and draw inferences from the patterns in the data; A profile of human driver characteristics is derived from the data and the inferences drawn. The human driver profile is compared with the output command of at least one of the vehicle's autonomous or semi-autonomous driving system's lateral or longitudinal controllers to determine the differences. In response to the determined differences, an output is provided that can be used to adjust at least one of the vehicle lateral controller or longitudinal controller, tune at least one of the vehicle lateral controller or longitudinal controller, or select the controller with the output that most closely matches the human driver profile.
10. The system of claim 9, wherein, The process of deriving a human driver profile includes scoring or weighting the data.
11. The system of claim 9, wherein, The human driver characteristic profile is within preset thresholds and limits of the output commands of at least one of the vehicle's lateral or longitudinal controllers.
12. The system according to claim 9, wherein, Adjust at least one of the vehicle lateral controllers or longitudinal controllers, or tune at least one of the vehicle lateral controllers or longitudinal controllers, within the external safety limits for operating the autonomous or semi-autonomous vehicle.
13. The system according to claim 9, wherein, The data is related to at least one of the rates of change of vehicle acceleration, deceleration, pitch, roll, yaw, or one or more of these.
14. The system according to claim 9, wherein, The data is related to at least one of comfort, driving profile, space, or terrain.
15. The system of claim 9, further comprising generating an output to adjust at least one of the vehicle lateral controller or longitudinal controller, tuning at least one of the vehicle lateral controller or longitudinal controller, or selecting a controller having an output that most closely matches the human driver profile.
16. A non-transitory computer-readable medium having instructions stored thereon, the instructions being executable by an electronic processor to perform actions using human driving characteristics in an autonomous or semi-autonomous vehicle, the actions comprising: Receive data from at least one vehicle sensor in a vehicle driven by a human driver; Analyze the patterns in the data and draw inferences from the patterns in the data; A profile of human driver characteristics is derived from the data and the inferences drawn. The human driver profile is compared with the output command of at least one of the vehicle's autonomous or semi-autonomous driving system's lateral or longitudinal controllers to determine the differences. In response to the determined differences, an output is provided that can be used to adjust at least one of the vehicle lateral controller or longitudinal controller, tune at least one of the vehicle lateral controller or longitudinal controller, or select the controller with the output that most closely matches the human driver profile.
17. The medium according to claim 16, wherein, The process of deriving a human driver profile includes scoring or weighting the data.
18. The medium according to claim 16, wherein, The human driver characteristic profile is within preset thresholds and limits of the output commands of at least one of the vehicle's lateral or longitudinal controllers.
19. The medium according to claim 16, wherein, Adjusting at least one of the vehicle lateral controllers or longitudinal controllers, or tuning at least one of the vehicle lateral controllers or longitudinal controllers, within the external safety zone.
20. The medium according to claim 16, wherein, The data is related to at least one of the rates of change of vehicle acceleration, deceleration, pitch, roll, yaw, or one or more of these.
Citation Information
Patent Citations
Control of autonomous vehicle based on passenger and environment aware driving style profile
CN111434554A
Self-adaptive driver car-following method comprehensively considering perception process and driver behavior
CN115056776A
