Maneuver prediction-based speed profile adaptation for behavior-based automated speed control

Abstract

A maneuver-predictive speed profile adaptation system for behavior-based automated cruise control in a host vehicle comprises sensors and actuators connected to a controller. The controller executes a maneuver-predictive speed profile adaptation (MPSPA) application that detects static and dynamic information about the host vehicle and its environment. The MPSPA application processes sensor information, performing feature extraction, synchronization, normalization, generating a probabilistic path prediction, and detecting driver intent.The MPSA application generates a control command to the actuators to selectively change at least one longitudinal speed of the host vehicle in order to approximate historical preferences of the driver regarding lateral and longitudinal speeds, yaw, and forces observed by the driver while the host vehicle is driven in a fully autonomous and / or semi-autonomous mode.

Description

introduction

[0001] The present disclosure relates to cruise control or speed control systems in vehicles and, more specifically, to advanced driver assistance systems that perform speed control functions in vehicles.

[0002] Vehicles equipped with advanced driver assistance systems (ADAS) use a variety of sensors to detect the vehicle's environment and actuators to adjust or adapt the vehicle's performance in relation to its surroundings, including through the use of adaptive cruise control (ACC).

[0003] While current ADAS-based ACC systems and procedures fulfill their intended purpose, there is a need for a new and improved system and procedure for ADAS-based ACC that incorporates maneuver prediction and speed adaptation to ensure that vehicle performance, using ADAS and ACC, adapts adaptively to the preferences inputted by a driver, thereby increasing driver comfort while maintaining the efficiency of ADAS and ACC, improving the accuracy and responsiveness of ACC, resulting in a smoother and more intelligent driving experience, while maintaining or reducing system complexity and increasing redundancy.

[0004] DE 10 2023 128 248 A1 discloses a method that includes determining a combined probability for a potential maneuver based on partial probabilities corresponding to various vehicle state variables and environmental conditions, using one or more tuning parameters specific to the respective combination of maneuver and vehicle state variable. The method includes determining an adjusted probability for the potential maneuver, at least partially, based on the combined probability and one or more context-related state variables corresponding to the current operating context for the vehicle, adapted to the driver's behavior and driving style.Furthermore, the procedure includes identifying the potential maneuver as an expected maneuver to be performed by a driver of the vehicle based on the adjusted probability, and automatically initiating one or more actions in a manner influenced by the expected maneuver.

[0005] DE 10 2005 043 838 A1 discloses a method for the automatic detection of vehicle driving maneuvers using detected state variables. The method comprises defining a Bayesian network formed from nodes and edges to describe the probability of certain driving maneuvers occurring as a given dependency on state variables, in which nodes initialized with a local probability distribution represent a state variable and the edges represent causal dependencies between the nodes. Furthermore, the method comprises calculating the probability of the occurrence of driving maneuvers using the Bayesian network by inputting detected or derived state variables into the Bayesian network.

[0006] EP 3 752 401 B1 discloses a method for selecting a driving profile for a motor vehicle, a driver assistance system for this purpose, and a motor vehicle equipped therewith. Based on an operation performed by the driver, a deviation between the driving profile currently desired by the driver and the previously used profile from several predefined options is detected. After the deviation is detected, a prompt is sent to the driver asking whether the deviation should be learned by the driver assistance system for the current situation. After the driver confirms the prompt, the driver assistance system is adjusted accordingly for the current situation in at least one parameter influencing the selection of the driving profile to be used, such that the deviation is taken into account when automatically selecting the driving profile to be used in future situations corresponding to the current situation. Summary

[0007] From several perspectives, a maneuver-predictive speed profile adaptation (MPSPA) system for behavior-based automated cruise control in a host vehicle comprises a host vehicle, one or more sensors, and one or more actuators. The one or more sensors detect static and dynamic state information about the host vehicle, and the one or more actuators modify the static and dynamic performance of the host vehicle. The system further includes a controller comprising a processor, memory, and input / output (I / O) ports. The I / O ports communicate with the one or more sensors and the one or more actuators. The processor executes program code stored in memory. This program code includes a maneuver-predictive speed profile adaptation (MPSPA) application.(maneuver prediction speed profile adaptation) with at least a first, a second, a third, and a fourth control logic. The first control logic acquires static and dynamic information about the host vehicle and its surrounding environment. The second control logic processes this static and dynamic information. This processing includes at least feature extraction, synchronization, normalization, and the generation of a probabilistic path prediction. The third control logic detects the intention of the driver. The fourth control logic generates a control command to one or more actuators.The control command selectively modifies at least one longitudinal speed of the host vehicle to approximate the driver's historical preferences regarding lateral and longitudinal speeds, yaw, and forces observed by the driver while the host vehicle is operating in a fully autonomous and / or semi-autonomous mode. The second control logic further includes control logic for filtering and performing feature extraction on data from the one or more sensors. The filtering causes the system and the MPSPA application to reduce the use of computing resources from a first level to a second level that is lower than the first, by selecting only specific portions of the data from the one or more sensors that can indicate a state change of the host vehicle. The feature extraction further determines that the state change of the host vehicle has a certain magnitude.of a magnitude that requires additional processing. The second control logic further includes control logic for synchronizing the data from the one or more sensors. Synchronizing the data ensures that the data from the one or more sensors are precisely time-aligned and associated, so that data from each of the one or more sensors is temporally correlated with data from each of the other sensors. The second control logic further includes control logic for generating a probabilistic path prediction using a probability function for each state variable that follows a sigmoid function, where the host vehicle's state variables include: the velocity, ν. x , of the host vehicle, the steering angle, δ, of the host vehicle, the torsion bar torque, τ, and the torsion bar torque rate, τ̇, and PM,x(k)=11+e−βM,x(|x|−αM,x) where x is a variable ν x , δ, τ, τ̇ is and M is the maneuver type, which includes cornering or turning, changing lanes, U-turns, evasive maneuvers and / or staying in lane. A maneuver probability is obtained by multiplying the state probability functions as follows: PM(k)=gd∏x=1nPM,x(k) where P M,x where x is a probability function of the variable for the maneuver M, k is a sampling time, β M,x a tuning parameter, the sigmoid slope of the variable x for the maneuver M, is α M,x a tuning parameter, the sigmoid center of the variable x for the maneuver M, n is a number or set of variables, and g d one direction of the maneuver M is such that for a maneuver to the right g d = -1 and for a maneuver to the left g d= 1. The third control logic further includes a control logic for utilizing a machine learning (ML) model for data-driven classification to integrate a shadow planner that enhances adaptive cruise control (ACC) by predicting the driver's intended path and adjusting a speed profile of the host vehicle to match the driver's intention, based on an analysis of driver inputs to the host vehicle's control systems and the behavior of surrounding vehicles. The third control logic also includes a control logic for performing real-time path classification to evaluate the type of maneuver actively pursued by the driver and the host vehicle's ADAS, as well as a control logic for performing an enumeration calculation that qualifies classified real-time paths as a left turn and / or a right turn.

[0008] In another aspect of the present disclosure, the first control logic further comprises a control logic for obtaining information about the position, motion, and acceleration of the host vehicle in at least three degrees of freedom from an inertial measurement unit (IMU). The first control logic further comprises a control logic for obtaining a steering angle of the host vehicle from one or more steering angle sensors (SAS), a control logic for obtaining an accelerator pedal position from an accelerator pedal position sensor, and a control logic for obtaining a brake pedal position from a brake pedal position sensor.

[0009] In another aspect of the present disclosure, the second control logic further comprises a control logic for integrating environmental knowledge into environmental knowledge categories, which include: road information, road type, the number or quantity of lanes, the presence / absence of construction sites, information on surrounding objects, turn or direction signals for a closest-in-vehicle path (CIPV), and navigation information. An effect of each environmental knowledge category is represented by a gain, with the multiplication of all gains of environmental knowledge categories defining a final environmental gain, which is expressed as follows: gM,e=gM,r×gM,s×gM,n where g M, r a probability enhancement of road information for the maneuver M is, g M,sa probability increase from surrounding objects for the maneuver M is, g M,n a probability enhancement of navigation information for the maneuver M is and g M, e This is a probability enhancement for maneuver M based on environmental knowledge. The following equation expresses a maneuver probability function taking environmental knowledge into account: PM(k)=gM,e×gd∏x=1nPM,x(k).

[0010] In another aspect of the present disclosure, the second control logic further includes control logic for normalizing the sensor data. Normalizing the sensor data reduces redundant information in the sensor data from a first redundancy level to a second redundancy level that is lower than the first. The second control logic further includes control logic for generating a confidence calculation. A result or output of the confidence calculation defines a probability that a certain type of maneuver M will be executed by the host vehicle.

[0011] In another aspect of the present disclosure, the third control logic further comprises a control logic for generating a path or trajectory prediction based on real-time paths, data from a global positioning system (GPS) and navigation information, including a selected navigation route.

[0012] In another aspect of the present disclosure, the fourth control logic further comprises control logic for generating a control command to one or more actuators. The control command comprises a predicted lateral command and a predicted longitudinal command. The fourth control logic further comprises control logic for activating and controlling one or more actuators of a transmission control system, an engine control system, and a brake control system of a host vehicle to change at least one longitudinal velocity of the host vehicle while an actual classified real-time path maneuver M is executed according to the predicted lateral command and the predicted longitudinal command.

[0013] Further areas of application will become apparent from the description provided herein. It should be understood that the description and specific examples are intended for illustrative purposes only and are not meant to limit the scope of this disclosure. Brief description of the drawings

[0014] The drawings described herein are for illustrative purposes only and are not intended to limit the scope of the present disclosure in any way. Fig. Figure 1 is a schematic diagram of a maneuver prediction-based speed profile adaptation system for behavior-based automated speed control according to one aspect of the present disclosure; Fig. Figure 2 is a perspective top view of a road intersection showing a host vehicle using the system of Fig. 1 according to one aspect of the present revelation, takes a curve; Fig. Figure 3 is a flowchart that illustrates the logical sequence of an application for Maneuver Prediction Speed ​​Profile Matching (MPSPA) of the system of Fig. 1 represents one aspect of the present revelation; Fig. 4A is a first graphical representation of lateral and longitudinal commands to actuators of a host vehicle, which follows a curve using the MPSPA application of Fig. 3 according to one aspect of the present revelation; and Fig. 4B is a second graphical representation of lateral and longitudinal commands to actuators of a host vehicle, which follows a curve using the MPSPA application of Fig. 3 according to another aspect of the present revelation. Detailed description

[0015] The following description is merely exemplary and is not intended to limit the present disclosure, application or uses.

[0016] Referring to Fig. Figure 1 describes a system 10 for maneuver prediction-based speed profile adaptation for behavior-based adaptive cruise control (ACC) 11. The system 10 includes a vehicle 12. The host vehicle 12 is illustrated as a passenger car; however, it should be understood that the host vehicle 12 can be any type of vehicle, including but not limited to cars, trucks, SUVs, vans, motorhomes, tractor units, semi-trailers, delivery vehicles including vehicles used in warehouses, tricycles, motorcycles, aircraft, amphibious vehicles, or any other such vehicle 12. Furthermore, the host vehicle 12 can be an aircraft, a watercraft, or the like, without deviating from the scope or intent of this disclosure.

[0017] The system 10 further comprises one or more sensors 14 arranged on, attached to, or otherwise integrated into the host vehicle 12. Additional sensors 14 may be located remotely from the host vehicle 12 and transmit information to the host vehicle 12, as described in more detail below. The sensors 14 of the host vehicle 12 may comprise any of a wide variety of sensor types, including, but not limited to, electromagnetic (EM) sensors 14 such as cameras, infrared cameras, video cameras, light-based detection and distance (LiDAR) sensors, radio-based detection and distance (RADAR) sensors, sound-based navigation and distance (SO-NAR) sensors, and the like.In some examples, cameras and / or other sensors 14 of the host vehicle 12 are mounted with outward-facing fields of view (FOVs), and data collected by such cameras includes optical information about an environment in which the host vehicle 12 operates. In additional, non-restrictive examples, the cameras and / or other sensors 14 are directed toward an interior or passenger compartment of the host vehicle 12, thereby providing information about the occupants of the host vehicle 12 and the driver of the host vehicle 12. Additional sensors 14 may, without restriction, include: inertial measurement units (IMUs) 16, suspension control units such as sensors 18 of a semi-active damping suspension (SADS), sensors 20 of a global positioning system (GPS), wheel speed sensors 24 that measure the rotational speeds, respectively.Speeds of one or more wheels 26 of the host vehicle 12 can be measured, throttle valve and / or accelerator pedal position sensors 28, brake pedal position sensors 30, steering position sensors 32 that can measure the position, steering rate and steering speed of a steering system 34, tire pressure monitoring systems 36 and the like.

[0018] The IMUs 16 can measure the motion, acceleration, and other parameters of the host vehicle 12 in multiple degrees of freedom. In a specific example, the IMUs 16 can measure position, motion, acceleration, etc., in at least three degrees of freedom. Similarly, the SADS sensors 18 can be IMUs 16 capable of measuring in three or more degrees of freedom. In some examples, the SADS 18 can be suspension hub accelerometers or similar devices. The sensors 14 of the host vehicle 12 can therefore detect and record wheel speed data, the position and location of the host vehicle 12, and static and dynamic state information of the host vehicle 12, such as speed, acceleration, and the like.

[0019] As used herein, the terms “front”, “rear”, “inner”, “inward”, “outer”, “outward”, “above”, and “below” are terms used in relation to the orientation of the host vehicle 12, as shown in the drawings of the present application. Thus, “front” refers to a direction towards a front of a vehicle 12, “rear” refers to a direction towards a rear of a vehicle 12, “inner” and “inward” refer to a direction towards an interior or passenger compartment 38 of a vehicle 12, “outer” and “outward” refer to a direction towards an exterior of a vehicle 12, “below” refers to a direction towards a floor of the host vehicle 12, and “above” refers to a direction towards a top of the host vehicle 12.

[0020] The system 10 further comprises one or more controllers 40 that communicate with the various sensors 14 of the host vehicle 12, process information received from them, and generate output signals used to assist the driver 42 in maintaining attention and avoiding highway hypnosis or white-line fever. The controllers 40 are integrated into the host vehicle 12. More precisely, the controllers 40 are non-generalized electronic control devices comprising a pre-programmed digital computer or processor 44, a non-transient computer-readable medium or memory 46 used to store data such as control logic, software applications, instructions, computer code, data lookup tables, etc., and input / output ports (I / O ports) 48.Computer-readable medium or computer-readable memory 46 includes any type of medium accessible to a computer, such as read-only memory (ROM), random-access memory (RAM), a hard disk, a compact disc (CD), a digital video disc (DVD), or any other type of storage. “Non-transient” computer-readable memory 46 excludes wired, wireless, optical, or other communication links that transmit transient electrical or other signals. Non-transient computer-readable memory 46 includes media on which data can be permanently stored and media on which data can be stored and subsequently overwritten, such as a rewritable optical disc or an erasable storage device. Computer code includes any type of program code, including source code, object code, and executable code.The processor 44 is configured to execute the code or instructions. The host vehicle 12 may have additional controllers 40, such as a dedicated Wi-Fi controller, an engine control module, a transmission control module, a body control module, an infotainment control module, or the like. The I / O ports 48 may be configured to communicate via wired communication, wirelessly via Wi-Fi protocols according to IEEE 802.11x, or the like, without deviating from the scope or intent of this disclosure.

[0021] The controller 40 further comprises one or more applications 50. An application 50 is a software program configured to perform a specific function or set of functions. The application 50 may comprise one or more computer programs, software components, sets of instructions, procedures, functions, objects, classes, instances, associated data, or parts thereof, adapted for implementation in suitable, machine-readable program code. The applications 50 may be stored within the memory 46 or in additional or separate memory 46. Examples of applications 50 include audio or video streaming services, games, browsers, social media, etc.In other examples, the applications 50 are used to manage functions of the body control system, functions of the suspension control system 54, functions of the steering control system 56, functions of the powertrain control system 58, including functions of the transmission control system 59 and / or the engine 60, functions for controlling the braking system 62 of the host vehicle 12, or the like in an exemplary vehicle 12. More specifically, the host vehicle 12 is equipped with a variety of control systems that manage static and dynamic performance characteristics of the host vehicle 12 via a variety of onboard actuators with which the host vehicle 12 is equipped.

[0022] The actuators 64 can take any of a variety of different forms and manage many different and / or interrelated control systems of the host vehicle 12 without deviating from the scope or intent of this disclosure. It is understood that the actuators 64 can be electrically, hydraulically, pneumatically, mechanically, electromechanically, electrohydraulically, electropneumatically, magnetorheologically, hydropneumatically, electromagnetically, and / or by any combination of the above-mentioned types, and can be used to modify one or more static and / or dynamic performance characteristics of the host vehicle 12.

[0023] In some non-limiting examples, the suspension control system 54 comprises one or more suspension system actuators 64, such as active or semi-active dampers 66, which are capable of changing a damping force transmitted from wheels 26 of the host vehicle 12 to a body 68 of the host vehicle 12 while the host vehicle 12 is being driven on a road surface. The actuators of the steering control system 56 may include electric motors, electro-hydraulic, electro-pneumatic, or other such motors or steering actuators 70, which exert a torque on a steering shaft 72 or a steering rack 74 of the host vehicle 12 and thereby change the direction of travel of the host vehicle 12 by changing the position or angular orientation of the steerable wheels 26 of the host vehicle 12.In contrast, the onboard actuators 64 of the control system for the transmission 59 or the engine 60 may include the transmission 59 or the engine 60 itself and / or actuating components contained therein, which are capable of changing a torque output or torque ratio of the engine 60 or transmission 59, or the like. In some non-limiting examples, actuators 76 of the control system for the transmission 59 or the engine 60 may include a throttle valve or an electronic throttle valve 78 capable of changing a torque output of the engine 60, a transmission actuator 80 capable of changing the gear ratios and torque outputs transmitted from the engine 60 via the transmission 59, and the like. It should also be understood that, although the in . Fig. The host vehicle 12 shown is equipped with an internal combustion engine (ICE) 60, and the engine 60 can be any type of engine 60 or propulsion machine, such as an ICE engine 60, an electric motor, a hybrid electric motor 60, combinations thereof, or any other type of engine 60, without deviating from the scope or intent of this disclosure. Similarly, the actuators 64 of the braking system 62 comprise brakes 82 of the host vehicle 12, which are capable of selectively slowing the rotational speed of the wheels 26 of the host vehicle 12 and thereby changing the speed of the host vehicle 12 itself.

[0024] The host vehicle 12 can be operated in any one of a variety of different modes, including a fully manual mode in which the driver 42 has complete control over the static and dynamic performance characteristics of the host vehicle 12. In other non-limiting examples, the host vehicle 12 can be operated in fully or semi-autonomous modes that control some or all of the static and dynamic performance characteristics of the host vehicle 12. More specifically, the system 10 of the present disclosure operates on a host vehicle 12 with advanced driver assistance systems (ADAS) 84 that are capable of controlling functions of an automatic cruise control (ACC) 11, steering, braking, and any number of other means of controlling onboard systems of the host vehicle 12.

[0025] Now, referring to Fig. 2 and Fig. 3 and with continued reference to Fig. 1 The system 10 uses one or more applications 50, in particular a Maneuver Predictive Speed ​​Profile Adjustment (MPSPA) application, which uses data from sensors 14, driver preference data 42, location data and the like to adjust the speed profile of the ACC 11 of the host vehicle 12, while at the same time enabling drivers 42 to make course or direction changes conveniently and precisely via steering inputs without disabling the ACC.

[0026] System 10 and MPSPA application 52 enhance the control of the ACC 11 of the host vehicle 12 by incorporating maneuver prediction and speed profile adjustment. System 10 and MPSPA application 52 function similarly to a shadow planner, predicting where the driver 42 intends to go and adjusting the speed profile of the host vehicle 12 to best match the driver's intention. By analyzing inputs from the driver 42 and surrounding vehicles 12', as well as predicting the maneuvers of surrounding vehicles 12', System 10 and MPSPA application 52 adaptively adjust the speed profile of the ACC 11 to ensure efficient and comfortable operation of the host vehicle 12. In a non-restrictive example, the host vehicle 12 can slow down while the driver 42 makes a left turn using ACC.System 10 and the MPSPA application 52 improve the accuracy and responsiveness of the ACC, resulting in a smoother and more intelligent driving experience than systems 10 that are not equipped with it. System 10 and the MPSPA application 52 activate the ACC 11 while the host vehicle 12 is actively operated in complex manual driving situations such as turning, navigating, or driving along different paths, and the like.

[0027] In several aspects, System 10 and the MPSPA application 52 perform maneuver prediction through a mathematical process or derivation. To create a probabilistic maneuver prediction, a probability function is used for each vehicle state variable, expressed as a sigmoid function. Current state variables of the host vehicle 12 include: the speed of the host vehicle 12, ν x, Steering angle of the host vehicle 12, δ, Torsion bar torque, τ, and Torsion bar torque rate, τ̇. PM,x(k)=11+e−βM,x(|x|−αM,x) where x is a variable ν x , δ, τ, τ̇ is and M is the maneuver type that includes turning, changing lanes, turning around, evasive maneuvers and / or keeping in lane.

[0028] The maneuver probability is obtained by multiplying the state probability functions as follows: PM(k)=gd∏x=1nPM,x(k) where P M,x where x is a probability function of the variable for the maneuver M, k is a sampling time, β M,x a tuning parameter, the sigmoid slope of the variable x for the maneuver M, is, a M,x a tuning parameter, the sigmoid center of the variable x for the maneuver M, n is a number or set of variables, and g d one direction of the maneuver M is such that for a maneuver to the right g d= -1 and for a maneuver to the left g d = 1.

[0029] System 10 and MPSPA Application 52 also integrate environmental knowledge to increase the confidence levels of the predictions generated by System 10 and MPSPA Application 52. Environmental knowledge is used in various categories, including, but not limited to, road information such as road type, number or quantity of lanes, presence / absence of construction sites, information about surrounding objects such as direction signals for a nearest vehicle path (CIPV), navigation information such as a selected navigation route, and the like. The effect of each category of environmental knowledge is represented by a gain that depends on the maneuver performed by the host vehicle 12.If it is determined that some part or all of the environmental knowledge is irrelevant for a particular maneuver, the gain is set to one (1). Conversely, if the environmental knowledge is relevant for the particular maneuver, the gain is greater than one based on an importance level. For example, if a host vehicle 12 is traveling on a highway, the gain for road information for turning is 1, but the gain for road information for changing lanes is greater than 1 because the probability of changing lanes on the highway is higher than the probability of turning. Multiplying the gains gives a final environmental gain, which can be expressed as: gM,e=gM,r×gM,s×gM,n where g M,r a probability enhancement of road information for the maneuver M is, g M,sa probability increase from surrounding objects for the maneuver M is, g M,n a probability enhancement of navigation information for the maneuver M is and g M,e This is a probability enhancement for maneuver M based on environmental knowledge. The following equation expresses a maneuver probability function taking environmental knowledge into account: PM(k)=gM,e×gd∏x=1nPM,x(k).

[0030] To maintain the comfort and confidence of the driver 42 in the system 10 and the MPSPA application 52, the system 10 and the MPSPA application 52 actively adjust the function of the ACC 11 by adapting the driving style. The driving style adaptation can be considered a subroutine or algorithm of the MPSPA application 52. As part of the driving style adaptation, the probability of each predicted maneuver, along with the associated state variables of the host vehicle 12, is stored over time in a buffer of size L. By calculating an average and a standard deviation of the buffered state variables from the initial tuning parameters, αM,x0 and βM,x0 When the tuning parameters α are taken, they are adjusted. M,x and β M,x Updated as follows: αM,xd=αM,x0+λM,xαΔαM,x; ΔαM,x=∑j=1L(x(j)−αM,x0)L; βM,xd=βM,x0+λM,xβΔβM,x; ΔβM,x=KM,xβ∑j=1L(x(j)−βM,x0)2L; λM,x=f(1PM); where KM,xβ a distribution level is defined and λM,xα and λM,xβ Learning rates are. Finally, the system 10 and the MPSPA application 52 set the tuning parameters based on the road curvature, ρ, the inclination, ϕ, the gradient, θ, and the road friction, µ. αM,x=f(ϕ,ρ,θ,μ,αM,xd); βM,x=f(ϕ,ρ,θ,μ,βM,xd).

[0031] Subsequently, an update of ACC 11 is performed based on the maneuver prediction calculations above, so that a longitudinal control command (i.e., a command that controls a longitudinal acceleration and / or deceleration of the host vehicle 12) can be expressed as follows: [ΔVx,ref,Δax,ref]'=f(E,R,PM,V,τ,...)

[0032] In a more specific, but not restrictive, embodiment, the longitudinal control command can be expressed as follows: [ΔVx,ref,Δax,ref]'=(k1⋅V⋅k2(R)⋅(1−k3⋅E)+k4⋅τ)⋅k5(PM), taking into account the speed V of the host vehicle 12, a current speed of the host vehicle 12 in m / s, a road type R such as motorway, urban, rural or the like, the environmental complexity E; the driving environment complexity (e.g. traffic density, weather), the torque (τ) of the driver 42, which is a torque exerted by the driver 42 on the steering wheel of the host vehicle 12 (Nm), a maneuver prediction (P M), which is a type of predicted maneuver (e.g., turning, changing lanes, U-turn), and the like. It is understood that each type of maneuver can affect the longitudinal steering command differently. Example: Turning: May require a deceleration to ensure safety. Changing lanes: May require a slight acceleration to merge safely. U-turn: Typically requires a significant deceleration. Longitudinal steering command [ΔV x,ref ,Δa x,ref ]': desired acceleration or deceleration command (m / s²) 2 ) and the derivative.

[0033] The design parameters and calibrations for the above factors are described in more detail below: k1 is a speed scaling factor. The value k1 is a constant that scales a basic command based on the speed of the host vehicle 12. The k1 speed scaling factor determines how much acceleration or deceleration is applied relative to the current speed of the host vehicle 12. Higher values ​​of the k1 speed scaling factor increase the responsiveness of the system 10 and the MPSPA application 52 at higher speeds, while lower values ​​of the k1 speed scaling factor make the system 10 and the MPSPA application 52 more conservative, especially in urban environments.

[0034] k2 is a road type setting factor. The k2 road type setting factor adjusts the ACC 11 command based on the road type. Different road types require different driving behavior. In some non-restrictive examples, where the host vehicle 12 is driven on a highway, the k2 road type setting factor is close to one (e.g., 1.0), whereas when driving in the city, the k2 road type setting factor is less than one (e.g., 0.8), and when the host vehicle 12 is driven on country roads or in rural conditions, the k2 road type setting factor is slightly greater than one (e.g., 1.1).

[0035] k3 is a factor for environmental complexity, or environmental complexity factor. The k3 environmental complexity factor takes into account the complexity of the driving environment of the host vehicle 12, which influences how much the ACC 11 command is reduced under difficult conditions. In several non-restrictive examples, higher values ​​of the k3 environmental complexity factor cause the system 10 and the MPSPA application 52 to drive more cautiously in complex environments (e.g., in heavy traffic), while lower values ​​of the k3 environmental complexity factor allow the system 10 and the MPSPA application 52 to drive more aggressively in simpler environments.

[0036] k4 is an influencing factor for the driver's torque, or driver torque influence factor. The constant of the k4 driver torque influence factor determines the extent to which a steering input (torque) from the driver 42 influences the longitudinal command of the ACC 11. In some non-restrictive examples, higher values ​​of the k4 driver torque influence factor lead to a greater influence of the driver's inputs 42 on the acceleration / deceleration of the host vehicle 12, whereas lower values ​​of the k4 driver torque influence factor lead to increased levels of autonomous control, so that the acceleration / deceleration of the host vehicle 12 is less affected by the driver's inputs 42.

[0037] k5(P M ) is the maneuver prediction setting factor. The k5(P mThe maneuver prediction setting factor modifies the command based on the predicted maneuver type (e.g., turning, lane change). In several non-restrictive examples, turning is defined as (k5 < 1) (e.g., 0.7); a lane change as (k5 ∼ 1) (e.g., 1.0); and a U-turn has or is associated with a significant reduction in the value (e.g., 0.5).

[0038] In Fig. Figure 2 shows in more detail the system 10 and the MPSPA application 52 in the context of a host vehicle 12 making a left turn at an intersection 100. The host vehicle 12 is operated with automatic cruise control (ACC) 11 activated, and, in order to demonstrate the Fig. To comfortably navigate the left-hand curve shown in section 2, the system 10 and the MPSPA application 52 change the longitudinal speed of the host vehicle 12 by actuating the brakes 82, the engine 60, the transmission 59, or other such functions of the powertrain control system 58 of the host vehicle 12. Fig. 3 use in blocks 200, 202, 204 and 206 of Fig. 3. The system 10 and the MPSPA application 52 use the sensors 14 of the host vehicle 12 to determine a current static and / or dynamic operating state of the host vehicle 12. More specifically, the MPSPA application 52 and the system 10 use one or more IMUs 16, one or more steering position or steering angle sensors (SAS) 32, as well as accelerator pedal position sensors 28 and brake pedal position sensors 30 to generate state information about the host vehicle 12. Subsequently, the system 10 and the MPSPA application 52 in block 208 process the data from the sensors 14. To further refine the data from the sensors 14, the system 10 and the MPSPA application 52 can use camera sensors, driver monitoring information 42, wheel speed sensors 26, GPS sensor 20 data, and live traffic information.

[0039] Within block 208, the system 10 and the MPSPA application 52 filter the sensor 14 data in block 210, perform feature extraction in block 212 based on the filtered data from block 210, and then synchronize the sensor 14 data in block 214. By filtering the data in block 210, the MPSPA application 52 and the system 10 can reduce the use of computing resources from a first level of utilization to a second level of utilization that is lower than the first, by selecting only specific parts of the sensor 14 data that can indicate a state change or a command from the host vehicle 12 or the driver 42.To further refine the sensor 14 data and process it selectively, feature extraction in block 212 is applied to the filtered data from block 210 to determine whether a state change or command from the host vehicle 12 or the driver 42 is of sufficient magnitude to require additional processing. In some examples, the features found in the feature extraction in block 212 might include lane changes, route-controlled turns, highway on-ramps, and the like. Synchronization of the sensor 14 data ensures that the data received from the various sensors 14 of the host vehicle 12 are correctly and accurately aligned and associated in time, so that the sensor 14 data from each sensor 14 of the host vehicle 12 is correlated and time-aligned with the data received from the other sensors 14 of the host vehicle 12.System 10 and MPSPA application 52 then generate a probabilistic path prediction in block 216. As previously described, to generate the probabilistic path prediction, a probability function expressed in sigmoid terms is used for each state variable of the host vehicle 12. PM,x(k)=11+e−βM,x(|x|−αM,x) where x is a variable ν x , δ, τ, τ̇ is and M is a maneuver type that includes turning, lane changing, U-turn, evasive maneuvers and / or lane keeping, where the current state variables of the host vehicle include 12: speed, ν x of the host vehicle 12, steering angle, δ of the host vehicle 12, torsion bar torque, τ, and torsion bar torque rate, τ̇.

[0040] In Block 218, System 10 and MPSPA Application 52 normalize the sensor 14 data to improve the quality, integrity, flexibility, performance, and usability of the sensor 14 data from a first level to a second level that is higher than the first. Furthermore, normalizing the sensor 14 data reduces redundant information from a first level of redundancy to a second level that is lower than the first. In Block 220, System 10 and MPSPA Application 52 perform a confidence calculation, utilizing environmental knowledge in a variety of different categories as previously described. A confidence calculation output defines the probability that a specific type of maneuver M will be performed by the host vehicle 12.Subsequently, the System 10 and the MPSPA application 52 proceed from Block 220 within Block 208 to Block 222, where the System 10 and the MPSPA 52 perform a detection of the driver's intent 42.

[0041] In several aspects, the process for detecting the driver's intent 42 utilizes a machine learning (ML) model 224 to classify the driver's intent 42 in a data-driven classification process. More specifically, the system 10 and the MSPA application 52 integrate a shadow planner to improve the ACC 11 by predicting the driver's intended path 42 and adjusting a speed profile of the host vehicle 12 to match the driver's intent, based on an analysis of the driver's inputs 42 to the host vehicle 12's control systems and the behavior of surrounding vehicles 12'. In block 226, the system 10 and the MPSPA application 52 perform a real-time path classification to assess the type of maneuver that the driver 42 and / or the ADAS 84 of the host vehicle 12 are actively pursuing / tracking.Various types of maneuvers by the host vehicle 12 are classified in a data-driven manner and can include real-time path classifications such as "turn left," "turn right," and the like. The intention of the driver 42 can be expressed as one or more of a lateral command 300 and a longitudinal command 400, which are specified in . Fig. 4A and Fig. 4B are described in more detail. System 10 and MPSPA application 52 then perform an enumeration calculation in block 228. The enumeration calculation in block 228 qualifies the classified real-time paths, defining them as left turns, right turns, or the like. In block 230, System 10 and MPSPA application 52 then generate a trajectory prediction based on the real-time paths, GPS data, and navigation information, such as a selected navigation route. The results or outputs of the driver intent detection processes within block 222 are expressed as changes in vehicle speed ΔV. x,ref , changes in vehicle acceleration Δa x,refand the like, generated and sent to the ACC 11 to change the speed of a host vehicle 12 by activating and controlling one or more actuators 64 of the control systems for the transmission 59 and / or control systems for the engine 60, the brake control system 62 and the like.

[0042] If one now turns Fig. 4A and Fig. 4B more specifically, and with continued reference to Fig. Figures 1-3 are graphical representations of the operation of System 10 and the MPSPA application 52 during the in Fig. The exemplary turning maneuver shown in section 2 is illustrated in more detail. Fig. 4A and Fig. Figure 4B shows lateral commands 300 and longitudinal commands 400, respectively. Lateral commands 300 are commands to the actuators 64 of the host vehicle 12 to change the lateral trajectory of the host vehicle 12. That is, lateral commands 300 are commands that apply one or more torques to the wheels 26 of the host vehicle 12 on opposite sides of the host vehicle 12 (i.e., the left and right sides of the host vehicle 12) by applying a differential torque to the wheels 26 of the host vehicle 12 by means of the actuators 64 of the control system for the transmission 59 and / or the engine 60 and / or the actuators 64 of the brake control system 62, resulting in a rotation or yaw of the host vehicle 12.In additional, non-restrictive examples, the lateral commands 300 are inputs to the steering system 34 or commands that act on it via steering actuators 70, which exert a torque on a steering shaft 72 and / or a steering rack 74 of the host vehicle 12, resulting in a rotation or yaw of the host vehicle 12 in a lateral rather than longitudinal direction.

[0043] In contrast, the longitudinal commands 400 define commands of system 10 and the MPSPA application 52 to one or more of the control systems for the transmission 59 and the engine, and more specifically to the actuators 76 of the engine control system and / or transmission actuators 80, to modify a torque output of the engine 60 and the transmission 59, thereby changing the rotational speed, velocity, and / or acceleration of the host vehicle 12. Likewise, the longitudinal commands 400 may also include the application of torque to the wheels 26 of the host vehicle 12 by means of actuators 64 of the control systems for the transmission 59 and / or the engine 60 and / or actuators 64 of the brake control system 62, thereby causing changes in the rotational speed of the wheels 26 and resulting in changes in the rotational speed, velocity, and / or acceleration of the host vehicle 12.

[0044] As in Fig. 4A and Fig. Figure 4B shows the lateral commands 300 and longitudinal commands 400, with the order of magnitude of the commands along the Y-axis and time along the X-axis. Specifically, on Fig. Referring to 4A, the vertical dashed line 402 defines an event in which the ADAS 84 and, in particular, the ACC 11 of the host vehicle 12 act to set a speed of the host vehicle 12. More precisely, the event can be cornering or turning, or the like, such as that described in Fig. 2 shown. It is understood that in a fully manually controlled vehicle 12, when initiating a turn, the driver 42 in a vehicle 12 sets the longitudinal speed of the host vehicle 12 from a previous cruising speed to a desired speed, where the cruising speed of the host vehicle 12 is greater than the desired speed of the host vehicle 12. Upon reaching the desired speed, the driver 42 then initiates the turn by inputting or otherwise generating commands that act on the steering system 34 via steering actuators 70, which apply a torque to a steering shaft 72 and / or a steering rack 74 of the host vehicle 12, resulting in a rotation or yaw of the host vehicle 12, thereby steering the host vehicle 12 through the turn.Upon completion of the turn, or upon completion of a sufficient portion of the turn so that the steering inputs can be derived or otherwise reduced, such that the wheels 26 of the host vehicle 12 approach a straight-ahead position and are in the direction of travel of the host vehicle 12, the driver 42 applies throttle inputs or accelerates to bring the host vehicle 12 back up to cruising speed. Fig. 4A represents a host vehicle 12 operating in a largely autonomous mode, such as a Level 2 autonomous driving mode, in which the ADAS 84 applies lateral and longitudinal inputs to the actuators 64 of the host vehicle 12 to cause the host vehicle 12 to perform the actions described in Fig. The second curve shown is driven.

[0045] In contrast, monitoring as in Fig.Figure 4B shows the system 10 and the MPSPA application 52 monitoring the sensors 14 and actuators 64 of the host vehicle 12, both while the host vehicle 12 is operated manually and while the host vehicle 12 is operated autonomously or semi-autonomously using the ADAS and the ACC. Over time, the MPSPA application 52 and the system 10 are trained to recognize situations in which turning maneuvers are initiated and adaptively adjust longitudinal and lateral inputs 404 to achieve lateral and longitudinal velocity, yaw, and forces observed by the driver 42 that correspond to each driver's historical preferences 42. That is, in some examples, the sensors 14 of the host vehicle 12 detect information about each driver 42 and store the information about drivers 42 within the memory 46 of the controllers 40 of the host vehicle 12 as preferences of the drivers 42.The information about drivers 42 can include information about typical driving styles and procedures for operating the host vehicle 12 of each driver 42. From the preferences of the drivers 42, the system 10 and the MPSPA application 52 generate a predicted maneuver, including a predicted lateral command 302 and the predicted longitudinal command 400. In several aspects, the predicted lateral command 302 defines an output command of the system 10 and the MPSPA application 52 to modify a lateral movement of the host vehicle 12 in such a way as to approximate or mimic the lateral commands 304 applied by the driver 42 and stored in memory 46. In some non-restrictive examples, the preferences of the driver 42 can be stored in the memory 46 of the controller 40 of the host vehicle 12 based on the individual driver 42 or an identifier of the driver 42.This means that each individual driver 42 can log in to the host vehicle 12 or otherwise indicate their presence, whereupon the system 10 and the MPSPA application 52 retrieve the preferences of the driver 42 corresponding to the currently logged-in driver 42 from memory 46 and adjust the functions and operations of the ACC 11 and the ADAS of the host vehicle 12 according to the stored preferences of the currently logged-in driver 42.

[0046] In additional, non-limiting examples, the preferences of the driver 42, including the lateral commands 304 of the driver 42, are stored locally in the memory 46 within the controllers 40 of the host vehicle 12, as well as in additional or auxiliary memories 46 of the controllers 40 that are separate and remote from the host vehicle 12. That is to say, the lateral commands of the driver 42 are also stored in additional or auxiliary memories 46 of controllers 40 that are remote from the host vehicle 12. Such remote controllers 40 may, in some non-limiting examples, be located in or on servers 406 for cloud computing, satellite-based servers, and / or at any other location in electronic wireless communication with controllers 40 of the host vehicle 12, without deviating from the scope or intent of the present disclosure.

[0047] In several aspects, the system 10 and the MPSPA application 52 of the present disclosure can be operated in vehicles 12 with Level 1 and / or Level 2 autonomy capabilities. That is, the host vehicle 12 can be equipped with systems that provide the driver 42 with assistance in steering, braking, and / or acceleration in predefined situations.Such Level 1 autonomy may include, for example: adaptive cruise control (ACC) 11, which maintains a safe distance to a surrounding vehicle 12' in front of the host vehicle 12 by controlling the acceleration and braking of the host vehicle 12; automatic lane keeping, which helps to keep the host vehicle 12 in its current lane by detecting lateral movement of the host vehicle 12 in the lane and changing the steering position to correct such movement; and / or parking assistance, which can assist in reversing the host vehicle 12 into a parking space without intervention or input from the driver 42. In contrast, Level 2 autonomy provides partial assistance to the driver 42 by providing some steering, acceleration, and / or braking control through the use of ACC, a lane keeping assist (LKA) system that provides gentle steering inputs to keep the host vehicle 12 in its current lane.An automated parking aid for parallel or perpendicular parking is provided via steering and / or braking and acceleration control and traffic jam assistance, which assists the host vehicle 12 in maintaining a set speed and following a preceding host vehicle 12 in slow-moving traffic. In a non-restrictive example, the system 10 and the MPSPA application 52 of the present disclosure use mission or route planning information to adaptively control longitudinal commands 400 via the ACC 11 in a vehicle 12 with Level 1 autonomy. In contrast, in a vehicle 12 with Level 2 autonomy, the system 10 and the MPSPA application 52 of the present disclosure adaptively control at least the longitudinal commands 400 via the ACC and, in some cases, also the lateral commands 300.

[0048] A System 10 and an MPSPA application 52 of the present disclosure offer several advantages. These include the ability to utilize an ADAS-based ACC 11 that incorporates maneuver prediction and speed adjustments to ensure that vehicle performance adapts to the input preferences of the operator or driver using ADAS and ACC 11, thereby increasing operator comfort while maintaining the efficiency of the ADAS and ACC 11, improving the accuracy and responsiveness of the ACC, resulting in a smoother and more intelligent driving experience, while maintaining or reducing the complexity of the System 10 and increasing its redundancy.

[0049] The description of the present revelation is by its very nature merely exemplary, and variations that do not deviate from the core of the present revelation are to be considered within the scope of the present revelation. Such variations are not to be regarded as a deviation from the spirit and scope of the present revelation.

Claims

[1] System (10) for maneuver prediction-based speed profile adaptation for behavior-based automated speed control in a host vehicle (12) comprises: a host vehicle (12); one or more sensors (14), wherein the one or more sensors (14) detect static and dynamic status information about the host vehicle (12); one or more actuators (64), wherein the one or more actuators (64) modify the static and dynamic performance of the host vehicle (12); a controller (40) comprising a processor (44), a memory (46) and input / output ports (I / O ports), wherein the I / O ports are connected to the one or more sensors (14) and the one or more actuators (64), wherein the processor (44) executes program code portions stored in the memory (46), the program code portions comprising a Maneuver Predictive Speed ​​Profile Adjustment (MPSPA) application (52), comprising: a first control logic for acquiring static and dynamic information about the host vehicle (12) and about an environment surrounding the host vehicle (12); a second control logic for processing the static and dynamic information about the host vehicle (12) and about the environment surrounding the host vehicle (12), wherein the processing includes at least: feature extraction, synchronization, normalization and the generation of a probabilistic path prediction; a third control logic for detecting a driver's intention; and a fourth control logic for generating a control command to the one or more actuators (64), wherein the control command selectively modifies at least one longitudinal velocity of the host vehicle (12) to approximate historical preferences of the driver regarding lateral and longitudinal velocities, yaw, and forces observed by the driver while the host vehicle (12) is driven in a fully autonomous and / or semi-autonomous mode; the second control logic further includes: a control logic for filtering and performing feature extraction on data from the one or more sensors (14); wherein the filtering causes the system (10) and the MPSPA application (52) to reduce the use of computing resources from a first level to a second level that is lower than the first by selecting only certain parts of the data from the one or more sensors (14) that can indicate a change in the state of the host vehicle (12), and wherein the feature extraction further determines that the change in the state of the host vehicle (12) is of a magnitude that requires additional processing; a control logic for synchronizing the data from the one or the multiple sensors (14), wherein the synchronization of the data causes the data from the one or the multiple sensors (14) to be precisely aligned and associated in time, so that data from each of the one or the multiple sensors (14) are temporally correlated with data from each of the other sensors (14) of the one or the multiple sensors (14); a control logic for generating a probabilistic path prediction using a probability function for each state variable that follows a sigmoid function, where: State variables of the host vehicle (12) include: the speed, ν x , of the host vehicle (12), the steering angle, δ, of the host vehicle (12), the torsion bar torque, τ, and the torsion bar torque rate, τ̇, PM,x(k)=11+e−βM,x(|x|−αM,x) where x is a variable ν x, δ, τ, τ̇ is and M is a maneuver type that includes turning, lane changing, U-turn, evasive maneuvers and / or lane keeping, and where a maneuver probability is obtained by multiplying the state probability functions as follows: PM(k)=gd∏x=1nPM,x(k) where P M,x where x is a probability function of the variable for the maneuver M, k is a sampling time, β M,x a tuning parameter, the sigmoid slope of the variable x for the maneuver M, is, a M,x a tuning parameter, the sigmoid center of the variable x for the maneuver M, n is a number or set of variables, and g d one direction of the maneuver M is such that for a maneuver to the right g d = -1 and for a maneuver to the left g d = 1 is; the third control logic further includes: a control logic to utilize a machine learning (ML) model for data-driven classification to integrate a shadow planner that enhances adaptive cruise control (ACC) by predicting the driver's intended path and adjusting a speed profile of the host vehicle (12) to match the driver's intention, based on an analysis of driver inputs to the host vehicle's (12) control systems and the behavior of surrounding vehicles; a control logic for performing real-time path classification to assess the type of maneuver that the driver and the ADAS of the host vehicle (12) are actively pursuing; and a control logic for performing an enumeration calculation that qualifies classified real-time paths as a left turn and / or a right turn. [2] System (10) according to claim 1, wherein the first control logic further comprises: a control logic for obtaining information about the position, movement and acceleration of the host vehicle (12) in at least three degrees of freedom from an inertial measurement unit (IMU); a control logic for obtaining a steering angle of the host vehicle (12) from one or more steering angle sensors (SAS) (32); a control logic for obtaining an accelerator pedal position from an accelerator pedal position sensor (28); and a control logic for obtaining a brake pedal position from a brake pedal position sensor (30). [3] System (10) according to claim 1, further comprising: A control logic for integrating environmental knowledge into environmental knowledge categories, which include: road information, road type, a number or quantity of lanes, presence / absence of construction sites, information on surrounding objects, direction signals for a nearest vehicle path (CIPV), and navigation information, wherein an effect of each environmental knowledge category is represented by a gain, where a multiplication of all gains of the environmental knowledge categories defines a final environmental gain, which is expressed as: gM,e=gM,r×gM,s×gM,n where g M,r a probability enhancement of road information for the maneuver M is, g M,s a probability increase from surrounding objects for the maneuver M is, g M,n a probability enhancement of navigation information for the maneuver M is and g M,ea probability enhancement of environmental knowledge for the maneuver M, and where the following equation expresses a maneuver probability function taking environmental knowledge into account: PM(k)=gM,e×gd∏x=1nPM,x(k). [4] System according to claim 3, wherein the second control logic further comprises: a control logic for normalizing the sensor data, wherein the normalization of the sensor data reduces redundant information in sensor data from a first redundancy level to a second redundancy level that is lower than the first redundancy level; and a control logic for generating a confidence calculation, wherein an output of the confidence calculation defines a probability that a certain type of maneuver M is performed by the host vehicle (12). [5] System (10) according to claim 3, wherein the third control logic further comprises: a control logic for generating a trajectory prediction based on real-time paths, data from a global Positioning system (GPS) and navigation information, including a selected navigation route. [6] System (10) according to claim 5, wherein the fourth control logic further comprises: a control logic for generating a control command to the one or more actuators (64), wherein the control command comprises a predicted lateral command and a predicted longitudinal command; and a control logic for activating and controlling one or more actuators (64) of a transmission control system, an engine control system and a brake control system of a host vehicle (12) to change at least a longitudinal velocity of the host vehicle (12) while performing an actual classified real-time path maneuver M according to the predicted lateral command and the predicted longitudinal command.

Images