Methods for avoiding an imminent collision of a vehicle and accident avoidance system

Abstract

Method for determining an impending collision of a vehicle, wherein in a first step (S1) a state of a driver (1) of the vehicle is recorded, and in a second step (S2) an evaluation (2) of the recorded state is carried out in such a way as to determine whether a collision is intentionally caused, wherein in a third step (S3), if an intentional collision has been determined based on the evaluation (2; 4), a braking system (100) present in the vehicle is activated.

Description

[0001] The present invention relates to a method for avoiding an impending collision of a vehicle and to an accident avoidance system.

[0002] Emergency braking systems are well-known and already legally mandated in certain vehicles, such as trucks. Many passenger cars are now also equipped with such systems. These systems are designed to brake the vehicle under specific or predetermined circumstances, if necessary to a complete stop. For this purpose, vehicle parameters such as the current speed are monitored, as well as environmental parameters such as existing and adjacent lanes, other vehicles surrounding the vehicle, pedestrians, etc. This monitoring is carried out using appropriate devices such as radar, lidar, cameras, or a combination thereof. An evaluation system located in the vehicle, for example in a corresponding control unit, assesses the collected data and then determines whether the vehicle should be braked, for example due to a calculated imminent collision, and if so, how hard.Such a system can, for example, prevent or at least mitigate accidents caused by inattention or a lack of reaction from the driver to a specific event.

[0003] The state of the art is disclosed in DE 10 2017 124 900 A1, DE 10 2018 004 479 A1, DE 10 2018 117 380 A1 and DE 10 2019 004 822 A1.

[0004] While existing emergency braking systems and methods for activating a vehicle's emergency braking system cover a wide range of situations to which they respond, they still require improvement. Therefore, an object of this invention is to provide a method for preventing an impending collision of a vehicle, as well as an accident avoidance system. This object is achieved according to the invention by the features of the independent claims. Advantageous embodiments are the subject of the dependent claims.

[0005] A method is proposed for detecting an impending collision of a vehicle. In a first step, the driver's condition is recorded. In a second step, this recorded condition is evaluated to determine whether a collision is imminent. In a third step, if an imminent collision is determined based on the evaluation, a braking system present in the vehicle is activated.

[0006] The object of the invention is to prevent imminent or imminent, intended or unintentional collisions of a vehicle by analyzing and evaluating the driver's state and initiating braking of the vehicle by activating the braking system under predetermined conditions. The braking system can also be a so-called emergency braking system. Depending on the evaluation of the driver's state, it is determined whether a collision of the vehicle will occur imminently, i.e., within a predetermined period, e.g., up to a few seconds, if the driver takes no action, e.g., by braking.

[0007] In one implementation, the evaluation of the detected condition results in a probability value for an impending collision, and in the third step, the braking system is activated if a predefined maximum value is reached or exceeded. For the evaluation, a classification is performed, for example, using a classifier, which determines a probability value indicating that a collision will occur soon or within a specified time period.

[0008] In one implementation, the assessment of whether a collision is imminent is carried out by determining an aggression and / or fear classifier from the driver's detected state. This classifier represents the probability of an aggression and / or fear state, and if a predefined value is exceeded, a collision is determined to be imminent. By differentiating between possible types of driver states, the point at which the braking system or emergency braking system is activated can be individually adapted to the driver's current state. For example, the threshold can also be adjusted accordingly, whereby, for instance, the threshold at which the braking system is activated might be lower in the case of a detected aggression state than in the case of a fear state. The threshold can depend on a recognized or known characteristic of the driver.This means that the threshold value can be determined based on the individual's baseline state. For example, the assessment can take into account that a driver is rather anxious or aggressive, but despite a higher rating of their condition, they do not pose a danger. This baseline state can be predefined as a default setting or determined using known algorithms, for example, based on current biometric data, camera data, or a combination thereof.

[0009] According to the invention, if the vehicle is determined to be in an aggressive state, it can be assumed that a detected imminent collision is intended to be brought about intentionally. For example, such a prerequisite could be that the driver's condition is far above a predetermined threshold for the detection of an aggressive state, or that certain biometric data indicate a significantly elevated stress level. The detection of (increased) aggressiveness is particularly helpful when an emergency braking system is present that monitors the vehicle's surroundings and detects a critical situation. Such a critical situation could be individual pedestrians, crowds of people, or areas with restricted access, but which the vehicle is approaching.In situations where both the driver's condition indicates increased aggression and a critical situation is detected in the vehicle environment, according to the invention, an intended collision is assumed to be probable and the braking system or the emergency braking system is activated immediately.

[0010] Several methods and possibilities for recognizing a person's emotional state are already known. These are often implemented as algorithms, making them suitable for the present invention if, for example, they are implemented as software or computer program products in an evaluation unit within the vehicle.

[0011] Thus, an assessment based on, for example, image analysis using artificial intelligence, neural networks or learning algorithms can determine that a person is aggressive, exhibits or will exhibit violent behavior, or is afraid.

[0012] A determination that the driver is exhibiting a state of aggression or anxiety can indicate that a collision is imminent, even if the vehicle's emergency braking system has not (yet) detected an impending collision. This allows the reaction time of the emergency braking system to be accelerated.

[0013] By setting a limit value, an individual threshold for activating the braking system or emergency braking system can be defined for each possible condition and each person. This allows the system's reaction or activation to be individualized for each driver.

[0014] In one implementation, the driver's state is monitored in addition to parameters for activating an emergency braking system. The determination of whether a collision will occur is based on an evaluation of a combination of the driver's state and the emergency braking system activation parameters. As described in the introduction, emergency braking systems, also known as emergency brake assist systems, are already known. These systems essentially monitor the vehicle's surroundings. Corresponding processing units are designed to determine, based on this data in combination with vehicle data such as speed, steering angle, navigation information, etc., whether a collision with another vehicle, a pedestrian, and, depending on the implementation, a stationary object is imminent. If an imminent collision is detected, the emergency braking system is activated, which then decelerates the vehicle.The emergency braking system is therefore an extension of the braking system, adding a control or regulation component that depends on predefined parameters. The fusion or combination of the parameters for activating the emergency braking system with the driver's state data leads to a significant improvement in collision detection. It can even expand the range of situations in which a collision is detected compared to existing systems.

[0015] In one implementation, the driver's condition is recorded using sensors already present in the vehicle, and at least one or a combination of the following characteristics is recorded and evaluated: facial expression, eye movement, abrupt acceleration of the vehicle, gestures, acoustic utterances, emotional state, biometric data, including heart rate, body temperature, perspiration, respiration, and respiratory rate.

[0016] The assessment of the driver's condition can be carried out using a wide variety of existing or specially installed aids, which are collectively referred to below as driver condition monitoring systems. For example, a camera monitoring the vehicle's interior can also be used to monitor the driver's facial expressions, gestures, and pupil activity, provided it has sufficient resolution and orientation. This can be achieved by implementing a suitable algorithm in the camera and / or an evaluation unit, or by adjusting its position. Thus, no additional hardware is required. Similarly, an existing entertainment system with voice or speech recognition can be used to detect outbursts of anger, etc. A corresponding algorithm, implemented in an evaluation unit, can also be provided for this purpose.Furthermore, existing or retrofittable measuring elements can be used to measure the driver's biometric data, e.g., temperature sensors or heart rate sensors on the steering wheel, seat, or contactless. Existing or retrofitted elements can also be used to detect abrupt acceleration of the accelerator pedal. This is particularly useful in combination with the detection of a potential collision during vehicle acceleration or the potential risk of injury to pedestrians, etc., to avoid unwanted intervention during overtaking maneuvers, etc.

[0017] In one implementation, a neural network is used to evaluate the driver's state data and determine the driver's state based on this evaluation. Learning algorithms can therefore be implemented to improve the accuracy of the state assessment and to provide individualization for each driver.

[0018] Furthermore, a computer program product is proposed which performs the described procedure and can be executed on an evaluation unit of the vehicle.

[0019] Furthermore, an accident avoidance system for a vehicle is proposed, comprising at least one driver state detection device designed to detect the driver's state, and at least one evaluation device that is in communication and / or operational connection with the driver state detection device and a braking system of the vehicle, and in which the computer program product is implemented.

[0020] In one embodiment, the driver state monitoring device comprises at least one of the following: a driver monitoring camera located inside the vehicle, a device for monitoring the driver's biometric data, a device for monitoring accelerator pedal actuation, a device for monitoring acoustic utterances, and a device for monitoring the driver's emotional state. Advantageously, the evaluation unit is designed such that the processing of the data for determining the driver's state and the parameters for activating the emergency braking system, if present, can occur in real time; that is, the computing capacity, performance, etc., are selected accordingly.

[0021] Further features and advantages of the invention will become apparent from the following description of exemplary embodiments of the invention, with reference to the figures in the drawing, which shows details of the invention, and from the claims. The individual features can be implemented individually or in any combination in a variant of the invention.

[0022] Preferred embodiments of the invention are explained in more detail below with reference to the accompanying drawing. Fig. Figure 1 shows an exemplary representation of data to be recorded from a vehicle driver according to one embodiment of the present invention. Fig. Figure 2 shows a schematic representation of the fusion between recorded data of the driver and parameters of the emergency braking system according to one embodiment of the present invention. Fig. Figure 3 shows a schematic representation of a process flow according to one embodiment of the present invention.

[0023] In the following figure descriptions, identical elements or functions are marked with the same reference symbols.

[0024] In Fig. Figure 1 depicts a driver whose condition is determined in different ways. Fig. Figure 2 shows the fusion of the data collected from driver 1 regarding their condition and the parameters relevant to the emergency braking system. A more detailed description of the relationships follows.

[0025] The state of driver 1 can be divided into two types. One type is an active state, which can also be described as driver 1's behavior. The other type is a passive state, which essentially describes driver 1's physical state, which can be described and measured using biometric data. Such data includes, for example, heart rate, respiratory rate, and emotional state such as stress, anger, or fear, which can also be determined by assessing facial expressions. The active state is essentially the behavior, which can also result from the physical condition, e.g.,Acoustic expressions such as cursing, loud shouting, outbursts of anger, frightened expressions, but also gestures or abrupt use of the accelerator pedal and thus abrupt acceleration of the vehicle, especially in situations where this is classified as dangerous or impermissible by an existing emergency braking system.

[0026] In Fig. The driver is monitored by a driver condition monitoring device for facial recognition 20, a driver condition monitoring device for body temperature 21, and a driver condition monitoring device for heart rate 22. Different sensors, either already present in the vehicle or specifically designed for this purpose, can be used to determine the various parameters. Such sensors can operate without contact, for example, to measure a pulse, heart rate, or respiratory rate, or for optical observation like cameras. They can also operate via contact, for example, as contact sensors on the steering wheel to measure heart rate or to detect whether hands are wet, which can indicate emotional agitation.The methods for determining the state of a vehicle occupant, in particular a driver, are known from the prior art and can be applied accordingly to this invention. For example, facial recognition can be used to determine the emotional state, e.g., based on the eye area, pupil movement, corner of the mouth, etc. Correct recognition for the observed person can be based on learned data or be individually trained or predefined for each person.

[0027] In Fig. Figure 2 shows that the driver state data 201, 202, 203 determined for the driver 1 are evaluated, for example by a corresponding evaluation unit 101 in the vehicle, such that the driver's state can be classified into a defined state 2, e.g., an aggressive state or a fear state. This evaluation is advantageously carried out using a corresponding algorithm, which can also be trained and extended using machine learning. This algorithm is advantageously executed as a computer program or computer program product on an evaluation unit 101 in the vehicle, so that the described procedure can be carried out.

[0028] If the vehicle has an emergency braking system or an emergency brake assist system, the parameters 30-33 recorded for activating the emergency braking system, or a parameter combination 3 thereof, can be combined or fused with the recorded driver state data 201, 202, 203, or the defined state 2 of the driver 1 determined therefrom. Thus, a further classification 4 can be made by combining the available data, which determines whether a collision is imminent based on the defined state 2 and the parameter data 3, and whether the vehicle's braking system 100 should be activated.

[0029] The parameters 30 to 33 used to activate the emergency braking system can vary depending on the type and equipment of the vehicle and may include, for example, parameters for vehicle speed 30, pedestrian detection 31, lane detection 32, and vehicle detection 33. Parameters for vehicle speed 30 can be acquired by means present in the vehicle and generally capture at least the current speed of the vehicle. Parameters for pedestrian detection 31, lane detection 32, and also vehicle detection 33 can be acquired by appropriate sensors such as radar sensors or camera(s) on the vehicle. To determine whether the acquired pedestrian data is relevant for assessing whether a collision is imminent, it can be processed in a corresponding evaluation unit 101. To assess whether the lane detection data leads to a collision with, for example, a pedestrian, the vehicle can be evaluated using a different evaluation unit 101.Data from vehicles in adjacent lanes or in front of the vehicle can also be processed in one or the same evaluation unit 101. Furthermore, vehicle detection 33 can be performed in conjunction with lane detection 32, but also independently, detecting vehicles surrounding the vehicle, e.g., in front of, beside, or behind it. The possibilities for parameter acquisition for activating an emergency braking system are well known, so they will not be discussed in detail here.

[0030] Through data fusion and classification 4 from individual or combined state data 201-203; 2 of the driver 1 with individual or combined parameters 30-33 available depending on the vehicle's equipment; 3 to activate an emergency braking system, an improvement in the detection of an impending collision is provided, whether it is determined that the collision is unintentionally caused or intentionally caused.

[0031] The evaluation unit 101 is advantageously implemented in the vehicle, e.g., as part of a control unit. It is advantageously in communication and operational link with at least one driver status monitoring device 20-22, enabling data and signal exchange. Furthermore, it can process the data received from the driver status monitoring devices 20-22 as close to real time as possible, in order to provide the fastest possible reaction to a detected, potentially imminent collision.

[0032] Fig.Figure 3 shows a schematic flowchart of the procedure, wherein in a first step S1 a recording of driver condition data 201-203, i.e. a state of the driver 1, of the vehicle, is carried out, and in a second step S2 an evaluation 2 of the recorded state is carried out in such a way as to determine whether a collision is imminent, and in a third step S3, if an imminent collision has been determined on the basis of the evaluation 2; 4, a braking system 100 present in the vehicle is activated. Reference symbol list 1 driver 20 Driver status detection device for facial recognition 21 Driver condition monitoring device for body temperature 22 Driver condition monitoring device for heart rate 201-203 Driver Condition Data 30 vehicle speed 31 Pedestrian recording 32 Lane detection 33 Vehicle identification 2. Classification of the driver's condition 3 Classification of parameters for emergency braking system 4. Merged Classification 100 Braking system / Emergency braking system 101 Evaluation unit S1-S3 process steps

Claims

[1] Method for determining an impending collision of a vehicle, wherein in a first step (S1) a state of a driver (1) of the vehicle is recorded, and in a second step (S2) an evaluation (2) of the recorded state is carried out in such a way as to determine whether a collision is being intentionally caused, wherein in a third step (S3), if an intentional collision has been determined based on the evaluation (2; 4), a braking system (100) present in the vehicle is activated. [2] Method according to claim 1, wherein the evaluation (2) of the detected state results in a value of a probability of an impending collision, and wherein in the third step (S3) the braking system (100) is activated when a predetermined maximum value is reached or exceeded. [3] Method according to one of the preceding claims, wherein the assessment (2; 4) of whether a collision is imminent is carried out by determining an aggression and / or fear classifier (2) from the detected state of the driver (1), which represents a probability of a state of aggression and / or a state of fear, and wherein, if a predetermined value is exceeded, it is determined that a collision is imminent. [4] Method according to one of the preceding claims, wherein the detection of the driver's state (1) is carried out in addition to the detection (4) of parameters (30, 31, 32, 33) for activating the braking system (100), and wherein the determination of whether a collision will occur is based on an evaluation (4) of a combination of detection (2; 3) of the driver's state (1) and parameters (30, 31, 32, 33) for activating the braking system (100). [5] Method according to one of the preceding claims, wherein the state of the driver (1) is detected by means of sensors present in the vehicle and at least one or a combination of the following features is detected and evaluated: facial expression, eye movement, abrupt acceleration of the vehicle, gestures, acoustic utterances, emotional state, biometric data, including heart rate, body temperature, perspiration, respiration, respiratory rate. [6] Method according to any of the preceding claims, wherein a limit value is set for each possible state of aggression and an individual threshold for the activation of the braking system or the emergency braking system is set for each person. [7] Method according to one of the preceding claims, wherein the driver's state data (201, 202, 203) are evaluated by means of a neural network and the driver's state is determined on the basis thereof.

Images