Procedures and systems for detecting hazardous situations

Abstract

The invention relates to a method for determining a hazardous situation in the vicinity of an ego-vehicle (1), wherein the ego-vehicle (1) has sensors (2) for detecting an area surrounding the ego-vehicle (1), and wherein the method comprises the following steps: - detecting a third vehicle (3) in the area surrounding the ego-vehicle (1) (S10); - detecting and / or receiving angular information about the yaw angle, pitch angle, roll angle and / or sideslip angle of the third vehicle (3) (S11); - processing the angular information by a computer unit (4) of the ego-vehicle (1) (S12); and - calculating a hazard probability value based on the angular information, wherein the hazard probability value indicates the probability that a hazard emanates from the third vehicle (3) to the ego-vehicle (1) (S13).

Description

The invention relates to a method and a system for determining hazardous situations in the environment of an ego vehicle.It is known to detect the surrounding area around the vehicle by means of a sensor system of a vehicle and to create a surrounding environment model in order to be able to make driving decisions of a driving assistance system or of an autonomous driving system based on this surrounding environment model. The sensor system may include, in particular, one or more radar sensors, LIDAR sensors (LIDAR: light detection and ranging), cameras, stereo cameras and / or ultrasonic sensors.DE 10 2017 114 876 A1 discloses a method for avoiding a potential collision between an ego vehicle and another road user.DE 10 2008 025 773 A1 discloses a method for estimating a location and movement state of an observed object, in which a distance to the object is determined, a displacement of at least a part of the object is detected in the image of a camera, and at least one parameter of the location and movement state is estimated taking into account a movement model of the observed object.DE 10 2016 001 253 A1 relates to a method for operating an autonomously or at least partially autonomously driving vehicle, wherein road users in the surroundings of the vehicle and their respective movement trajectory are detected.US 2022 / 0 135 026 A1 discloses a system and a method for controlling a vehicle on the basis of an algorithm for collision avoidance.In order to be able to make situation-appropriate driving decisions, the driving assistance system or the autonomous driving system requires as precise as possible information with regard to the position and the predicted movement of other road users. Known algorithms use the current position, the speed and the acceleration in order to estimate the future position or a movement trajectory of this road user based thereon.The prediction accuracy and the ability to identify hazardous situations as early as possible are essential for avoiding accidents and / or unsafe or uncomfortable driving maneuvers of the ego vehicle equipped with the driving assistance system or the autonomous driving system.Proceeding from this, it is an object of the invention to specify a method for ascertaining dangerous situations in the environment of an ego vehicle, which method offers improved accuracy and reliability of the prediction of the future position and / or driving trajectory of a third-party vehicle.The object is achieved by a method having the features of independent claim 1. Preferred embodiments are the subject matter of the dependent claims. A system for determining hazardous situations in the environment of an ego vehicle is the subject matter of independent patent claim 8.According to a first aspect, a method for ascertaining hazardous situations in the environment of an ego vehicle is disclosed. The vehicle has a sensor system for detecting a surrounding area of the vehicle. The method comprises the following steps:First, a third-party vehicle is detected in the surrounding area of the ego vehicle. This is preferably effected via the sensor system installed on the ego vehicle.Subsequently, angle information about the yaw angle, pitch angle, roll angle and / or sideslip angle of the third-party vehicle is acquired and / or received. At least a portion of this angle information can be detected via information provided by the sensor system of the ego vehicle. The receiving of at least a part of the angle information can take place via a wireless communication interface (e.g. V2X, etc.) from the third vehicle itself, from a further vehicle or from a permanently installed infrastructure unit (e.g. a traffic monitoring device on a road, etc.).The angle information is then processed by a computer unit of the ego vehicle. In particular, the angle information is used together with further information such as the current position, the speed and the acceleration of the third-party vehicle in order to estimate the future position and / or the travel trajectory of the third-party vehicle.Finally, the risk probability value is calculated based on the angle information. This risk probability value indicates the probability with which the third vehicle is at risk for the ego vehicle.The technical advantage of the proposed method is that, due to the additional use of the angle information, a more rapid and better prediction of the future position and movement of the third vehicle is possible and a hazardous situation can be detected more early, so that the driver assistance system or the autonomous driving system can warn the driver of the ego vehicle early and / or an emergency assistance system can initiate an accident avoidance maneuver, for example an avoidance maneuver or an emergency braking, in good time and precisely.According to one exemplary embodiment, the angular velocity of the yaw angle, pitch angle, roll angle and / or sideslip angle is detected and / or received. The angular velocity allows the determination of the time rate at which the yaw angle, pitch angle, roll angle and / or sideslip angle of the third-party vehicle changes. This makes it possible, for example, to detect abrupt braking of the third vehicle by the rapid change in the pitch angle, specifically before the speed of the vehicle changes.According to one exemplary embodiment, the future travel trajectory of the third-party vehicle by the ego vehicle is estimated on the basis of the angle information. It is thus possible, by comparing the travel trajectory of the third vehicle with the travel trajectory of the ego vehicle, to determine how high the risk potential is by the third vehicle.According to one exemplary embodiment, the probability of a collision of the ego vehicle with the third vehicle is estimated on the basis of an ascertained ego trajectory of the ego vehicle and the estimated future travel trajectory of the third vehicle. Based on this probability, a warning of the driver of the ego vehicle can take place and / or an accident avoidance maneuver, for example an avoidance maneuver or an emergency braking, can be initiated.According to one exemplary embodiment, the future movement state of the third-party vehicle is estimated by means of a kinematic model of the third-party vehicle in the ego vehicle. By means of the kinematic model of the third vehicle, it is possible to determine an expectable dynamics of the third vehicle, in particular with regard to the angle information of yaw angle, pitch angle, roll angle and / or sideslip angle. In the event that the detected and / or received angle information relating to the third-party vehicle deviates from the expected angle information beyond a predefined tolerance range, a vehicle state of the third-party vehicle may be assumed which indicates braking, skidding, etc., so that there is a potential risk for the ego vehicle, since there is a significant probability that the vehicle will not behave as originally assumed.According to one exemplary embodiment, the risk probability is calculated by means of a rule-based algorithm. The algorithm contains, for example, a predefined regulating mechanism with conditions. Depending on which conditions are fulfilled and / or how much a threshold value used in the regulating mechanism has been exceeded or undershot, a risk probability can be determined.According to one exemplary embodiment, the risk probability is calculated by means of a machine-learning calculation model. The machine-learning calculation model can be a neural network, for example. The machine-learning model is trained to determine a risk probability based on information provided by the sensor system and / or based on received information.According to the invention, the ground plane is estimated by the ego vehicle. Based on the estimated ground plane, a compensation of changes in the yaw angle, pitch angle and / or roll angle of the third-party vehicle takes place. In other words, a change in the yaw angle, the pitch angle and / or the roll angle of the third vehicle caused by the roadway course is determined by means of the ground plane estimate and subtracted from the detected and / or received angle information relating to the yaw angle, the pitch angle and / or the roll angle, so that the risk probability is determined at least substantially solely on the basis of angle information which is attributable to the vehicle dynamics independent of the roadway course. A more reliable information basis for the calculation of the risk probability is thus obtained.According to a further aspect, a system for ascertaining a hazardous situation in the environment of an ego vehicle is described. The system comprises at least one computer unit which is designed to perform the following steps:detecting a third-party vehicle in the surrounding area of the ego vehicle;detecting and / or receiving angle information relating to the yaw angle, pitch angle, roll angle and / or sideslip angle of the third-party vehicle;processing the angle information by a computer unit of the ego vehicle; andcalculating a risk probability value based on the angle information, wherein the risk probability value specifies the probability with which a risk from the third vehicle for the ego vehicle is imminent.In the context of the present disclosure, "third-party vehicle" is understood to mean a further vehicle which is different from the ego vehicle and is located in the surrounding region of the ego vehicle.The terms "approximately", "substantially" or "about" mean, for the purposes of the invention, deviations from the exact value in each case by + / - 10%, preferably by + / - 5%, and / or deviations in the form of changes which are insignificant for the function.Developments, advantages and possible applications of the invention also result from the following description of exemplary embodiments and from the figures. All features described and / or graphically depicted are fundamentally the subject matter of the invention, either alone or in any combination, independently of their summary in the claims or their reference back. The contents of the claims are also made part of the description.The invention is explained in more detail below on the basis of the figures using exemplary embodiments. The following are shown: FIG. 1 shows, by way of example, a schematic illustration of an ego vehicle having a sensor system for detecting a third-party vehicle in the surrounding region of the ego vehicle; FIG. 2 shows, by way of example and schematically, the representation of yaw angle, pitch angle and roll angle of the third-party vehicle; FIG. 3 is an exemplary illustration of the sideslip angle, the steering angle and the direction of the vehicle longitudinal axis with respect to a reference coordinate system; and FIG. 4 is a block diagram illustrating the method steps for ascertaining a hazardous situation in the environment of an ego vehicle by way of example.FIG. 1 schematically shows an ego vehicle 1 moving in the direction of travel FR on a roadway. The vehicle comprises a sensor system 2, by means of which a surrounding area 2.1 of the ego vehicle 1 can be detected. The sensor system 2 comprises one or more sensors which are configured to detect the surrounding area of the ego vehicle 1. The sensor system 2 can comprise, for example, one or more of the following sensors: radar sensor, LIDAR sensor, camera, stereo camera, ultrasonic sensor, etc.The ego vehicle 1 furthermore comprises a wireless communication interface 5, via which the ego vehicle 1 can receive information from an external data source or transmit information to an external data sink. The wireless communication interface 5 can be, in particular, a mobile radio interface which implements, for example, an LTE, LTE advanced or 5G mobile radio standard.The ego vehicle 1 can be designed to communicate with an external device 6 via the wireless communication interface 5, for example via a V2X communication channel (V2X: vehicle to analyzing). The external device 6 can be, for example, a third vehicle 3 or a stationary infrastructure unit, for example an infrastructure unit in the form of a traffic sign, a traffic light, etc., installed on the edge of the roadway using a sensor system.The ego vehicle 1 is designed to acquire angle information of a third-party vehicle 3 by means of the sensor system 2 and / or to receive it via the wireless communication interface 5. The angle information may include, for example, one or more of the following information:- yaw angle;- pitch angle;- roll angle;float angle.FIG. 2 outlines the yaw angle, the pitch angle and the roll angle. In a Cartesian coordinate system, the x-axis of which runs along the vehicle longitudinal axis, the y-axis of which runs along the vehicle vertical axis and the z-axis of which runs along the vehicle transverse axis, the roll angle refers to the angle about the vehicle longitudinal axis, the yaw angle refers to the angle about the vehicle vertical axis and the pitch angle about the vehicle transverse axis.In FIG. 3, the sideslip angle β is illustrated, wherein the sideslip angle β indicates the angle difference between the vehicle longitudinal axis and the motion vector of the third vehicle 3. ψ indicates the direction of the vehicle longitudinal axis relative to the x-axis of the reference coordinate system and δ indicates the steering angle of the steered wheels.In addition, the ego vehicle 1 can acquire and / or receive information on the rate of change of the yaw angle, the pitch angle, the roll angle and / or the sideslip angle of the third-party vehicle 3, i.e. the angular velocity of said angles. The detection can be carried out by a sensor system 2 of the ego vehicle 1 suitable for this purpose. Alternatively, the ego vehicle 1 may receive angular velocity information via the wireless communication interface 5, for example, via a V2X communication with the third-party vehicle 3.In addition, the ego vehicle 1 is designed to detect the local position, the direction of movement, the speed, the acceleration of the third vehicle 3 by means of the sensor system 2, in order to be able to estimate the future movement of the third vehicle 3 relative to the ego vehicle 1. The ego vehicle 1 can preferably also determine the object class of the third vehicle 3, wherein the object classes can comprise, for example, "car", "motorcycle", "truck" or "bicycle".Based on the acquired angle information, an algorithm implemented in a computer unit 4 of the ego vehicle 1 can determine the probability that the third-party vehicle 3 is at risk for the ego vehicle 1. The algorithm can be implemented, for example, as a rule-based algorithm or as a state machine. The algorithm may include a plurality of rules or conditions that are iteratively checked to determine the risk probability.A rule or condition of the algorithm can specify, for example, that a high risk probability is to be assumed for a third vehicle 3 driving ahead, which has a pitch angular velocity greater than a predefined threshold value, since this third vehicle 3 is expected to carry out a strong braking operation and it is thus to be assumed that the distance between the third vehicle 3 and the ego vehicle 1 will decrease sharply in the short time.A further rule or condition of the algorithm can specify that, in the event that the slip angle of a third-party vehicle 3 is greater than a threshold value, it is to be assumed that the tires of the third-party vehicle 3 have lost adhesion and the third-party vehicle 3 is thus moving or sliding in an uncontrolled manner, with the result that a high risk probability is to be assumed. As a further condition, for example, the external temperature and / or the roadway state (e.g. wet or dry) can be used in order to derive therefrom, for example, that the third-party vehicle 3 has slid on a wet or icy roadway.As an alternative to a rule-based algorithm or a state machine, a machine-learning algorithm can be used in a computer unit 4 of the ego vehicle 1 to determine the risk probability. The machine learning algorithm may have been trained with training data (e.g., labeled sensor information) in a training phase. The training data can comprise, in particular, sensor information which comprises third-party vehicles in different driving situations. The training data also comprise information on the probability of risk from the third-party vehicle 3. As a result, the machine-learning algorithm can estimate in an inference phase whether and with which probability a risk originates from a third-party vehicle 3.The ego vehicle 1 is preferably designed to estimate the roadway plane, in particular also in the area in which the third-party vehicle 3 is located. As a result, it is possible to detect vehicle movements caused by the roadway course and to compensate or suppress the changes in the yaw angle, the pitch angle, the roll angle that result from the roadway course when calculating the risk probability. Thus, for example, a change in the roll angle of the third vehicle 3 when the transverse inclination of the roadway changes cannot be an indication of a rollover of the third vehicle 3, whereas a significant change in the roll angle when driving through a curve can be an indication of a risk of a rollover.The algorithm by means of which the risk probability existing by the third-party vehicle 3 is determined can implement a kinematic model of the third-party vehicle 3. By means of the kinematic model of the third vehicle 3, in the current driving state of the vehicle (e.g. speed, acceleration, steering angle during cornering and / or road inclination, etc.), expectable angle information (e.g. the expectable yaw angle, the expectable pitch angle, the expectable roll angle and / or the expectable float angle) can be calculated. This calculation based on the kinematic model can be carried out continuously or intermittently. The angle information recorded by the sensor system 2 can be compared with the angle information that can be expected. By means of the comparison, driving states of the third vehicle 3 can be determined which show a risk potential for the ego vehicle 1, for example a strong braking operation by means of an abrupt change in the pitch angle, a skidding of the vehicle by means of a large sideslip angle and / or a strong change in the sideslip angle.The algorithm for determining the risk potential can take into account the object class of the third-party vehicle 3, namely in such a way that a kinematic model is used for the calculation of the kinematic state of the third-party vehicle 3, which model is adapted to the object class of the third-party vehicle 3. For example, if the third vehicle 3 is a passenger car, a kinematic model of a passenger car can be used, whereas a truck is used as the third vehicle a kinematic model of a truck.According to a preferred exemplary embodiment, the ego vehicle 1 is designed to estimate the state of movement of the third-party vehicle and / or the future travel trajectory of the third-party vehicle 3. A plurality of travel trajectory estimates and a probability value for the respective estimated travel trajectories are preferably determined, which probability value indicates how likely the third-party vehicle 3 is to travel along the respective travel trajectory. In other words, a set of estimated travel trajectories and probabilities assigned to these are determined. Based on these travel trajectories, it is possible to determine how high the risk probability by the third-party vehicle 3 is, for example how close the third-party vehicle 3 and the ego vehicle 1 come, whether braking and / or avoidance is necessary, etc.Taking into account the angle information leads to a earlier and more accurate estimation of the risk of the third-party vehicle 3 to the ego vehicle 1, as a result of which the autonomous driving system or a driving assistance system can perform an earlier and possibly more appropriate driving maneuver than in the case of not considering the angle information of the third-party vehicle 3. Furthermore, an emergency assistance system can be activated earlier, an earlier warning can be issued to the driver, measures before the impact and / or driving maneuvers such as avoidance or emergency braking can be initiated earlier.FIG. 4 shows a flow chart which illustrates the method steps for ascertaining a hazardous situation in the environment of an ego vehicle.First, a third-party vehicle in the surrounding area of the ego vehicle is detected (S 10).Subsequently, angle information in the form of the yaw angle, the pitch angle, the roll angle and / or the sideslip angle of the third-party vehicle is acquired and / or received (S 11).The angle information is then processed by a computer unit of the ego vehicle (S 12).Finally, a risk probability value is calculated based on the angle information, wherein the risk probability value specifies the probability with which a risk for the ego vehicle originates from the third-party vehicle (S 13).The invention has been described above with reference to exemplary embodiments. It is to be understood that numerous changes and modifications are possible without thereby departing from the scope of protection defined by the claims.List of reference characters1 Ego vehicle 2 Sensor system environment 2.1 3 Third vehicle 4 Computer unit 5 Wireless communication interface 6 External device FR Direction of travel

Claims

Method for determining a hazardous situation in the environment of an ego vehicle (1), wherein the ego vehicle (1) has a sensor system (2) for detecting an environmental region of the ego vehicle (1), wherein the method comprises the following steps: - detecting a third vehicle (3) in the environmental region of the ego vehicle (1) (S10); - detecting and / or receiving angle information relating to the yaw angle, pitch angle, roll angle and / or sideslip angle of the third vehicle (3) (S11); - processing the angle information by a computer unit (4) of the ego vehicle (1) (S12); and - calculating a risk probability value based on the angle information, wherein the risk probability value specifies the probability with which a risk from the third vehicle (3) for the ego vehicle (1) is imminent (S 13), wherein a ground plane is estimated by the ego vehicle (1) and wherein based on the estimated ground plane a compensation of changes in the yaw angle, pitch angle and / or roll angle of the third vehicle (3) takes place.Method according to Claim 1, characterized in that the angular velocity of the yaw angle, pitch angle, roll angle and / or sideslip angle is detected and / or received.Method according to Claim 1 or 2, characterized in that the future travel trajectory of the third-party vehicle (3) through the ego vehicle (1) is estimated on the basis of the angle information.Method according to Claim 3, characterized in that the probability of a collision of the ego vehicle (1) with the third-party vehicle (3) is estimated on the basis of an ascertained ego trajectory of the ego vehicle (1) and the estimated future travel trajectory of the third-party vehicle (3).Method according to one of the preceding claims, characterized in that the future movement state of the third-party vehicle (3) is estimated by means of a kinematic model of the third-party vehicle (3) in the ego vehicle (1).Method according to one of the preceding claims, characterized in that the risk probability is calculated by means of a rule-based algorithm.Method according to one of the preceding claims, characterized in that the risk probability is calculated by means of a machine-learning calculation model.System for determining a hazardous situation in the environment of an ego vehicle (1), wherein the system comprises at least one computer unit (4) which is designed to carry out the following steps: - detecting a third-party vehicle (3) in the surrounding region of the ego vehicle (1); - detecting and / or receiving angle information relating to the yaw angle, pitch angle, roll angle and / or sideslip angle of the third-party vehicle (3); - processing the angle information by a computer unit (4) of the ego vehicle (1); and - calculating a risk probability value based on the angle information, wherein the risk probability value specifies the probability with which a risk is being posed by the third-party vehicle (3) for the ego vehicle (1)), wherein a ground plane is estimated by the ego vehicle (1) and wherein based on the estimated ground plane a compensation of changes in the yaw angle, pitch angle and / or roll angle of the third-party vehicle (3) takes place.

Images