DE102012222715B4 - Device and method for the predictive detection of a vehicle collision - Google Patents

Abstract

Method for predictively detecting a collision of a vehicle (300), comprising the following steps: emitting (400) pulses (106) of electromagnetic radiation by means of a transmitter (102) attached to the vehicle (300); receiving (402) reflected pulses (107), by means of a multi-channel sensor (108) attached to the vehicle (300), from a plurality of directional channels (110); determining (420), based on the reflected pulses (107), a group (800) of adjacent directional channels (110) in which the reflected pulses (107) were reflected from a same object (200); determining (424) time-to-collision information (150) for the object (200) based on in the directional channels (110) of the group (800) received reflected pulses; and judging (428, 430, 434) whether a collision will occur based on the time-to-collision information (150).

Description

The present invention relates to a method and a device for predictive detection of a vehicle collision.

The DE 10 2005 021 225 A1 describes a method and a device for recognizing the surface condition of objects of road traffic or people. In this case, signals are emitted from a vehicle which are reflected by a person or an object, the reflected signals being evaluated in the vehicle. The emitted signals are transmission pulses, the echo pulses of which are received, the signal characteristics of the echo pulses being used to determine the surface condition and from this whether a person or an object is positioned in front of the vehicle.

The DE 10 2004 032 048 A1 discloses a method and a device for determining the alignment of a first object with a second object on the basis of emitted and reflected echo pulses.

The EP 1 600 793 A2 discloses a radar sensor for motor vehicles for collision avoidance using transmitted and reflected radar pulses. To increase the redundancy for safety-relevant applications, the transmitted and received radar radiation is emitted in at least two or more different types of modulation and evaluated in a receiving part.

Under the title "Device and procedure to predict and respond to an accident" disclosed US 6 012 008 A. which is an active detection system for predicting a collision in order to be able to trigger an impact cushion before the collision occurs. The acquisition is based on the reflection of pulse signals (eg radar, laser), the transit times of which are measured. In order to rule out the generation of false messages, the detection is limited to a detection envelope area near the vehicle surface, which allows collision warnings to be given approximately 10 to 40 ms before a collision occurs.

While this period is sufficient to deploy an impact cushion, others require useful measures in the event of a collision, such as the tightening of belts, adjusting the head restraints, giving a warning tone to the driver, or raising any open side windows, however, requires considerably more time to use. Therefore, it is desirable to provide an early warning of an impending collision without increasing the likelihood that false messages will be generated.

Disclosure of the invention

The above object is achieved by a method for predictive detection of a vehicle collision according to claim 1, and likewise by a device for predictive collision detection for a vehicle according to claim 14.

The inventive device for predictive collision detection for a vehicle comprises a pulse transmitter for emitting pulses of electromagnetic radiation from the vehicle, a multi-channel sensor with a plurality of directional channels for receiving reflected pulses on the vehicle, and a grouper for determining a group based on the reflected pulses reflected pulses from adjacent directional channels in which the reflected pulses were reflected by a same object. The apparatus further includes a time-to-collision determiner for determining time-to-collision information for the object based on reflected pulses received in the directional channels of the group, and an assessor for judging whether a collision will occur on the time-to-collision information. The use of information from a plurality of directional channels makes it possible to detect a collision with particularly high reliability even when there is a large distance between the vehicle and the potential collision object. Measures such as closing windows or tightening belts that require a long period of time compared to an impact cushion can thus be used safely. Furthermore, the logical grouping of information from a plurality of directional channels limits the amount of data processing, so that powerful processors and the like are not required, which makes it possible to reduce the production costs.

The pulse transmitter preferably has a light-emitting diode. This makes it possible to implement the predictive collision detection device in a compact form that is easy to manufacture, consumes little energy and e.g. fits under a cover of a rear light or other light of the vehicle, so that the manufacture of the vehicle can be simplified and its design can be refined. If you use an infrared light-emitting diode in particular, the effects of adverse weather conditions such as e.g. Fog on the operation of the device.

The inventive method for predictive detection of a collision Vehicle comprises steps of emitting pulses of electromagnetic radiation by means of a transmitter attached to the vehicle, receiving reflected pulses by means of a multichannel sensor attached to the vehicle from a plurality of directional channels, and determining, based on the reflected pulses, a group of adjacent directional channels in which the reflected pulses were reflected from a same object. It further includes steps of determining time-to-collision information for the object based on reflected pulses received in the directional channels of the group and assessing whether a collision will occur based on the time-to-collision information.

Preferably, the step of judging whether a collision will occur includes judging whether the time to collision is shorter than a predetermined time to collision threshold according to the time to collision information. In this way, the time-to-collision threshold can be selected in accordance with a required operating time for measures that are to be used in the event of an assessment that a collision is imminent.

According to a preferred development, the method further comprises a step of determining a reflection distance for each reflected pulse, based on a transit time from sending out to receiving the reflected pulse, the step of determining the group being carried out by selecting adjacent direction channels in which the received reflected pulses correspond to essentially the same reflection distance. As a result, the object can be reliably distinguished from other objects that are in the same direction but at a different distance from the vehicle, e.g. of a second object, which is partially behind the object corresponding to the group.

According to a preferred development, the method further comprises a step of rejecting reflected pulses which correspond to a reflection distance greater than a predetermined maximum distance. This enables the prediction of a collision with an object that is so far away that the driver of the vehicle - or, if the object is another vehicle, the driver of the other vehicle - can still be prevented by changing the driving speed and / or direction can react so that a collision is avoided. For example, data corresponding to said pulses are deleted from a memory or marked as to be disregarded in subsequent method steps. Additionally or alternatively, reflected pulses which correspond to a reflection distance smaller than a predetermined minimum distance are rejected in a similar manner, which leads to incorrect predictions of a collision based on close proximity due to fog, rain and the like. reflected pulses.

According to a preferred development, the steps of transmitting, receiving and determining the reflection distance are carried out repeatedly for each reflected pulse in order to provide a reflection distance function over time for each directional channel. This makes it possible to increase reliability by basing predictions on more data. Preferably, the method further comprises a step of smoothing each reflection distance function. In this way, discontinuities, e.g. this is because potential collision objects appear, disappear or reform, causing the false appearance of an object approaching at high speed, which makes it possible to avoid incorrect predictions of a collision.

According to a preferred development, the method further comprises a step of determining a reflection direction speed function for each direction channel by differentiating the reflection distance function for the respective direction channel over time. This enables particularly reliable predictions by evaluating the relative speed of an object approaching the vehicle based on data from both the present and the past.

According to a preferred development, the step of determining the group is carried out by selecting adjacent direction channels in which the respective reflection direction speed functions indicate an essentially identical reflection speed. This enables the object to be reliably distinguished from other objects that are positioned in a similar direction but move at a different speed relative to the vehicle, e.g. from a second object next to the object corresponding to the group.

According to a preferred development, the method further comprises a step of determining a reflection direction acceleration function for each direction channel by differentiating the reflection direction speed function for the respective direction channel over time. Here, the step of judging whether a collision will occur includes judging whether an object's impact speed on the vehicle will be greater than a predetermined impact speed threshold based on the respective reflection direction acceleration values. Functions and the time-to-collision information. This avoids, for example, incorrect predictions of a collision if the object slows down as it approaches the vehicle at high speed, such as if the object is another vehicle that is being decelerated by its driver.

According to a preferred development, the method further comprises a step of determining an object center position. The step of determining the time-to-collision information is carried out based on the reflection distance function and the reflection direction speed function for a direction channel corresponding to the object center position. A particularly reliable prediction is made possible in this way because the object center position will generally coincide approximately with the center of gravity of the object, the future movement of which can be predicted with the greatest precision based on existing data with regard to distance, speed and preferably acceleration.

According to a preferred development, the method further comprises a step of determining a pulse quality index for each reflected pulse in the group. The step of determining the object center position is carried out on the basis of the reflected pulse with the highest pulse quality index among the reflected pulses of the group. This leads to a particularly high reliability of predictions, since a high pulse quality indicates that the reflected pulse is probably not due to an artifact.

According to a preferred development, the method further comprises a step of determining a transverse speed of the object center position in a direction transverse to the directional channel corresponding to the object center position. The step of judging whether a collision will take place includes judging whether the lateral speed is lower than a predetermined lateral speed threshold. This makes it possible to avoid false predictions of a collision by setting the cross-speed threshold so that it is not judged that a collision will take place if the cross-speed is sufficiently large for the object, the area covered by the directional channels within the time-to range -Collision information to leave displayed time until collision.

Figure list

• 1 10 is a schematic block diagram of a predictive collision detection device according to an embodiment of the invention;
• 2nd shows a schematic side view of a vehicle according to an embodiment, which the device for predictive collision detection 1 installed, with another vehicle behind it;
• 3rd shows a schematic plan view of a vehicle according to an embodiment, which consists of several copies of the device for predictive collision detection 1 installed and drives in the middle of other vehicles on a street;
• 4th 10 is a flowchart of a method for predictively detecting a vehicle collision in accordance with an embodiment;
• 5 illustrates a step of discarding reflection distance information within the method 4th ;
• 6 illustrates a step of smoothing a reflection removal function within the process 4th ;
• 7 Figure 11 is a graph showing a step of differentiating a reflection direction velocity function within the method 4th , illustrated;
• 8th is a schematic top view illustrating steps within the method of determining a group of adjacent directional channels in which the reflected pulses were reflected from a same object and determining an object center position 4th ; and
• 9 FIG. 12 is a schematic plan view illustrating a step of determining a lateral speed component of the object center position in a direction transverse to that of the object center position within the method 4th .

Unless stated otherwise, the same reference symbols designate the same elements throughout the figures.

Detailed description of the embodiments

1 shows, as a schematic block diagram, an apparatus 100 for predictive collision detection for installation in a vehicle 300 (shown in 2nd , not in 1 ), together with a potential collision object 200 , which is shown here as another vehicle. The device 100 for predictive collision detection comprises a detector device 101 , a control unit 120 and several actuators 191-194 to carry out different actions in the event of a collision, such as closing side windows of the vehicle (first actuator 191 ), Setting the Backrests of seats (second actuator 192 ), Generate a warning tone (third actuator 193 ) or pretensioning belts (fourth actuator 194 ).

The detector device 101 that, for example, with a backward direction in the vehicle 300 can be installed, includes an infrared light emitting diode 104 and a pulse transmitter including a lens 102 for sending regular pulses 106 infrared light in a detection field that, for example, essentially the area behind the rear of the vehicle 300 can cover. 3rd shows an example of a vehicle in which two more incarnations 100 ' , 100 " the device 100 for predictive collision detection 1 are built in when driving on a road 204 Corresponding detection fields for detecting vehicles driving in the blind spot on the driver's side 200 ' or for detecting vehicles 200 " cover that in front of the vehicle 300 drive.

The pulse transmitter 102 the detector device 101 is from a detector device processor 105 controlled in which different programs 103 , 126 to operate the detector device 102 are saved. Furthermore, the detector device 101 a multi-channel sensor 108 on, also with the detector device processor 105 is connected and controlled by this and an infrared filter 109 , a lens 111 and an infrared light-sensitive detection plate including a plurality of infrared light-sensitive sections, each of which has a directional channel 110 for receiving reflected pulses 107 represented from a certain direction within the detection field. For example, the detection field can be fan-shaped with an opening angle of 40 ° in a horizontal plane above the floor, divided into, for example, sixteen directional channels of 2.5 ° each, as in FIG 8th and 9 shown. The detector device processor 105 is designed to execute programs that have the functionality of a distance determining device 103 to determine a reflection distance 202 , illustrated in 2nd , from the detector device 101 to a reflective object 200 from a runtime of the transmitted pulses 106 from transmission to reception as reflected pulses 107 , and a pulse quality index determination device 126 to determine an index indicating the quality of the reflected pulses, both separately for each directional channel 110 .

The control unit 120 includes a control unit processor 121 with several saved program blocks 112 , 113 , 115-119 , 122-124 , to predict whether a collision between a potential collision object 200 and the vehicle will take place where the device 100 for predictive collision detection is built in, based on that from the detector device 101 provided reflection distance and pulse quality index data. The control unit 120 can be in the same place as the detector device 101 installed, as exemplified in 2nd shown, or as software together with other software on a computer or control device of the vehicle which also serves other purposes 300 . The individual program blocks 112 , 113 , 115-119 , 122-124 are to be explained below.

The control unit 120 includes a rejection device 113 to discard all data as in 5 shown, the reflection distances are smaller than a minimum distance 502 of 0.3 m, for example, and reflection distances greater than a maximum distance 500 of, for example, 50 m after the data for the reflection distance 202 from the detector device 101 were received. This will be a close range 504 and a wide area 506 excluded from the detection field. Also included is a smoothing device 115 for smoothing the separately per directional channel from the detector device 101 via the rejection device 113 as a function over time 600 received stream of reflection distance data 202 , as in 6 illustrates that there is a smooth reflection distance function 202 ' results. The control unit 120 further comprises a distance differentiating device 117 to differentiate the smoothed reflection distance function 202 ' over time 600 , so that a reflection direction velocity function 700 per directional channel 110 results, illustrated in 7 which the to the direction of the directional channel 110 even parallel speed component of the reflecting object in the respective directional channel 110 indicates. The control unit 120 further includes a speed differentiator 118 to differentiate the reflection direction velocity function 700 over time 600 , so that for each directional channel 110 a reflection direction acceleration function 702 over time 600 results in minima and maxima 704 the reflection direction velocity function 700 is zero.

The control unit 120 also includes a grouper 112 to determine how in 8th illustrates a group 800 of directional channels 110 , for which the respective reflection distance function 202 indicates the same reflection distance, ie in the respective object positions 802 at a much the same reflection distance 202 from the detector device 202 the device 100 were detected for predictive collision detection. The term “same” is to be interpreted here with the meaning of not differing from one another by more than one tolerance value, which can be determined in advance on the basis of a typical dimension of a potential collision object to be expected in a given application, such as a typical dimension of a motor vehicle, for example 2 m, in the case of an application in the motor vehicle sector as described for the present embodiment in 2nd and 3rd is shown. In addition or as an alternative to using the reflection removal functions 202 for the individual directional channels the grouper can 112 configured, also the reflection direction velocity function 700 and / or the reflection direction acceleration function 702 as criteria for determining the group 800 to use.

The control unit 120 further comprises an object center position determination device 119 to determine a center of an object 804 that as close to an actual center of gravity of the object 200 Is accepted. In the present embodiment, the object center position determination device is 119 trained that directional channel 110 to select in which the reflected pulse is currently 107 with the highest quality index, as by the detector device 101 provided, is received. An alternative configuration can provide a directional channel 110 at the center of the group 800 or, in the case of an even number of directional channels in the group 800 , an imaginary position between the two directional channels in the center of the group 800 than the center position 804 to select.

The control unit 120 further includes a time-to-collision detector 123 configured from the reflection distance and reflection direction velocity function 202 , 700 in the directional channel (s) 100 the center position 804 time-to-collision information 150 to calculate, which indicates a period of time after which that of the group 800 represented object 200 will collide with the vehicle on the assumption that the directional channel 110 the center position 804 and the direction of reflection speed 700 stay the same. Also included is a transverse speed determination device 124 to determine a transverse speed of the center position 804 of the object 200 in a direction transverse to the directional channel corresponding to the object center position 110 .

The control unit also includes 120 an assessor 116 to judge whether a collision of the object 200 with the vehicle 300 will take place based on the time-to-collision information 150 by the transverse speed determination device 124 determined transverse speed and the reflection direction acceleration function 702 .

Operation of the device 100 for predictive collision detection in the following with reference to an in 4th specified flow chart are explained. After switching on in step 442 sends the detector device 101 in step 400 regular pulses of infrared light, eg 30 pulses per second on a wavelength of approx. 900 nm, in the detection field behind the vehicle 300 out. In step 402 receives the detector device 101 on the multi-channel sensor 108 reflected pulses 107 in a plurality of directional channels 110 . It is believed that reflected pulses 107 only in a subset of the full number of directional channels of the multi-channel sensor 108 be received. In step 404 the processor calculates 101 the detector device 101 a reflection distance 202 separate for each directional channel 110 from which a reflected pulse 107 was received. In step 406 the processor calculates 101 the detector device 101 a pulse quality index for each directional channel 110 from which a reflected pulse 107 was received. Data with the numbers of the directional channels 110 , Values for the reflection distance 202 and pulse quality index values are sent to the control unit 120 continuously from the detector device 101 fed.

In step 408 all information related to objects that are considered unimportant is deleted because they are either too close to the vehicle 300 or too far from it. This applies to all data where the reflection distance is less than 0.3 m and where it is more than 50 m. These values can of course be implemented as parameters, simply depending on, for example, the type of vehicle 300 and the installation position of the detector device 101 inside the vehicle 300 to be configurable.

In step 410 are for each directional channel 110 the corresponding values of the reflection distance 202 that are not in step 408 were treated as a function over time and subjected to smoothing, in particular to discrepancies as in 6 shown eliminating, resulting in a smooth reflection removal function 202 ' results. If a detected object is covered by a closer object, the distance value in the respective channel suddenly drops, causing a discontinuity as if an object had approached very quickly. To smooth such discontinuities, for example, a filter with a moving average is used, which has a buffer with eight time steps, works according to a "first in - first out" principle and per directional channel 110 is applied.

In step 412 becomes the smoothed reflection distance function 202 ' separately per directional channel 100 after the time 600 differentiated to a reflection direction velocity function 700 for the directional channel, which is a speed component of the object parallel to the Direction of the directional channel 110 corresponds. Negative speed results that indicate an object moving away from the vehicle are ignored. Since relative speeds are calculated, the speed of the vehicle relative to the ground surface is irrelevant. In step 414 A moving average filter that has an eight time step buffer smoothes the relative reflection direction velocities 700 . As in step 410 the filter works according to a "first in - first out" principle and is used for each directional channel 110 applied.

In step 416 the smoothed reflection direction velocity function is separated per direction channel 110 after the time 600 differentiates to a reflection direction acceleration function 702 for the directional channel 110 to compute an acceleration component of the object parallel to the direction of the directional channel 110 corresponds. This provides information about whether the object is getting faster or slowing down. In step 418 A moving average filter that has an eight time step buffer smoothes the relative reflection direction accelerations. As in step 410 and 414 the filter works according to a "first in - first out" principle and is used for each directional channel 110 applied.

In step 420 a group of directional channels is determined which correspond to the same reflecting object as in 8th illustrated. Approaching objects, such as other vehicles, generally occupy more and more directional channels 110 in the course of their rapprochement. Adjacent directional channels, the object data with the same direction of reflection speed 700 and reflection distance 202 are represented as a single object 200 logically grouped. Corresponding difference threshold values which dictate when reflection direction speeds 700 and reflection distances 202 are to be considered as the same, are defined as parameters in advance.

In step 422 is under the directional channels 110 the directional channel where the quality index is highest, ie the signal quality of the reflected pulses 107 is highest as a center position 804 of the object 200 identified accordingly. This ensures sufficient signal availability over the next steps. Furthermore, it can be assumed that the point with the highest quality represents the projected center of gravity of the object in most cases.

Using data from step 414 and 422 will in step 424 calculates time-to-collision information that indicates when the object is on the vehicle 300 will bounce, assuming no changes in direction and speed. That from the time-to-collision information 150 The time window presented is continuously updated.

In step 426 becomes a transverse speed component of the object center position in a direction parallel to the ground and perpendicular to the direction of the directional channel 110 the center position 804 calculated. An object with a significant transverse speed will change between directional channels over time. To the movement of the object 200 across the directions of the directional channels 110 to calculate the change is center position 804 observed. A high cross speed reveals that the object 200 not with the vehicle 300 will collide. So shows 9 an example where a group 800 over time to a new position 800 ' changes, the places of reflection 802 ' in different directional channels 110 includes. The speed vector 900 between the respective center positions 804 , 804 ' is called the speed vector of the object 200 relative to the vehicle 300 treated.

In decision step 428 becomes the time-to-collision information 150 compared to a reflection direction threshold for the use of a desired action. The reflection direction threshold value is a parameter value which is predetermined in accordance with the triggering times of the respective desired actions, for example 200 ms for an electromechanical belt tensioner. This step checks whether a collision is to be expected within the time given by the reflection direction threshold, without a possible deceleration of the vehicle or a speed component transverse to the directions of the direction channels 110 to consider. In the positive case ("Y") decision step goes 428 with decision step 430 away. In the negative case ("N") there is a restart 440 of the procedure.

In step 430 will that in step 426 calculated cross-speed compared to a cross-direction threshold, that of the reflection direction speed 700 depends. This step evaluates whether the lateral movement of the object is sufficiently small to cause a collision of the object 200 with the vehicle 300 unavoidable. In the positive case ("Y") decision step goes 430 with step 432 away. In the negative case ("N") there is a restart 440 of the procedure. For example, in 9 the speed vector 900 such a significant component can be seen transversely to the directions of the directional channels 110 (ie approximately perpendicular to the center line of the vehicle 300 ) that a collision is not to be expected.

In step 432 is calculated using the reflection direction acceleration from step 416 and the time-to-collision information 150 from step 424 a collision speed is calculated. In decision step 434 it is determined whether the in step 423 calculated collision speed is greater than a predetermined collision speed threshold. In the positive case ("Y") decision step goes 434 with step 436 in which the actions intended for the event of a collision are used. In the negative case ("N") there is a restart 440 of the procedure. Thus, the decision steps combine 428 , 430 , 434 cumulative tests against three thresholds, before making the final deployment decision.

Claims (16)

Method for predictively detecting a collision of a vehicle (300), comprising the following steps: Emitting (400) pulses (106) of electromagnetic radiation by means of a transmitter (102) attached to the vehicle (300); Receiving (402) reflected pulses (107) by means of a multi-channel sensor (108) attached to the vehicle (300) from a plurality of directional channels (110); Determining (420), based on the reflected pulses (107), a group (800) of adjacent directional channels (110) in which the reflected pulses (107) were reflected by a same object (200); Determining (424) time-to-collision information (150) for the object (200) based on reflected pulses received in the directional channels (110) of the group (800); and Judge (428, 430, 434) whether a collision will occur based on the time-to-collision information (150). Procedure according to Claim 1 wherein the step of judging (428, 430, 434) whether a collision will occur includes judging (428) whether the time to collision is shorter than a predetermined one according to the time-to-collision information (150) Time-to-collision threshold. Procedure according to Claim 1 or 2nd , further comprising a step of determining (404) a reflection distance (202) for each reflected pulse based on a transit time from sending (400) to receiving (402) the reflected pulse (107), the step of determining (420) the group (800) is carried out by selecting adjacent directional channels (110) in which the received reflected pulses (107) correspond to a substantially equal reflection distance. Procedure according to Claim 3 , further comprising a step of rejecting (408) reflected pulses (107) which correspond to a reflection distance (202) greater than a predetermined maximum distance (500), and / or reflected pulses which correspond to a reflection distance (202) less than a predetermined minimum distance ( 502). Procedure according to Claim 3 or 4th , wherein the steps of transmitting (400), receiving (402) and determining (404) the reflection distance (202) are repeated (440) for each reflected pulse to determine a reflection distance (202) for each directional channel (110) Function over time (600). Procedure according to Claim 5 further comprising a step of smoothing (410) each reflection distance (202) function. Procedure according to Claim 5 or 6 further comprising a step of determining (412) a reflection direction velocity (700) function for each direction channel (110) by differentiating the reflection distance (202) function for the respective direction channel (110) by time (600). Procedure according to Claim 7 wherein the step of determining (420) the group (800) is performed by selecting adjacent direction channels (110) in which the respective reflection direction speed (700) functions indicate a substantially equal reflection speed (700). Procedure according to Claim 7 or 8th further comprising a step of determining (416) a reflection direction acceleration (702) function for each direction channel (110) by differentiating the reflection direction velocity (700) function for the respective direction channel (110) by time (600), the The step of judging (428, 430, 434) whether a collision will occur includes judging (434) based on the respective reflection direction acceleration (702) functions and the time-to-collision information (150) whether there is a Impact speed of the object (200) on the vehicle (300) will be greater than a predetermined impact speed threshold. The method of any preceding claim, further comprising a step of determining (422) an object (200) center position (804), the step of determining (424) the time-to-collision information being performed based on the reflection distance - (202) function and the reflection direction velocity (700) function for a direction channel (110) corresponding to the object center position. Procedure according to Claim 10 , further comprising a step of determining (406) one Pulse quality index for each reflected pulse of the group (800), the step of determining (422) the object center position (804) being performed based on the reflected pulse with the highest pulse quality index among the reflected pulses of the group (800). Procedure according to Claim 10 or 11 further comprising a step of determining (426) a cross speed of the object (200) center position (804) in a direction transverse to the directional channel corresponding to the object center position (804), the step of judging (428, 430, 434 ) to assess (430) whether a collision will take place includes whether the lateral speed is less than a predetermined lateral speed threshold. Computer program product with computer-executable instructions which, when executed by a computer or a control device (120) of a vehicle (300) in which a pulse transmitter (102) and a multi-channel sensor (108) are installed, cause the computer or the control device (120) in Connection to the pulse transmitter (102) and the multi-channel sensor (108) to carry out the steps of the method according to one of the preceding claims. Device (100) for predictive collision detection for a vehicle (300), comprising: a pulse transmitter (102) for sending pulses (106) of electromagnetic radiation from the vehicle (300); a multi-channel sensor (108) with a plurality of directional channels (110) for receiving (402) reflected pulses (107) on the vehicle (300); a grouper (112) for determining, based on the reflected pulses (107), a group (800) of reflected pulses (107) from adjacent directional channels (110) in which the reflected pulses were reflected from a same object (200); a time-to-collision detector (123) for determining time-to-collision information (150) for the object (200) based on reflected pulses (107) received in the directional channels (110) of the group (800) ; and a judge (116) for judging whether a collision will take place based on the time-to-collision information (150). Device (100) for predictive collision detection Claim 14 , wherein the pulse transmitter (102) has a light emitting diode (104), in particular an infrared light emitting diode. Vehicle (300) in which the device (100) for predictive collision detection according to Claim 14 or 15 is installed to detect a collision at a rear of the vehicle (300).

Images