Determination method for predicting injuries, and associated computing unit and system

Abstract

The invention relates to a determination method for determining a prediction of an injury to a road at-risk user (112) in the event of a collision with a vehicle (104), as well as to an associated central computing unit (100) and to an associated determination system. Here, the determination method comprises the steps of receiving (S2) sensor data from sensor devices (105) of a vehicle (104) concerning a collision of the vehicle (104) with the at-risk road user (112), determining (S3) impact parameters on the basis of the sensor data, transmitting (S4), by means of the vehicle (104), the impact parameters to a central computing unit (100), the computing unit (100) comprising trained artificial intelligence (102) which has been trained to determine injury parameters of the at-risk road user (112) as output data on the basis of the impact parameters as the input data, inputting (S5) the impact parameters into the trained artificial intelligence (102), and receiving injury parameters of the at-risk road user (112) as output data and outputting (S7) the impact parameters to an external unit.

Description

[0001] Determination methods for predicting injuries, as well as associated computing unit and system

[0002] Technical field

[0003] The present invention relates to a determination method for determining a prediction of injury to a vulnerable road user in a collision with a vehicle, as well as an associated central computing unit and an associated determination system.

[0004] State of the art

[0005] Systems for determining the injuries of road users in an accident are currently limited primarily to assessing the injuries of the driver of the vehicle involved in the crash. Injuries to vulnerable road users, such as pedestrians or cyclists, are often overlooked. Consequently, no relevant information can be transmitted to emergency services or relevant facilities, such as hospitals, and valuable time is wasted in treating the injured road user. For example, it is often impossible to determine which emergency services are needed to care for the injured road user or which facilities are suitable for their subsequent treatment.

[0006] The eCall system, which is standard equipment in new vehicles, also only transmits information that may be used to provide care for the vehicle occupants.

[0007] Description of the invention

[0008] The present invention relates, in a first aspect, to a method for determining the likelihood of injury to a vulnerable road user. As described below, the prediction of injury can, for example, involve predicting the severity of the injury, the type of injury, or the severity of injury to a specific part of the vulnerable road user's body. In this context, a vulnerable road user is not the occupant of the vehicle discussed further below, but rather a road user at risk, such as a pedestrian or cyclist, who might collide with the vehicle.

[0009] The process begins with the step of receiving sensor data from a vehicle's sensor systems regarding a potential collision with a road user at risk. This reception can involve, for example, receiving a transmission signal from the sensor systems or reading a corresponding signal from a memory. The sensor systems can be standard in-vehicle sensors such as camera systems, lidar systems, accelerometers, etc., which are already present in modern vehicles for driver assistance systems. Therefore, receiving sensor data can simply be the standard processing of sensor data within a driver assistance system.In this context, sensor data relating to a collision between the vehicle and the endangered road user can include, for example, sensor data received at the time of the collision or impact of the endangered road user, as well as data received shortly before or shortly after this collision.

[0010] The process then includes the step of determining impact parameters based on sensor data. Such determination can, for example, involve processing within the vehicle's perception system. Here, the received sensor data, such as images captured by a camera, can be processed and evaluated using, for example, the perception model of the perception system. This determination can, for instance, lead to the classification of objects, the determination of the vehicle's relative speed to the endangered road user during the impact, the impact angle, or even the movement of the endangered road user after the impact.Such processing of sensor data, referred to here as determination, is, as previously described, often already carried out in modern vehicles as part of the perception system used for driver assistance systems. Accordingly, the impact parameters determined here can include, for example, the classification of the object, more precisely whether it is a pedestrian or cyclist, the relative speed during the impact, the impact angle, or similar parameters.

[0011] The determination process then involves transmitting the impact parameters to a central processing unit. This transmission could, for example, be wireless via a suitable radio standard. The central processing unit could be, for instance, a cloud system. This central processing unit incorporates a trained artificial intelligence. This artificial intelligence could, for example, be a neural network trained using supervised learning. The network was trained using training data, which includes the impact parameters as input and injury parameters as output, to determine the corresponding injury parameters based on the input data. As will be described below, such training data can be obtained, for example, from simulations.Accordingly, the trained artificial intelligence also uses impact parameters as input data and injury parameters as output data. As will be described in more detail below, such injury parameters can, for example, indicate the degree of injury to a certain body part, such as the head, or even predict which body part will be injured.

[0012] The determination procedure then involves entering the impact parameters into this trained artificial intelligence, thereby receiving the corresponding injury parameters of the endangered road user as output data.

[0013] These data can then be output to an external unit, for example, for further processing. Output in this context could mean, for instance, transmitting the injury parameters to the aforementioned external unit using a suitable transmission standard. The external unit could, for example, be a corresponding computing unit located in an ambulance, a hospital, or a control center.

[0014] Thus, the aforementioned method can be used to predict the injury profile of an endangered road user, such as a pedestrian, in an accident with a vehicle, such as a car, and thereby accelerate and improve the treatment of the patient, i.e., the endangered road user.

[0015] According to another aspect, the determination process can also include the creation of an animation of the predicted collision based on the impact parameters by the central processing unit. As already described above, the aforementioned impact parameters can include information on the collision speed or relative speed of the vehicle to the endangered road user, a collision angle, or even the body orientation relative to the vehicle at the time of impact. Accordingly, this information can be used to generate a moving animation of the impact. This animation can then also be output to the external unit.

[0016] This allows the viewer of this animation to better assess the injury pattern of the endangered road user.

[0017] According to another aspect, impact parameters can be determined using a vehicle perception system. A perception system, also known as a driver assistance system, could be, for example, the system typically used in vehicles for driver assistance systems. As mentioned above, an impact parameter can include at least one of the following: a collision speed, such as the relative speed between the vehicle and the endangered road user at the time of the collision, a collision angle, or the orientation of the endangered road user relative to the vehicle at the time of impact.

[0018] Such data can be determined, for example, from camera images from cameras that are part of the vehicle as sensor devices.

[0019] According to another aspect, the determination process can also include training the artificial intelligence. Here, the sensor data, for example, contains a multitude of impact parameters as input data and corresponding injury parameters as output data. Within the framework of supervised learning, the artificial intelligence, such as a neural network, can then be trained using this training data. Such training data can be obtained through a variety of simulations, whereby the relevant parameters, such as collision speed, collision angle, and body orientation relative to the vehicle, are varied in each simulation, and the corresponding injury pattern and thus the associated injury parameters are determined through the simulation.

[0020] According to another aspect, the determination process can also include receiving feedback data from a user. This user could be, for example, a doctor who, after the injured road user has been admitted to a hospital or during an on-site examination, examines the injured road user's injury pattern and compares it with the predicted injury pattern. The user, such as the doctor, can then enter feedback data, for example via an input device, and this data can be used to train the artificial intelligence and improve the prediction.

[0021] Another aspect of this approach is that outputting injury parameters to an external unit can involve controlling that unit. This allows external devices, such as external processing units or display devices, to be controlled through this output. Alternatively, output can be achieved by first generating a unique identifier, such as a QR code, using the central processing unit. This identifier is then output, and in a subsequent step, the injury parameters are accessed by reading this unique identifier. This allows the relevant information to be output to a variety of external units and associated with other information, such as a medical record.

[0022] According to another aspect, outputting injury parameters to an external unit can also involve controlling a display device. This display device could be located, for example, in a control center, an ambulance, or a hospital.

[0023] According to another aspect, the injury parameters can include at least one aspect: the injured body part itself, that is, an indication of which body part is likely to be injured; the degree of injury to the person as a whole, that is, an overall assessment of the severity of the injury; and the degree of injury to one or more body parts. For example, it is possible to predict the severity of the injury to each body part.

[0024] According to a further aspect, the present invention relates to the central computing unit discussed above for determining a prediction of an injury to a vulnerable road user. This can, for example, be a cloud-based computing unit.

[0025] Facilities of this central computing unit can, for example, be software modules that are trained to perform the corresponding functions.

[0026] The central processing unit initially comprises a receiver for receiving impact parameters from a vehicle. This reception can occur, for example, via a suitable receiver for wired or wireless communication. Alternatively, it can also involve reading the relevant information from a memory. The central processing unit then includes a trained artificial intelligence, which has been trained to determine injury parameters of the endangered road user as output data, based on the impact parameters as input data. Reference is made to the discussion above regarding the trained artificial intelligence.

[0027] The central processing unit then includes an output device for transmitting the violation parameters to an external unit. This can again be a suitable device for wired or wireless transmission. However, output can also involve writing the corresponding information to a memory. This output device can also include a device for generating a unique identifier, such as a QR code, which is then output by the output device. Upon receiving this identifier and querying it via the receiving device, the violation parameters can then be output by the output device.

[0028] Another aspect concerns a system for determining a prediction of a road user at risk.

[0029] This determination process initially comprises the vehicle's sensor systems, which are configured to output sensor data regarding a collision between the vehicle and the endangered road user. These systems can include, for example, one or more cameras, acceleration sensors, etc. Reference is made to the explanations above in this regard. The determination system then includes a device for determining impact parameters based on the sensor data. As discussed above, this can, for example, be a vehicle perception system. The device for determining impact parameters is generally located within the vehicle.

[0030] The determination system then includes a transmission device for transmitting the impact parameters to the aforementioned central processing unit. This transmission can, for example, be a suitable wired or wireless connection.

[0031] Furthermore, this system includes the aforementioned central computing unit itself.

[0032] Brief description of the characters

[0033] Figure 1 shows the components of a determination system.

[0034] Figure 2 shows a flowchart of a determination procedure.

[0035] Detailed description of embodiments

[0036] Figure 1 shows the components of a tracking system. On the one hand, the components of the central processing unit 100 described above are shown, namely the receiving device 101, the trained artificial intelligence 102, and the output device 103. Furthermore, the components of the vehicle 104, which are part of the tracking system, are shown. These are the vehicle's sensor devices 105, the vehicle's tracking device 106, and the vehicle's transmission device 107. Figure 1 also shows a simulation device 108, which generates the training data discussed above and thus trains the artificial intelligence 102 of the central processing unit 100. Various possible external devices, such as external devices of an ambulance 109 or a hospital 110, are also shown.

[0037] The units described above and their functionality will be discussed further below in conjunction with the flowchart in Figure 2.

[0038] In step S1, the artificial intelligence 102 is first trained using training data. For this purpose, the simulation unit 108 generates various training data by simulating a collision between a vulnerable road user and a vehicle. The relevant impact parameters, such as the relative speed, the angle of impact, and the orientation of the vulnerable road user relative to the vehicle at the time of impact, are varied. The simulation outputs a multitude of injury patterns as injury parameters. Using this simulation-generated training data, which has impact parameters as input and injury parameters as output, the artificial intelligence 102 of the central processing unit 100 is then trained. This is illustrated by arrow 111.

[0039] Then, in the event of a collision between a vehicle 104 and a vulnerable road user 112, various sensor data are received from a sensor device 105 of the vehicle 104 in step S2.

[0040] In step S3, the vehicle 104's determination device 106 then determines impact parameters based on the sensor data. In this case, this is a perception system of the vehicle 104, which determines the aforementioned impact parameters based on, for example, camera data as sensor data.

[0041] Then, in step S4, these impact parameters are transmitted by the vehicle 104 to the receiving unit 101 of the central processing unit 100 via the transmission device 107. This is illustrated by arrow 113.

[0042] In step S5, the received impact parameters are then entered into the artificial intelligence 102 as input data. The output data thus consists of injury parameters for the endangered road user. In this case, these are degrees of injury, indicated on a scale, for a variety of body parts of the endangered road user, such as a parameter for head injuries, one for arm injuries, one for spinal injuries, etc.

[0043] Then, in step S6, the central processing unit 100 can additionally...

[0044] Animation of the collision is generated based on the impact parameters, which depicts the course of the collision and, for example, the impact of the corresponding body parts on the body of vehicle 104.

[0045] In step S7, the injury parameters and, if applicable, the animation can then be output to an external device, such as an external device in an ambulance 109 or in a hospital 110, using output device 103. This is indicated by arrows 114 and 115.

[0046] In the next step, S8, following an examination of the endangered road user, feedback on the actual injury pattern can be entered by a user, such as a doctor at hospital 110, and transmitted to the central computer 100. The transmission is indicated by arrow 116.

[0047] The feedback can then be received in step S8 and used in step S9 to optimize the trained artificial intelligence 102.

[0048] Reference symbol central computing unit

[0049] Reception facility trained artificial intelligence

[0050] Output device

[0051] vehicle

[0052] Vehicle sensor systems

[0053] Vehicle identification device

[0054] Vehicle transmission device

[0055] Simulation facility

[0056] ambulance

[0057] hospital

[0058] Training vulnerable road users

[0059] Transmission of impact parameters

[0060] Output of injury parameters

[0061] Output of injury parameters

[0062] Submitting feedback

[0063] Training artificial intelligence

[0064] Receiving sensor data

[0065] Determining impact parameters

[0066] Transmitting the impact parameters

[0067] Entering the impact parameters

[0068] Creating an animation

[0069] Output of injury parameters and animation

[0070] Receiving feedback data

[0071] Using feedback data for optimization

Claims

Patent claims 1. Determination procedure for determining a prediction of injury to a vulnerable road user (112), comprising the steps: - Receiving (S2) sensor data from sensor devices (105) of a vehicle (104) regarding a collision of the vehicle (104) with the endangered road user (112); - Determining (S3) impact parameters based on sensor data; - Transmitting (S4) the impact parameters by the vehicle (104) to a central computing unit (100), wherein the central computing unit (100) has a trained artificial intelligence (102) which has been trained to determine injury parameters of the endangered road user (112) as output data based on the impact parameters as input data; - Input (S5) of the impact parameters into the trained artificial intelligence (102) and receipt of injury parameters of the endangered road user as input data; - Output (S7) the injury parameters to an external unit.

2. Determination method according to claim 1, further comprising - Create (S6), by the central processing unit (100), an animation of the collision based on the impact parameters, and - Output (S7) the animation to the external unit.

3. Determination method according to one of the preceding claims, wherein - the determination (S3) of impact parameters is carried out using a vehicle perception system (104) and includes at least one of: - a collision speed, - a collision angle, and - Body orientation of the endangered road user (112) relative to the vehicle.

4. Determination method according to any of the preceding claims, further comprising: - Training (S1) of the artificial intelligence (102) using training data, wherein the training data comprise a variety of impact parameters as input data and associated injury parameters as output data, wherein the training data were obtained by means of simulation.

5. Determination method according to any of the preceding claims, further comprising: - Receiving (S8) feedback data from a user, and - Use (S9) of the feedback data to optimize the trained artificial intelligence (102).

6. Determination method according to one of the preceding claims, wherein the output (S7) of the violation parameters to an external unit comprises controlling the external unit.

7. Determination method according to one of the preceding claims, wherein the output (S7) of the violation parameters to an external unit comprises controlling a display device as an external unit.

8. Determination method according to any of the preceding claims, wherein the infringement parameters comprise at least one of: - an injured body part; - a degree of injury; - a degree of injury to a body part.

9. Central computing unit (100) for determining a prediction of an injury to an endangered road user (112), comprising: - a receiving device (101 ) for receiving impact parameters from a vehicle (104), - a trained artificial intelligence (102) which was trained to determine injury parameters of the endangered road user as output data, based on the impact parameters as input data, - an output device (103) for outputting the injury parameters to an external unit.

10. Determination system for determining a prediction of injury to a vulnerable road user (112), comprising: - Sensor devices (105) of a vehicle (104) which are configured to output sensor data regarding a collision of the vehicle (104) with the endangered road user (112), - a determination device (106) for determining impact parameters based on the sensor data, - a transmission device (107) for transmitting the impact parameters to the central computing unit (100) according to claim 9, and - the central computing unit (100) according to claim 9.

Images