Method for regulating the supply of outside air to the interior of a vehicle

Abstract

The invention relates to a method for regulating the supply of outside air to an interior of a vehicle (1), wherein the interior load of harmful substances is continuously determined during the journey of the vehicle (1) from signals acquired by a harmful substance sensor (2) arranged in the interior. The invention provides that the harmful substance load of the outside air on the journey section ahead of the vehicle (1) is predicted, wherein the supply of outside air is automatically regulated in accordance with the predicted harmful substance load of the outside air.

Description

TECHNICAL FIELD

[0001] The invention relates to a method for regulating the supply of outside air to a vehicle interior, wherein the interior harmful substance load is continuously determined during the vehicle journey in dependence on a measured signal of a harmful substance sensor arranged in the interior. BACKGROUND

[0002] A device and a method for controlling the ventilation of a motor vehicle interior in dependence on a signal of a harmful substance sensor which is essentially supplied with outside air are disclosed from DE 41 06 078 Al. Here, a switch between an air intake mode and an air circulation mode is carried out in dependence on the harmful substance concentration. By means of a computer, a first parameter which is associated with the harmful substance concentration in the interior is determined in consideration of the sensor signal, the respective operating mode (air intake or air circulation) and predetermined call-up experience values and is compared with a second parameter which is derived from the sensor signal and is associated with the harmful substance concentration of the outside air. Depending on the comparison result, the air intake mode or the air circulation mode is activated.

[0003] Furthermore, a method for controlling the air quality in a vehicle interior on the basis of a predicted air quality is known from US 2016 / 0 318 368 Al. Here, air quality data are received by means of vehicle sensors which belong to the vehicle and sensors of other vehicles or external environmental sensors or information sources which are arranged at a distance. Furthermore, a first air quality value for the vehicle environment and a second air quality value for the vehicle interior are determined in dependence on the air quality data, and a control signal is determined on the basis of these air quality values. The control signal is transmitted to the vehicle air conditioning system in order to automatically control the air quality in the vehicle interior. SUMMARY

[0004] It is the task of the invention to specify a method for regulating the supply of outside air to a vehicle interior.

[0005] This task is solved by a method having the features according to the invention.

[0006] Advantageous designs of the invention are the subject matter of further embodiments.

[0007] A method for regulating the supply of outside air to a vehicle interior provides that the interior harmful substance load is continuously determined during the vehicle journey in dependence on a measured signal of a harmful substance sensor arranged in the interior. According to the invention, the harmful substance load of the outside air on the driving section in front of the vehicle is predicted, wherein the outside air supply is automatically regulated in dependence on the predicted harmful substance load of the outside air.

[0008] Here, the harmful substance load of the vehicle exterior air is determined on the basis of optical / visual information, which is determined on the basis of the measured signals of at least one vehicle-side optical / visual detection unit. In particular, the optical information acquired is further processed and evaluated in order to determine the harmful substance load of the exterior air.

[0009] It is thus possible to extract semantic information from the acquired optical information, wherein, for this purpose, in particular machine learning methods, such as object recognition methods, are employed. For example, templates can be stored on the vehicle side, which are compared with objects detected in the optical information in order to recognize which objects are located in the vehicle surroundings.

[0010] By using this method, the driving comfort in the vehicle interior is improved and the health of the passengers in the interior is protected. In particular, the passengers' health is protected in that the supply of exterior air, i.e. the intake air of the vehicle interior ventilation device, is stopped or at least reduced when the exterior air to be expected in the future has a high harmful substance concentration. Thus, the supply of exterior air to the interior is reduced by means of the method before harmful substances enter the vehicle interior with the exterior air. In addition, in one possible refinement, identification features / fingerprints are extracted from the optical information in order to recognize objects emitting harmful substances in the future acquired signals of the at least one optical detection unit. Thus, it is possible to recognize by means of the identification features objects which were not defined and known and thus not taken into account during the development of the exterior air supply regulation model.

[0011] In another embodiment of the method, stationary objects and moving objects emitting harmful substances are recognized in the optical information of the vehicle surroundings as semantic information, wherein, for example, a truck in front of the vehicle and a trailer loaded with agricultural fertilizer (so-called manure) are recognized as moving objects emitting harmful substances and the supply of exterior air to the interior is stopped.

[0012] Furthermore, the method provides in another possible embodiment that the instantaneous position of the vehicle is continuously determined, so that, for example, on the basis of the instantaneous position it can be recognized that the vehicle is located in an industrial area, for example directly adjacent to an industrial plant. In this case, too, the supply of exterior air to the vehicle interior is at least reduced.

[0013] In addition, the method provides in another possible embodiment that the instantaneous position is combined with time information, for example clock time and day of the week, so that, for example, on the basis of the instantaneous position and the current day of the week it can be recognized that the vehicle is located in an industrial area, for example directly adjacent to an industrial plant, in particular on a weekday. In this case, too, the supply of exterior air to the vehicle interior is at least reduced.

[0014] Based on the extracted semantic information, the extracted identification features and / or in dependence on the determined vehicle instantaneous position, a model of the time delay in the vehicle is trained, which predicts the interior harmful substance load as a target parameter. This model thus takes into account the time delay and the time integration of the harmful substance load of the outside air of the vehicle.

[0015] Furthermore, the method provides that a time delay between a high interior harmful substance load and the cause extracted by the model is determined. The model thus trained is used in the vehicle to predict the expected future interior harmful substance load of the vehicle by means of input information of the identification features, in particular the so-called perception fingerprint, in particular by means of semantic information and / or the vehicle instantaneous position. It is thereby possible to regulate the outside air supply in such a way that as little harmful substance as possible enters the interior to thus protect the health of the passengers.

[0016] In another possible design, the trained model, in particular of the vehicle, is supplied to a central computing unit, so that the model can be pooled and aggregated with at least one further model of another vehicle.

[0017] The model and / or the aggregated model can then be transmitted to other vehicles of the fleet by means of the central computing unit and provide for a regulation of the outside air supply of each further vehicle of the fleet. BRIEF DESCRIPTION OF DRAWINGS

[0018] Embodiments of the application will be explained in detail below with reference to the drawings, in which:

[0019] Figure 1 A vehicle with a harmful substance sensor and an optical detection unit as well as various different stationary and mobile objects in the surroundings of the vehicle are schematically shown. DETAILED DESCRIPTION

[0020] A vehicle 1 is shown in the only figure, which has a harmful substance sensor 2 and an optical detection unit 3 in the form of a camera, wherein a truck 4 and a bicycle 5 as mobile objects O1, an industrial plant 6 as a stationary object O2 and a central computing unit 7 are also shown.

[0021] The harmful substance sensor 2 is arranged in the interior of the vehicle 1 and continuously acquires a signal during the driving of the vehicle 1, in dependence on which the interior harmful substance load is determined. Instead of or in addition to the harmful substance sensor 2, a sensor can also be arranged in the interior of the vehicle 1, in dependence on the signal acquired thereby, a human-perceptible odor is recognized.

[0022] The interior harmful substance load determined in dependence on the signal measured by the harmful substance sensor 2 is a function of the time delay and the time integration of the harmful substance load of the environment outside the vehicle 1, i.e. of the outside air.

[0023] As soon as the harmful substance load of the interior is determined from the signals of the harmful substance sensor 2, the ventilation of the interior of the vehicle 1 is controlled as is known from the prior art.

[0024] In order to achieve the regulation of the outside air supply to the interior of the vehicle 1 in such a way that the health of the passengers is protected in relation to the harmful substance load by keeping the harmful substance load of the interior as low as possible, the following method is provided.

[0025] Here, the harmful substance load of the outside air in the driving section ahead of the vehicle 1 is predicted and the outside air supply is automatically regulated in dependence on the predicted harmful substance load.

[0026] In particular, the method provides that the outside air supply to the interior of the vehicle 1 is stopped when the outside air is expected to have a high harmful substance concentration in the future. Thus, by means of the method, the outside air supply is reduced before the harmful substances enter the interior of the vehicle 1 with the outside air.

[0027] As described above, the harmful substance sensor 2 on the vehicle side continuously acquires signals during driving of the vehicle 1, from which the harmful substance load of the interior is determined.

[0028] The vehicle 1 has an optical detection unit 3 in the form of a camera, the detection range of which is directed in front of the vehicle 1 and by means of which signals are continuously acquired during driving of the vehicle 1, from which signals the surroundings of the vehicle 1 and objects O1, O2 located therein are detected.

[0029] In addition, the vehicle 1 comprises a satellite-supported positioning unit and a digital map, which are not shown in detail, so that the instantaneous position of the vehicle 1 can be determined.

[0030] The optical information evaluated is determined from the signals of the optical detection unit 3.

[0031] Here, semantic information can be extracted from the optical information by means of a machine learning method, in which, in particular, object recognition in relation to known objects O1, O2 is used. That is, it can be recognized which object O1, O2 is located in the surroundings of the vehicle 1. The objects O1, O2 detected in the surroundings of the vehicle 1 can be distinguished in particular as moving objects O1 and stationary objects O2.

[0032] In addition, it can be determined from the semantic information which object O1, O2 emits harmful substances. If, for example, a bicycle 5 is detected as a moving object O1 in the surroundings of the vehicle 1, it is recognized from the determined features of the bicycle 5 that it is a bicycle 5 that does not emit harmful substances.

[0033] Furthermore, the identification features, in particular the perception identification features, can be extracted from the optical information, for example by means of a folding / convolutional neural network, without having to assign predetermined objects Oi, O2. Thereby, during the development of the model created by means of this method, which will be explained below, non-predetermined and known objects Oi, O2 can be identified by means of the identification features or by means of a plurality of extracted identification features.

[0034] The model of the time lag within the vehicle 1 is trained by means of a machine learning method depending on the semantic information, the perception identification features and / or depending on the position information determined based on the instantaneous position of the vehicle 1. The model is trained here such that as a target parameter the determined interior harmful substance load is predicted. The model can be, for example, a recurrent model or a reinforcement learning model.

[0035] In the model, a time delay and a time integration of the harmful substance load of the outside air is taken into account, for example by using temporal information.

[0036] Such a model allows to determine in an extrapolated manner a time lag between a higher interior harmful substance load, in particular above a predetermined threshold, and a cause of the higher harmful substance load of the outside air determined by means of the model. Such a cause of a high harmful substance load is, for example, a truck 4 and / or a preceding towing vehicle with a trailer loaded with agricultural fertilizer (so-called manure).

[0037] The model trained in this way is used in the vehicle 1 to predict an expected interior harmful substance load of the vehicle 1 by means of the input information of the perception identification features, in particular the semantic information and / or the position information. If a comparatively high interior harmful substance load is predicted in the future, the interior ventilation is adjusted such that the outside air supply to the interior of the vehicle 1 is reduced or switched off in correspondence with the predicted interior harmful substance load.

[0038] In one embodiment, the model can be pre-trained during the development of the vehicle 1, wherein the model can also be trained and / or continuously trained in a user-specific and / or region-specific manner in other vehicles of the vehicle fleet.

[0039] It is also conceivable that a user-specific and / or region-specific general model is summarized and aggregated in a central computing unit 7 of the vehicle manufacturer, for example, by means of distributed learning (so-called federated learning) depending on the experience of other vehicles of the vehicle fleet.

[0040] The model also allows to identify the cause of a comparatively high harmful substance concentration, for example a truck 4 and / or an industrial plant 6, which can be region-specific. Such information can be used, for example, to identify the source of a higher harmful substance concentration in the environment of the vehicle 1 in a country-specific and / or region-specific manner.

[0041] Thus, by means of the central computing unit 7, the respective model and / or the collective model can be provided to the vehicle 1 of the fleet and to other vehicles.

[0042] The method thus provides that a model is trained on the basis of the optical information of the optical detection unit 3, the geographical position, i.e. the instantaneous position of the vehicle 1, and in dependence on the harmful substance sensor 2, which model predicts the harmful substance load in the interior, i.e. the harmful substance concentration in the interior, on the basis of the optical information, in particular.

[0043] For example, the moving object Ol and the stationary object 02 are associated with a higher harmful substance load of the outside air and also, with a delay, with a higher harmful substance load in the interior, wherein the model learns said associations.

Claims

1. A method for regulating the supply of external air to the interior of a vehicle (1), wherein: The interior pollutant load is continuously determined during driving of the vehicle (1) based on signals detected by a pollutant sensor (2) arranged in the interior, wherein the pollutant load of the outside air on the driving section ahead of the vehicle (1) is predicted, and the outside air supply is automatically adjusted based on the predicted interior pollutant load. Its characteristics are: The interior harmful substance load determined based on the signal measured by the harmful substance sensor (2) is a function of the time delay and time integral of the harmful substance load of the outside air of the vehicle (1), The harmful substance load of the air outside the vehicle (1) is determined based on optical information, which is determined based on signals collected by at least one vehicle-side optical detection unit (3), and semantic information is extracted from the collected optical information. extracting identification features based on the optical information to identify objects (O1, O2) emitting harmful substances in future collected signals of the at least one optical detection unit (3), continuously determining the instantaneous position of the vehicle (1), Based on the extracted semantic information and the extracted identification features and according to the determined instantaneous position of the vehicle (1), a model of the delay in the vehicle (1) is trained in the following manner, i.e., the interior harmful substance load is predicted as a target parameter, and A time delay between a high interior pollutant load and a cause determined by the model, the cause being the pollutant-emitting object ( O1 , O2 ), is determined.

2. The method according to claim 1, wherein: Active objects (O1) and fixed objects (O2) that emit harmful substances in optical information of the vehicle (1) environment are acquired as semantic information.

3. The method according to claim 1 or 2, wherein: The trained model is fed to the central computing unit (7).

4. The method according to claim 1 or 2, wherein: The trained model is transferred to another vehicle in the fleet.

Images