System for determining vehicle occupancy

Abstract

One subject of the invention is a system for determining the occupancy of a motor vehicle by means of a presence sensor, the system being configured to carry out the following steps:-generating a point cloud (50) on the basis of data originating from the presence sensor, in particular a radar, any point associated with an external target located outside the predetermined detection area (60) is detected in the point cloud generated in the previous step.

Description

Technical Field

[0001] This invention relates to a system for determining the state of existence inside a motor vehicle. Background Technology

[0002] It is known to detect the presence of living persons (especially people) inside motor vehicles, and to issue an alarm if a person is left behind. This detection and alarm can improve passenger safety, particularly reducing the risk of children being left in motor vehicles. Such detection can, for example, utilize radar located inside the motor vehicle.

[0003] In particular, false alarms need to be prevented. Such false alarms can occur when a person outside the vehicle passes near the radar and the radar detects this person outside the vehicle. Therefore, if a person is outside the vehicle but is perceived as being inside, the detection can cause confusion. This is because the radar inside the vehicle can capture reflections of waves from a person outside the vehicle.

[0004] False alarms may also occur when strong winds cause vibrations inside the vehicle.

[0005] Of course, such false alarms are undesirable. Summary of the Invention

[0006] This invention is particularly designed to prevent such false alarms.

[0007] Therefore, the present invention relates to a system for determining the presence state inside a motor vehicle by means of a presence sensor, the system being configured to perform the following steps:

[0008] - Generate point clouds based on data from existing sensors (especially radar).

[0009] - Detect any points in the point cloud generated in the previous step that are associated with external targets located outside the predetermined detection area.

[0010] - In a first exemplary embodiment of the present invention: points associated with external targets and located within a predetermined detection area are removed from the point cloud to provide a filtered point cloud; the filtered point cloud is then processed in the step of detecting the presence state in the vehicle to detect the potential presence of targets inside the motor vehicle, or

[0011] - In a second exemplary embodiment of the invention: when a point associated with an external target and located within a predetermined detection area is confirmed to be detected in the point cloud, data indicating "not detected inside the vehicle" is forcibly generated, especially even if there are people in the vehicle.

[0012] By means of this invention, false alarms can be avoided when sensors (especially radar) detect points, for example, originating from people outside the vehicle. This invention allows points related to people outside the vehicle to be disregarded, which can be achieved by removing these points from the detected point cloud or by forcibly generating "not detected" information, regardless of whether anyone is in the vehicle.

[0013] In a first exemplary embodiment of the present invention, it can be said that the present invention enables the shielding of parasite points created by the presence of persons outside the vehicle.

[0014] In a second exemplary embodiment of the present invention, instead of masking points generated by people outside the vehicle, information indicating non-existence is forcibly provided to avoid issuing false alarms when no one is inside the vehicle.

[0015] Generally speaking, the present invention enables the prevention of false alarms caused by parasitic detection of people or objects moving in the vicinity of a vehicle.

[0016] According to one aspect of the invention, the detection area of ​​the sensor corresponds to the interior of the passenger compartment of the vehicle.

[0017] According to one aspect of the invention, the detection area is defined by a volume, such as a parallelepiped, particularly a cuboid, which substantially corresponds to the outline of the passenger compartment of a vehicle.

[0018] According to one aspect of the invention, presence sensors (in particular radar) are thus configured to detect presence throughout the passenger compartment of the vehicle, particularly the potential presence of occupants in the front and rear seats.

[0019] According to one aspect of the invention, presence detection is performed based on a point identified by a presence sensor and located within the detection area.

[0020] It should be noted that people passing by a vehicle, especially those outside the vehicle, can easily create parasitic points that can be detected by radar within the detection area. This is because radar waves reflect off the person and into the vehicle. The presence of these parasitic points can lead to false alarms.

[0021] According to one aspect of the invention, a sensor (in particular a radar) is configured to detect points in a detection area whose behavior represents the presence of a person (in particular breathing).

[0022] Parasitic points associated with the presence and movement of people near a car may be used by a system to determine the state of presence, but are not intended to be interpreted as points within the detection area.

[0023] According to one aspect of the invention, a system for determining the state of existence is configured to generate clusters of points of a point cloud and track them, particularly for identifying parasitic points.

[0024] In a first exemplary embodiment of the invention (i.e., removing points from the point cloud that are associated with an external target and located within a predetermined detection area to provide a filtered point cloud), parasitic points appearing within the detection area in the point cloud are removed during point cloud processing. If necessary, parasitic points outside the detection area may not be processed, i.e., not removed, because they do not affect the determination of the presence state inside the detection area.

[0025] Therefore, if no one is inside the vehicle, the image obtained through this point cloud processing will not contain points of interest, leading to the conclusion that no presence is detected inside the vehicle. Therefore, no alarm needs to be triggered.

[0026] This invention is advantageous because, by means of this treatment of parasites, there is no need for additional sensors to detect the presence of external objects / people (external targets), such as a person passing by the vehicle near its location. Therefore, this invention allows the use of radar, typically present in the passenger compartment, to remove parasites.

[0027] Generally speaking, tracking specifically refers to the process of continuously tracking a moving object or target in order to determine its position and orientation in three-dimensional space. This allows for the real-time tracking of an object's motion and positional changes.

[0028] For example, when a child is inside a car and an external person (external target) passes by the car, the sensors generate a point cloud, and the system is able to identify two sets of points: one set associated with the child (because these points have similar movements) and another set associated with the external person (because these points have similar movements). In this situation, the system triggers an alarm.

[0029] According to one aspect of the invention, the system is configured to use artificial intelligence to determine whether a set of points (particularly point-based movement) are related to a person (target) inside or outside the vehicle.

[0030] For example, the artificial intelligence can determine that the movement of a point with a specific vector is associated with a person outside the vehicle.

[0031] For example, several points very close to the edge of the detection area within a set of points can be considered to belong to external personnel. Specifically, the system is configured to identify parasitic points created by the reflection of waves emitted by the sensor onto external personnel, which exhibit specific movements of approaching and moving away, characteristic of people walking next to vehicles.

[0032] It can also be inferred that if some points remain in the image captured by the sensor for a relatively short period of time, such as less than 20 seconds or less than 10 seconds, they are parasitic points. This means that the target appears and disappears relatively quickly, and makes it possible to ensure that the target is outside the vehicle.

[0033] According to one aspect of the invention, the system uses artificial intelligence with learning capabilities to improve the identification of parasite sites.

[0034] According to one aspect of the invention, the system, particularly through artificial intelligence, is configured to compare detected point clouds, especially their dynamic features, with a stored catalog to determine whether the point cloud or a cluster of points in the point cloud belongs to a target in the vehicle.

[0035] An alarm is triggered when a presence is detected in a vehicle, regardless of whether a parasite is actually detected.

[0036] In a second exemplary embodiment of the present invention, if a point associated with an external target and located within a predetermined detection area is confirmed to be detected in the point cloud, even if a person is detected in the vehicle at the same time as the parasitic point associated with the external target is detected, data indicating "no presence detected inside the vehicle" is still forcibly generated.

[0037] According to one aspect of the invention, the post-processing steps of new presence detection are repeated periodically, so that potential alarms are forcibly prevented only during the period of external target interference detection.

[0038] According to one aspect of the invention, the system is designed to generate an alarm signal based on a detected presence state, and the signal is in particular an audible sound signal in the vehicle's environment, or a message sent to a smartphone or help center.

[0039] According to one aspect of the invention, the system includes a computer configured to process signals from presence sensors.

[0040] According to one aspect of the invention, the computer is configured to utilize artificial intelligence to process point clouds obtained through presence sensors.

[0041] Another subject of the present invention is a method for determining the presence state inside a motor vehicle using presence sensors, the method comprising the following steps:

[0042] - Generate point clouds based on data from existing sensors (especially radar).

[0043] - Detect any points in the point cloud generated in the previous step that are associated with external targets located outside the predetermined detection area.

[0044] - In a first exemplary embodiment of the present invention: points associated with external targets and located within a predetermined detection area are removed from the point cloud to provide a filtered point cloud; the filtered point cloud is then processed in the step of detecting the presence state in the vehicle to detect the potential presence of targets inside the motor vehicle, or

[0045] - In a second exemplary embodiment of the present invention: if a point associated with an external target and located within a predetermined detection area is confirmed to be detected in the point cloud, data indicating "not detected inside the vehicle" is forcibly generated (even if there is someone in the vehicle). Attached Figure Description

[0046] Other features, details, and advantages of the invention will become apparent from the following detailed description of several non-limiting exemplary embodiments, taken in conjunction with the accompanying drawings. In the drawings:

[0047] Figure 1 This is a schematic diagram of a vehicle including a system for determining a state of presence according to an exemplary embodiment of the present invention, wherein an external person approaches and passes by the vehicle;

[0048] Figure 2 It is shown by Figure 1 A flowchart of the steps to determine the existence of the system;

[0049] Figure 3 It is by Figure 1 A schematic diagram of the detection area used by the system;

[0050] Figure 4 It occurs in three consecutive time periods while an outsider is approaching. Figure 3 The depiction of the detection area in three images;

[0051] Figure 5 This is a block diagram illustrating the steps for determining the presence according to another exemplary embodiment of the present invention.

[0052] The features, variations, and various embodiments of the present invention can be combined with each other in various combinations, provided that they are not mutually incompatible or mutually exclusive. In particular, variations of the invention may be contemplated that include only selections of the features described below, isolated from the other features described, if the selection of these features is sufficient to provide a technical advantage and / or distinguish the invention from the prior art. Detailed Implementation

[0053] Figure 1 A system 1 according to the invention is shown for determining the presence state E of occupants inside a vehicle V (in this case, a motor vehicle). System 1 includes a presence sensor 10 and a computer 20 connected to the presence sensor 10.

[0054] The presence sensor 10 is configured to determine the presence of occupants 30 inside the vehicle V.

[0055] The presence sensor 10 is placed in the central area of ​​vehicle V, in this case, such as Figure 1As shown, a sensor 10 is placed at the center of the roof 2 of the vehicle V, and thus advantageously covers all seats of the vehicle V.

[0056] The presence sensor 10 can be any instrument that can be used to determine the presence of an occupant 30 inside the vehicle V. The presence sensor 10 can consist of a single sensing element or a network of sensing elements distributed throughout the vehicle V.

[0057] The sensor 10 may be, for example, a camera or camera network, or even an ultrasonic sensor or network of ultrasonic sensors.

[0058] In this case, sensor 10 is a device for transmitting and receiving radar waves.

[0059] In this case, the presence sensor 10 includes at least one antenna (not shown) designed to emit electromagnetic waves 11, and at least one sensor (not shown) designed to receive reflected electromagnetic waves 12 after the electromagnetic waves 11 are reflected, particularly after reflection from a person 30 inside the vehicle V or an external person 40 near the vehicle V.

[0060] In this context, the presence sensor 10 is more specifically a radar (radio detection and ranging) that uses millimeter electromagnetic waves. The radar, for example, is a continuous-wave Doppler radar.

[0061] Computer 20 includes at least one memory and at least one processor. Computer 20 may be, for example, an electronic control unit (ECU) of vehicle V. Computer 20 may also be a computer dedicated to system 1. Instructions for determining the state E are stored in memory and implemented by the processor. When implemented, these instructions enable the methods described below to be implemented.

[0062] Specifically, the computer 20 is programmed to control the presence sensor 10, particularly to monitor the transmission of electromagnetic waves 11.

[0063] System 1 is configured to perform the following steps:

[0064] - Generate point cloud 50 based on data from presence sensor 10. Figure 2 Step 101 in the middle).

[0065] - Detect any points in the point cloud 50 generated in the previous step that are associated with external targets located outside the predetermined detection area 60. Figure 2 Step 102 in the middle).

[0066] In a first exemplary embodiment of the present invention, the determining system 1 is configured to remove points from the point cloud that are associated with an external target and located within a predetermined detection area, in order to provide a filtered point cloud 51. Figure 2In step 103), the filtered point cloud 51 is then processed in the step of detecting the presence state E in vehicle V in order to detect the potential presence of targets inside the motor vehicle. Figure 2 Step 104 in the middle.

[0067] Steps 101, 102, and 103 can be considered preprocessing, and step 104 can be considered postprocessing.

[0068] If person 30 is actually present in vehicle V, then system 1 only sees points related to person 30 actually present in vehicle V for the detection area, because parasitic points related to external person 40 have been removed from the points used in step 104 of the detection presence state E.

[0069] By means of this invention, false alarms can be avoided when sensor 10 (in this case, radar) captures points originating from a person 40 outside the vehicle V. This invention allows points associated with a person outside the vehicle to be disregarded, i.e., by removing these points from the detected point cloud. In this first exemplary embodiment of the invention, it makes it possible to mask parasitic points arising from the presence of a person outside the vehicle.

[0070] Generally speaking, the present invention enables the prevention of false alarms caused by parasitic detection of a person 40 or object moving in the vicinity of the vehicle V.

[0071] As in Figure 3 As can be seen, detection area 60 corresponds to the interior of the passenger compartment of vehicle V. Therefore, system 1 uses detection area 60 and the presence or absence of points within it as the basis for detecting the presence state E. It should be understood that it is important that there are no parasitic points interfering with step 104 within detection area 60. Points in the point cloud can be represented in the x, z reference coordinate system.

[0072] The detection area 60 is defined by, for example, a cuboid volume that substantially corresponds to the outline of the passenger compartment of vehicle V.

[0073] Therefore, the presence sensor 10 is configured to detect the presence throughout the passenger compartment of the vehicle V, particularly the potential presence of people in the front and rear seats.

[0074] The detection of presence E is performed based on the point identified by presence sensor 10 and located within the detection area 60.

[0075] As in Figure 4As can be seen in the three images of the detection area 60 at three consecutive times when an external person is approaching, due to the reflection of radar waves from the person 40 toward the interior of the vehicle V, a person 40 outside the vehicle and passing close to the vehicle V is prone to generate parasitic points 61 in the detection area 60 that are captured by the radar 10. It is the presence of these parasitic points 61 that can cause false alarms.

[0076] exist Figure 4 In the left image, point cloud 50 is generated by person 40 outside vehicle V, who is initially relatively far from detection area 60, so that reflections do not generate parasitic points in detection area 60.

[0077] Next, person 40 approaches vehicle V, and the reflection of waves from person 40 produces a parasitic point 61 in detection area 60, which is captured by radar 10 (see [link]). Figure 4 (Image at the center).

[0078] After parasitic point 61 has been removed by system 1, what remains is the filtered point cloud 51 that does not affect the detection of E (in Figure 4 (In the image on the right).

[0079] Sensor 10 and system 1 are configured to detect points in detection area 60 where their behavior represents the presence of a person (particularly breathing).

[0080] In a known manner, System 1 is configured to generate point clusters of certain points in a point cloud and track them.

[0081] In the point cloud processing, parasitic points 61 appearing within the detection region 60 are removed. If necessary, parasitic points outside the detection region may not be processed, that is, they may not be removed, because they do not affect the determination of the existence state inside the detection region.

[0082] Therefore, if no one is inside the vehicle, the image obtained through this point cloud processing will not contain points of interest, leading to the conclusion that no presence is detected inside the vehicle. Therefore, no alarm needs to be triggered.

[0083] Clustering is performed using a Kalman filter. Of course, other types of filters or devices can be used.

[0084] Generally speaking, clustering points by their velocity vectors allows us to conclude that a group of points belong to the same object / person (target).

[0085] For example, when a child is inside a car and an external person (external target) passes by the car, the sensors generate a point cloud, and System 1 enables the identification of two sets of points: one set associated with the child (because these points have similar movements) and another set associated with the external person (because these points have similar movements). In this situation, the system triggers an alarm.

[0086] System 1 is configured to use artificial intelligence to determine whether a set of points (especially point-based movement) are related to a person (target) inside or outside the vehicle.

[0087] For example, the artificial intelligence can determine that the movement of a point with a specific vector is associated with a person outside the vehicle.

[0088] For example, several points very close to the edge of the detection area in a set of points can be considered to belong to an external person. Specifically, the system is configured to identify parasitic points generated by the reflection of waves emitted by sensor 10 on an external person 40, which exhibit specific movements of approaching and moving away, characteristic of a person 40 walking next to vehicle V.

[0089] It can also be inferred that if some points remain in the image captured by the sensor for a relatively short period of time, such as less than 20 seconds or less than 10 seconds, they are parasitic points. This means that the target appears and disappears relatively quickly, and makes it possible to ensure that the target is outside the vehicle.

[0090] System 1 uses artificial intelligence with learning capabilities to improve the identification of parasites.

[0091] System 1 can be specifically configured to compare the detected point cloud (especially its dynamic features) with a stored catalog to determine whether the point cloud or a cluster of points in the point cloud belongs to a target in the vehicle.

[0092] An alarm is issued when E is detected in vehicle V, regardless of whether a parasite point is detected.

[0093] exist Figure 5 In the second exemplary embodiment of the present invention shown, if a point associated with an external target and located within a predetermined detection area 60 is confirmed to be detected in the point cloud pre-generated in step 201 (in step 202), then system 1 is configured to forcibly generate data indicating "no presence detected inside the vehicle" (even if there is someone in the vehicle), bypassing the post-processing step of presence detection E (step 203).

[0094] By forcibly bypassing the post-processing step of presence detection E, even if the presence of a person in the vehicle is detected at the same time as the parasite point associated with the external person 40 is detected, the system 1 will not generate an alarm (step 204).

[0095] If no parasitic point is detected, the post-processing step 203 of presence detection E is performed normally, and if presence is detected in vehicle V, an alarm can be generated.

[0096] The post-processing steps for new presence detection are repeated periodically so that the above-mentioned mandatory measures prevent potential alarms only during the period when external target 40 interferes with detection.

[0097] Generally, the system is designed to generate an alarm signal based on the detected presence, and this signal is either an audible sound signal in the vehicle's environment or a message sent to a smartphone or help center.

Claims

1. A system (1) for determining the presence state inside a motor vehicle (V) by means of a presence sensor (10), said system being configured to perform the following steps: - Generate point clouds (50) based on data from the existing sensors, particularly radar. - Detect any points in the point cloud generated in the previous step that are associated with external targets located outside the predetermined detection area (60). - In a first exemplary embodiment of the present invention: points associated with an external target (40) and located within a predetermined detection area (60) are removed from the point cloud to provide a filtered point cloud (51), and then the filtered point cloud is processed in the step of detecting the presence state in the vehicle to detect the potential presence of the target (40) inside the motor vehicle, or - In a second exemplary embodiment of the present invention: if a point associated with an external target (40) and located within a predetermined detection area (60) is confirmed to be detected in the point cloud, data indicating "not detected inside the vehicle" is forcibly generated.

2. The system (1) according to the preceding claim, wherein, The detection area (60) of the sensor (10) corresponds to the interior of the passenger compartment of the vehicle (V), and is defined in particular by a volume that substantially corresponds to the outline of the passenger compartment of the vehicle, such as a parallelepiped, particularly a cuboid.

3. The system (1) according to any one of the preceding claims, wherein, The sensor (10), in particular a radar, is configured to detect points in the detection area (60) where the behavior can represent the presence of a person (in particular breathing).

4. The system (1) according to any one of the preceding claims, wherein, The system (1) used to determine the state of existence is configured to generate clusters of points of a point cloud and track them, in particular to identify parasitic points.

5. The system (1) according to any one of the preceding claims, wherein, In the processing of the point cloud, parasitic points that appear in the detection area (60) and are related to the external target (40) in the point cloud (50) are removed.

6. The system (1) according to any one of the preceding claims, wherein, The system is configured to use artificial intelligence to determine whether a set of points, particularly based on the movement of the points, is related to a person (target) inside the vehicle or a person (target) outside the vehicle. For example, the artificial intelligence determines that the movement of a point with a specific vector is associated with a person outside the vehicle.

7. The system (1) according to any one of the preceding claims, wherein, The system is configured to identify parasitic points created by the reflection of waves emitted by the sensor onto an external person (40), which exhibit specific movements of approaching and moving away, characteristic of a person walking next to the vehicle.

8. The system (1) according to any one of the preceding claims, wherein, The system, particularly via artificial intelligence, is configured to detect point clouds (50), especially their dynamic features. The point cloud or a cluster of points in the point cloud is compared with a stored directory to determine whether the point cloud or a cluster of points in the point cloud belongs to a target in the vehicle.

9. The system (1) according to any one of claims 1 to 4, wherein, If a point associated with an external target and located within the predetermined detection area (60) is confirmed to be detected in the point cloud (50), even if a person is detected in the vehicle (V) at the same time as the parasitic point associated with the external target is detected, data indicating "no presence detected inside the vehicle" is still forcibly generated.

10. The system (1) according to the preceding claim, wherein, The system (1) is configured to periodically repeat the post-processing steps of new presence detection, such that the above-mentioned enforcement prevents potential alarms only during the time period of external target interference detection.

11. The system (1) according to any one of the preceding claims, wherein, The system is designed to generate an alarm signal based on the detected presence, and the signal is either an audible sound signal in the environment of the vehicle or a message sent to a smartphone or help center.

12. A method for determining the presence state inside a motor vehicle (V) by means of a presence sensor (10), the method comprising the steps of: - Generate point clouds (50) based on data from the existing sensors, particularly radar. - Detect any points in the point cloud generated in the previous step that are associated with external targets located outside the predetermined detection area (60). - In a first exemplary embodiment of the present invention: points associated with external targets and located within a predetermined detection area (60) are removed from the point cloud to provide a filtered point cloud, which is then processed in the step of detecting the presence state in the vehicle to detect the potential presence of a target within the motor vehicle, or - In a second exemplary embodiment of the invention: if a point associated with an external target and located within a predetermined detection area (60) is confirmed to be detected in the point cloud, data indicating "not detected inside the vehicle" is forcibly generated, especially even if there are people in the vehicle.

Images