DE102023205608A1 - Computer-implemented method for machine learning of a seat setting that is ergonomically adapted to a user, computer-implemented method, computer program and control device for setting at least one property of at least one intelligent material of a seat for a seat setting that is ergonomically adapted to a user and seat - Google Patents

Abstract

Computer-implemented method for machine learning an ergonomically adapted setting of a seat (1) for a user, the seat comprising at least one intelligent material (2), wherein at least one property of the intelligent material (2) can be adjusted by means of at least one control device (3), and at least one seat surface and a backrest, the method comprising the steps of: obtaining occupancy-relevant data of the user (T1); providing metadata relating to an ergonomic sitting posture of the user (T2); processing the occupancy-relevant data and the metadata with an algorithm from the field of machine learning, wherein the algorithm learns the ergonomically adapted setting of the seat (1) and the signals required for this for the control device (3) after several process steps by linking the occupancy-relevant data and the metadata (T3); outputting the signals for the control device (3) (T4).

Description

The invention relates to a computer-implemented method for machine learning of a seat setting that is ergonomically adapted to a user. The invention also relates to a computer-implemented method, a computer program and a control device for setting at least one property of at least one intelligent material of a seat for a seat setting that is ergonomically adapted to a user. The invention also relates to a seat.

The terms introduced below retain their respective meaning for the entire subject matter of the invention.

Sitting in vehicles for long periods of time is often uncomfortable. In vehicles such as airplanes, individual breaks to leave the seat are very limited or not possible at all. What makes things even more difficult is that the number of vehicle users fluctuates considerably, for example in company cars or car sharing. From the user's perspective, their own body is subject to significant changes due to the predefined seats installed. Different body shapes, people's weights and sizes, physical illnesses or impairments, back shapes and other human characteristics vary greatly and therefore require, for example, individually adjusted seat hardness, lumbar support, neck support and the like.

The EN 10 2019 004 095 A1 discloses a vehicle seat with an adjustment device and a control unit for controlling the adjustment device with an electrical voltage, wherein the adjustment device comprises a first wire made of nitinol and a second wire made of nitinol.

The EP3 789 250 B1 discloses an identity and bio-management system in vehicles comprising a seat adjustment module and a data algorithm learning tool that develops new data algorithms based on a machine learning process using artificial intelligence.

The object of the invention was to find out how a seat setting can be automatically adjusted to a user in order to achieve an ergonomic posture of the user on the seat.

The subject matter of the independent and the subordinate claims each solves this problem. Further advantageous embodiments of the invention emerge from the definitions, the subclaims, the figures and the description of preferred embodiments.

According to one aspect, the invention provides a computer-implemented method for machine learning a seat setting that is ergonomically adapted to a user. This method corresponds to a training method of the algorithm from the field of machine learning. Machine learning relates to solving tasks from learned data patterns without a special solution algorithm having to be programmed procedurally or object-oriented for this purpose. The seat can be a vehicle seat, for example a vehicle seat of a passenger vehicle, for example a car, bus, shuttle, aircraft, ship, train, or a commercial vehicle, for example a truck, or a seat for a non-mobile application in which the seat is sat on for a longer period of time, for example a cinema seat. Ergonomic refers to an adjustment of the seat to increase efficiency, for example good all-round visibility when driving or in vehicles, and/or avoiding posture and movement damage, for example with regard to back muscles, spine, intervertebral discs, pelvis and/or blood circulation in the body. In order not to damage the body in the long term, correct posture when driving is important, for example. The user can, for example, be a vehicle driver or passenger.

The seat comprises at least one intelligent material, wherein at least one property of the intelligent material can be adjusted by means of at least one control device. The seat also comprises at least one seat and a backrest. Intelligent materials include materials whose properties can be influenced in a way by active control, for example via an electrical voltage. An intelligent material is, for example, a shape memory material. A shape memory material is a material that, after deformation, still contains the information of the original shape in its internal structure. Shape memory materials include shape memory alloys, also called memory metals, for example nickel-titanium or nickel-titanium-copper, shape memory polymers and shape memory ceramics. Other examples of intelligent materials are piezoelectric materials, optoelectronic materials, electroactive polymers, magnetostrictive materials, intelligent inorganic polymers, pH-sensitive polymers, temperature-responsive polymers, ferrofluids, photomechanical materials, dielectric elastomers, thermoelectric materials and chemoresponsive materials. According to one aspect, intelligent materials are integrated into a fabric cover of the seat. According to another aspect, intelligent materials are integrated into the seat and/or the backrest, According to a further aspect, it is integrated into other seat components, for example head/neck support, lumbar support, side bolsters. According to one aspect, an adjustable property of the intelligent material is its shape, volume and/or hardness. The invention disclosed here thus makes it possible, among other things, to specifically harden, soften and/or change the respective shape of the fabric cover, the seat surface, backrest, head/neck support, lumbar support and/or side bolsters of a seat.

One step of the method includes obtaining occupancy-relevant data of the user for this seat. The occupancy-relevant data includes a size of the user, for example a height, a leg length and/or a back length. The size of the user is detected or recorded by at least one imaging sensor as a data source. The imaging sensor is, for example, a camera. According to one aspect, the imaging sensor is integrated in the seat. According to another aspect, the imaging sensor, for example a camera, is arranged in the space in which the seat is used; for example, in the case of a vehicle seat, the camera can be arranged as an interior camera in a vehicle interior with a field of view that can be aligned with the seat.

A further step of the method includes providing metadata relating to an ergonomic sitting position of the user on the seat. The metadata includes at least an angle between the leg and the back of the user in the ergonomic sitting position. In addition to the shape and design of the seat, the ergonomic sitting position can also relate to a seat height and a distance to an application object, for example the distance to the steering wheel and/or to an airbag in the case of a vehicle seat. For example, in the ergonomic sitting position in the vehicle seat, one is not too close to an airbag, so that no disproportionate damage is sustained in the event of an accident. The respective metadata relates to the respective user and the respective seat. The angle between the leg and the back is, for example, chosen so that a seat height that is not too low is achieved for this user on this seat. According to a further aspect, the metadata includes height, gender, medical information and other user data. This improves individual adjustment.

A further step of the method involves processing the occupancy-relevant data and the metadata with an algorithm from the field of machine learning. By linking the occupancy-relevant data and the metadata, the algorithm learns the ergonomically adjusted setting of the seat based on the occupancy-relevant data after several process steps. The algorithm also learns the signals required for the control unit to adjust the properties of the intelligent material for the ergonomically adjusted setting of the seat. This supports the technical purpose of adjusting the ergonomic sitting posture. Machine learning is realized by algorithms that are implemented in software code sections. An open source platform for programming machine learning is, for example, TensorFlow written in the Python programming languages. The platform is, for example, a Linux, macOS, Microsoft Windows, Android or JavaScript platform. After several process steps of occupancy-relevant data with the associated metadata, occupancy data for a seat with regard to an ergonomic sitting posture can advantageously be determined in the application case, i.e. after training, without having to resort to corresponding metadata. For example, the learning algorithm is gradient-based and the process steps include forward feeds and backward feeds of the algorithm in the concept of supervised learning. The control unit is, for example, an electronic control unit. The signal is, for example, a control signal, for example an electrical voltage.

A further step of the method includes outputting the signals to the control unit. The control unit is, for example, an electronic control unit. Based on the signals, the control unit adjusts at least one property of the intelligent material, for example the volume of the intelligent material integrated into a seat surface, in order to adjust the seat height for the ergonomic sitting position.

Another aspect is that a seat setting is learned that is adapted to the length of the journey or flight, including a seat setting that is constantly adjusting. For example, after a certain time, the left or right seat muscles of a user are more stressed, sometimes the user falls a little deeper into the seat, sometimes the user sits up straighter.

According to a further aspect, the adjustable property of the intelligent material is a shape and/or a hardness of the intelligent material. For example, a precisely fitting hardness of the seat is set for a comfortable sitting position that is individually adapted to the user. According to a further aspect, the metadata, for example in the area of application of the vehicle seat, includes information on the driving style and/or driving mode, for example sport or comfort, in order to adapt the driving style and/or driving mode to achieve a seat adjustment adapted to the vehicle's needs using the intelligent material. For example, it learns that the intelligent material, integrated in the seat, for example, should harden in a sport mode.

According to another aspect, the occupancy-relevant data includes the user's weight. The weight is obtained from a weight sensor as a data source. According to one aspect, the weight sensor, for example a pressure sensor or strain gauge, is integrated in the seat. Depending on the weight, the hardness of the intelligent material can be adjusted, for example.

According to a further aspect, the metadata includes a length of the seat and/or a viewing angle of the user. The length of the seat is also important for ergonomic sitting posture: if it is too short, the legs become tense. According to one aspect, the length of the seat corresponds to the length of the user's thighs. According to another aspect, for example, the seat is adjusted in such a way that there is no contact between the seat, for example the vehicle seat, and the user's lower leg. For a vehicle seat, for example, the viewing angle is in a range of values in which a good all-round view of the road traffic, road users and/or instruments in the vehicle cockpit is achieved. According to one aspect, intelligent materials are integrated into the seat, by means of which the length of the seat can be adjusted by means of an adjustable change in shape. According to one aspect, the seat comprises a headrest into which intelligent materials are integrated, by means of which the height and/or inclination of the headrest can be adjusted by means of an adjustable change in shape in order to obtain a predetermined viewing angle for a specific user.

According to a further aspect, the algorithm comprises an artificial neural network. The artificial neural network comprises several layers. The layers comprise fully connected and/or convolutional layers. An artificial neural network is a collection of individual information processing units, called neurons, which are arranged layer by layer in a network architecture. At the beginning of training, each neuron has a random initial weight. In the individual process steps, for example forward and backward feeds, the artificial neural network learns increasingly better, for example based on the least squares method, the cross-entropy known from information theory or the gradient descent method, to establish associations between the occupancy-relevant data and the metadata. In a fully connected network, each neuron in a layer is connected to all neurons in the preceding layer. Each connection has its own weight. The artificial neural network alternatively or additionally comprises convolutional layers. In a convolutional neural network, also called a convolutional network, a filter is applied to a layer of neurons with the same weights regardless of position. According to a further aspect of the invention, after the learning phase, all weights are saved as a so-called checkpoint. The artificial neural network can then be restarted at any time with these optimized weights, which is particularly advantageous for the deployment phase.

According to a further aspect, the invention provides a computer-implemented method for adjusting at least one property of at least one intelligent material of a seat for an adjustment of the seat that is ergonomically adapted to a user. This method corresponds to a method of using the algorithm from the field of machine learning.

One step of the method includes recording occupancy-relevant data of the user using at least one sensor as a data source. The sensor can be integrated in the seat. For example, the sensor can be a weight sensor, for example a pressure sensor or a strain gauge, or an imaging sensor, for example a camera. The imaging sensor can be arranged on the seat, for example on a headrest, or in or on the application object, for example in an interior of a vehicle, for example on an interior mirror or on a cockpit instrument.

A further step in the process involves processing the data with an algorithm from the field of machine learning, for example with an artificial neural network. The algorithm is trained according to the training process described above to determine the properties of the intelligent material for the ergonomically adapted adjustment of the seat. The properties of the intelligent material are, for example, its volume, shape and/or hardness.

A further step of the method includes obtaining signals for adjusting this property of the intelligent material after executing the algorithm. This means that the algorithm also determines the signals that are required to adjust the particular property of the intelligent material. The signal is, for example, a control signal, for example an electrical voltage.

A further step in the process involves providing the signals to the intelligent material. The material is activated by the signals, for example to change its shape.

According to a further aspect, the invention provides a control device for adjusting at least one property of at least one intelligent material of a seat for an adjustment of the seat that is ergonomically adapted to a user. The control device is, for example, an electronic control device. According to one aspect, the control device is integrated in the seat.

The control unit comprises at least one first interface, also called a data input interface, to at least one sensor as a data source that records occupancy-relevant data from the user. The sensor is, for example, an imaging sensor, such as a camera, and/or a weight sensor, such as a pressure sensor or a strain gauge.

The control unit comprises a processor unit. The processor unit reads the data via at least the first interface and processes the data using an algorithm from the field of machine learning. The algorithm is to determine the property of the intelligent material for the ergonomically adapted adjustment of the seat according to the training method described above. The processor unit then determines signals for setting this property of the intelligent material.

The control unit comprises at least one second interface, also called a data output interface, to the at least one intelligent material, via which the signals are provided to the intelligent material.

According to a further aspect, the invention provides a computer program for setting at least one property of at least one intelligent material of a seat for an ergonomically adapted setting of the seat for a user. The computer program comprises program instructions that cause a computer to carry out the above-described method of use when the computer program is executed or loaded by the computer. The computer program serves a technical purpose when it runs on the computer, namely setting the ergonomically adapted setting of the seat. The computer program can be provided as interpretable code that is loaded onto the computer, for example directly into an internal memory of the computer, or as compiled code that is executed by the computer. The provision can be made, for example, on a machine-readable carrier on which the computer program is stored, or via a data carrier signal. The program instructions can be machine instructions, source text or object codes, written in assembly language, an object-oriented programming language, for example C++, or in a procedural programming language, for example C. According to one aspect, the computer program is executed by the control device described above. For example, the computer program is stored in a working memory of the control unit. According to a further aspect, the computer program is provided to the control unit by means of a data carrier signal. For example, the control unit comprises middleware which mediates between the application of the computer program and the application software of the control unit.

According to a further aspect, the invention provides a seat. The seat comprises a seat surface, a backrest, at least one intelligent material that is integrated into the seat surface and/or the backrest, and a sensor that records occupancy-relevant data of a user of the seat. The seat also comprises a control device as described above, which is connected to the sensor and the at least one intelligent material in a signal-transmitting manner and sets at least one property of the intelligent material for an adjustment of the seat that is ergonomically adapted to the user. According to a further aspect, the seat comprises a head/neck support into which the intelligent material is integrated. According to a further aspect, the seat comprises a lumbar support to support an ergonomic posture for the spine of the seated user. According to a further aspect, the seat comprises side bolsters into which the intelligent material is integrated and by means of which the seat is adapted to bodies of different widths, wherein according to one aspect the side bolsters sit close to the user's thighs in the ergonomic sitting position. The seat can be, for example, a seat in an airplane or a cinema seat.

In one aspect, the seat is a vehicle seat, for example a seat in a car, truck or shuttle. In another aspect, the seat includes electrically adjustable springs or motors that can additionally adjust seat width, seat height, lumbar support and/or neck support, seat tilt angle and/or distance to the steering wheel.

• 1 ein Ausführungsbeispiel eines hier offenbarten Sitzes,
• 2 ein Ausführungsbeispiel eines hier offenbarten Steuergeräts,
• 3 ein Ausführungsbeispiel eines hier offenbarten Trainingsverfahrens und
• 4 ein Ausführungsbeispiel eines hier offenbarten Einsatzverfahrens.

The invention is explained using the embodiments described in the following figures. They show

• 1 an embodiment of a seat disclosed here,
• 2 an embodiment of a control device disclosed here,
• 3 an embodiment of a training method disclosed here and
• 4 an embodiment of an application method disclosed here.

1 shows a seat 1, for example a vehicle seat. The seat 1 comprises a seat back and a backrest. Intelligent materials 2 are integrated into the seat back and the backrest, which change their shape when controlled by a signal, for example an electrical signal, or a set temperature difference. The adjustable change in shape is used to achieve an ergonomic seat setting that is individually adapted to a user.

To adjust the shape change and thus also the ergonomic seat setting, the seat includes a control unit 3. The control unit 3 reads the user's data relevant to seat occupancy via a first interface 5. The data is recorded via the first interface 5 using a sensor 4. The sensor 4 is, for example, a weight sensor integrated into the seat 1. The sensor 4 can also be an imaging sensor 4, for example a camera, see 2 .

The control unit 3, the 2 shows in detail, comprises a processor unit 6. The computer program disclosed here for setting at least one property of at least one intelligent material 2 of a seat 1 for an ergonomically adapted setting of the seat 1 for a user can be loaded into the processor unit 6 or executed by the processor unit 6. Depending on the occupancy-relevant data, the control unit 3 then determines signals for setting the corresponding property of the intelligent material. The signals are provided to the intelligent material 2 via a second interface 7. The intelligent material then adjusts its shape for the ergonomic seat adjustment, see also 2 .

3 shows the computer-implemented method disclosed here for machine learning an ergonomically adapted setting of the seat 1 for a user. In a step T1, occupancy-relevant data of the user for this seat 1, comprising a size of the user, is obtained from at least one imaging sensor 4 as a data source. In a step T2, metadata relating to an ergonomic sitting posture of the user on the seat is provided. In a step T3, the occupancy-relevant data and the metadata are processed using an algorithm from the field of machine learning. By linking the occupancy-relevant data and the metadata after several process steps, the algorithm learns the ergonomically adapted setting of the seat 1 based on the occupancy-relevant data and the signals required for this for the control unit 3 for setting the property of the intelligent material 2 for the ergonomically adapted setting of the seat 1. In a step T4, the signals for the control unit 3 are output.

4 shows the computer-implemented method disclosed here for setting at least one property of at least one intelligent material 2 of a seat 1 for an ergonomically adapted setting of the seat 1 for a user. In a step E1, occupancy-relevant data of the user are recorded using at least one sensor 4 as a data source. In a step E2, the data are processed with an algorithm from the field of machine learning, which is trained, for example, after steps T1 to T4 to determine the property of the intelligent material 2 for the ergonomically adapted setting of the seat 1. In a step E3, signals for setting this property of the intelligent material 2 are received from the algorithm after execution of the algorithm. In a step E4, the signals are provided to the intelligent material 2.

Reference symbols

1

Seat

2

intelligent material

3

Control unit

4

sensor

5

first interface

6

Processor unit

7

second interface

T1-T4

Process steps

E1-E4

Process steps

QUOTES INCLUDED IN THE DESCRIPTION

This list of documents listed by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions.

Cited patent literature

• DE 102019004095 A1 [0004]
• EP 3789250 B1 [0005]

Claims (10)

Computer-implemented method for machine learning an ergonomically adapted setting of a seat (1) for a user, the seat comprising at least one intelligent material (2), wherein at least one property of the intelligent material (2) can be adjusted by means of at least one control device (3), and at least one seat surface and one backrest, the method comprising the steps: • Obtaining occupancy-relevant data of the user for this seat (1) comprising a size of the user from at least one imaging sensor (4) as a data source (T1); • Providing metadata relating to an ergonomic sitting position of the user on the seat (1), the metadata comprising at least one angle between the leg and back of the user in the ergonomic sitting position (T2); • Processing the occupancy-relevant data and the metadata with an algorithm from the field of machine learning, whereby the algorithm, by linking the occupancy-relevant data and the metadata after several process steps, learns the ergonomically adapted setting of the seat (1) based on the occupancy-relevant data and the signals required for this for the control unit (3) for setting the property of the intelligent material (2) for the ergonomically adapted setting of the seat (1) (T3); • Outputting the signals for the control unit (3) (T4). Procedure according to Claim 1 , wherein the adjustable property of the intelligent material (2) is a shape and/or a hardness of the intelligent material (2). Method according to one of the preceding claims, wherein the occupancy-relevant data comprise a weight of the user obtained from a weight sensor (4) as a data source. Method according to one of the preceding claims, wherein the metadata comprises a length of the seat and/or a viewing angle of the user. Method according to one of the preceding claims, wherein the algorithm is an artificial neural network comprising several layers, the layers comprising fully connected and/or convolutional layers. Computer-implemented method for setting at least one property of at least one intelligent material (2) of a seat (1) for an ergonomically adapted setting of the seat (1) for a user, the method comprising the steps: • Recording occupancy-relevant data of the user using at least one sensor (4) as a data source (E1); • Processing the data with an algorithm from the field of machine learning, wherein the algorithm is trained according to the method according to one of the preceding claims to determine the property of the intelligent material (2) for the ergonomically adapted setting of the seat (1) (E2); • Receiving signals for setting this property of the intelligent material (2) after executing the algorithm (E3); • Providing the signals to the intelligent material (2) (E4). Control device (3) for setting at least one property of at least one intelligent material (2) of a seat (1) for an ergonomically adapted setting of the seat (1) to a user, the control device comprising • at least one first interface (5) to at least one sensor (4) as a data source, which records occupancy-relevant data of the user; • a processor unit (6) which reads the data via the at least first interface (5), processes the data with an algorithm from the field of machine learning, wherein the algorithm is carried out according to the method according to one of the Claims 1 until 5 is trained to determine the property of the intelligent material (2) for the ergonomically adapted adjustment of the seat (1), and determines signals for setting this property of the intelligent material (2); • at least one second interface (7) to the at least one intelligent material (2), via which the signals are made available to the intelligent material (2). Computer program for adjusting at least one property of at least one intelligent material (2) of a seat (1) for an adjustment of the seat (1) that is ergonomically adapted to a user, the computer program comprising program commands that cause a computer to carry out the method according to Claim 6 to execute when the computer program is executed or loaded by the computer. Seat (1) comprising • a seat surface; • a backrest; • at least one intelligent material (2) which is integrated into the seat surface and/or the backrest; • a sensor (4) which records occupancy-relevant data of a user of the seat (1); • a control device (3) according to Claim 7 , transmitting the signal with the sensor (4) and the little at least one intelligent material (2) and adjusts at least one property of the intelligent material (2) for an adjustment of the seat (1) ergonomically adapted to the user. Seat (1) after Claim 9 , wherein the seat (1) is a vehicle seat.

Images