Method for collecting data for machine learning

Abstract

The technical solution relates to the field of transportation, and more particularly to auxiliary devices and methods for collecting data for machine learning using vehicles such as, for example, a personal transporter. What is proposed is a method for collecting data for machine learning. A need exists to speed up the preparation of data suitable for machine learning and thus reduce the time taken to prepare machine learning models trained on big data. The technical result achieved by the claimed technical solution, in addition to providing a product and / or process which fulfils its intended purpose, is that of increasing the speed of preparation of data suitable for machine learning, including increasing the speed of preparation of big data which can be used for machine learning. In certain aspects of the invention, a further technical result achieved is that of also improving road traffic safety.

Description

A METHOD OF DATA COLLECTION FOR MACHINE LEARNING

[0001] AREA OF TECHNOLOGY

[0002] The technical solution relates to the field of transport, in particular to assistive devices and methods for collecting data for machine learning using vehicles, such as personal mobility devices (PMDs).

[0003] LEVEL OF TECHNOLOGY

[0004] Various methods for collecting data for machine learning are known, including using vehicles and various vehicle sensors. The data collected in this way can be raw (unadjusted or unaltered, providing no value until labeled or altered, which creates value) or further labeled or altered to enable its use for machine learning, for example, to create a neural network used to predict the maximum speed limit based on the collected data. However, there are no known solutions that allow for labeling and / or altering and / or deleting and / or not recording raw data directly during its collection.

[0005] Moreover, from the patent document DE102012220011B4, published on 17.11.2022 (D1), a control device for controlling an electric bicycle is known, comprising: a detection device intended for detecting sensor data; a computer device connected to the detection device and intended for determining at least one environmental parameter based on the detected sensor data; and an interface device connected to the computer device and intended for controlling the electric motor of the electric bicycle and limiting the speed of the electric bicycle based on the determined at least one environmental parameter.The solution known from D1 does not provide the ability to influence the collected raw data directly during its collection by the data assessor (data collection specialist), due to which any data collected in this way requires their further cleaning, which consists, for example, but not limited to, in their labeling, and / or modification, and / or even deletion, which inevitably affects the speed of machine learning, as well as its accuracy and efficiency, since for accurate and effective machine learning a sufficient amount of data suitable for machine learning is necessary. which takes time to prepare from raw data, which is especially important when using neural networks trained on big data.

[0006] Thus, there is a need to speed up the preparation of machine learning-ready data and thus reduce the training time of machine learning models trained on big data.

[0007] The solution known from D1 can be accepted as the closest analogue of the declared technical solution.

[0008] DISCLOSURE OF THE TECHNICAL SOLUTION

[0009] The technical problem solved by the claimed technical solution is the creation of a product and / or method and / or application thereof that avoids the disadvantages of the prior art and thus accelerates the process of preparing data suitable for machine learning, including accelerating the process of preparing big data that can be used for machine learning. In some aspects, other technical problems solved also include the creation of a product and / or method and / or application that ensures increased road safety. Another technical problem solved by the claimed technical solution is the creation of a product and / or method and / or application that expands the arsenal of technical means—methods for collecting data for machine learning and / or methods for monitoring and / or determining the maximum permissible speed of a vehicle.

[0010] The technical result achieved by implementing the claimed technical solution, in addition to the product and / or method fulfilling its intended purpose, is an increase in the speed of preparing data suitable for machine learning, including an increase in the speed of preparing big data that can be used for machine learning. In some respects, another technical result achieved is an increase in road safety.

[0011] The technical result is achieved due to the fact that a method for collecting data for machine learning is provided, characterized by at least: obtaining raw data for machine learning using at least one environment sensor, and recording at least a portion of the obtained raw data on a machine-readable medium using a data collection device; wherein during the data collection process, at least a control signal is generated for the data collection device for machine learning, the control signal contains at least a command for the processor of the data collection device, prompting the processor of the collection device data to execute program code that, when executed by the processor of the data collection device, provides an effect on at least a portion of the raw data being collected.

[0012] BRIEF DESCRIPTION OF DRAWINGS

[0013] Illustrative embodiments of the present utility model are described below in detail with reference to the accompanying drawings, which are incorporated herein by reference, and in which:

[0014] Fig. 1 shows exemplary diagrams of systems 100, 200 for collecting data for machine learning, a system 300 for training machine learning models, and a vehicle control system 400.

[0015] IMPLEMENTATION OF THE TECHNICAL SOLUTION

[0016] In the context of the present invention, the term "big data" preferably, without limitation, means large data sets characterized primarily by such characteristics as volume, diversity, processing speed and / or variability, which require the use of scalable technology for efficient storage, processing, management and analysis, as provided for, for example, but not limited to, ISO / IEC 20546:2019, GOST R ISO / IEC 20546-2021 and other standards.

[0017] In the context of the present invention, the term "maximum permissible speed" preferably, without limitation, means the maximum threshold value of the speed of movement (displacement) of a vehicle, determined by the current regulatory restrictions for such a vehicle on the section of the road along which such a vehicle moves using its own engine, while preferably, without limitation, without taking into account any non-penalty threshold for exceeding the maximum permissible speed.

[0018] In the context of the present invention, the term "permissible speed" preferably, without limitation, means a value or range of values ​​of the speed of movement (displacement) of a vehicle, which may or may not include the maximum permissible speed of such vehicle.

[0019] In the context of the present invention, the term "raw data" preferably, without limitation, means uncleaned and / or labeled and / or modified data that is not suitable for machine learning before being cleaned and / or labeled and / or modified, but that may become suitable for machine learning after being cleaned and / or labeled and / or modified.

[0020] In the context of the present invention, the term "traffic flow" preferably, without limitation, means a collection of vehicles, typically present on a particular section of the route at a particular time of day, preferably without limitation, moving on a section of the route without exceeding the maximum permissible speed in accordance with the regulatory restrictions established for them, or restrictions determined by the logical thinking of vehicle drivers.

[0021] In the context of the present invention, the term "pedestrian traffic flow" preferably, without limitation, means the traffic flow in combination with pedestrians normally present on a particular section of the path, preferably, without limitation, moving on a section of the path in accordance with the regulatory restrictions established for them, or restrictions determined by the logical thinking of pedestrians.

[0022] In the context of the present invention, the term "electric personal mobility vehicle" (EPMV) preferably, without limitation, means a vehicle having one or more propellers, intended for the movement of no more than one person by using at least one electric motor with a maximum permissible speed, in the absence of other restrictions, of no more than 25 km / h for sections of the route running outside a residential area or courtyard area, and with a maximum permissible speed of no more than 20 km / h for sections of the route running within a residential area or courtyard area.

[0023] In the context of the present invention, the term "electric bicycle" means a bicycle equipped with at least one electric motor used to propel the electric bicycle and / or to facilitate the effort expended in propelling the electric bicycle, with a maximum permissible speed, in the absence of other restrictions, of no more than 60 km / h for sections of the route running outside a residential area or courtyard area, and with a maximum permissible speed of no more than 20 km / h for sections of the route running within a residential area or courtyard area.

[0024] In a preferred embodiment of the present invention, there is provided a method for collecting data for machine learning, characterized at least by: obtaining raw data for machine learning using at least one environment sensor, and recording at least a portion of the obtained raw data on a machine-readable medium using a data collection device; wherein during the data collection process, at least a control signal is generated for the data collection device for machine learning, the control signal contains at least a command for the processor of the data collection device, causing the processor of the data collection device to execute program code, which, when executed by the processor the data collection device provides an effect on at least a portion of the raw data being collected.

[0025] In a particular embodiment of the present invention, any of the mentioned data collection methods is provided, characterized in that the action is at least: marking and / or changing at least a portion of the raw data recorded on a machine-readable medium, and / or deleting at least a portion of the raw and / or machine learning data recorded on a machine-readable medium, and / or not recording at least a portion of the collected raw data on a machine-readable medium.

[0026] In a particular embodiment of the present invention, any said method of collecting data is provided, characterized in that the data collection is carried out using a data collection device associated with at least one said environment sensor and / or containing at least one said environment sensor.

[0027] In a particular embodiment of the present invention, any of the mentioned data collection methods is provided, characterized in that the environment sensor is selected from or formed by any combination of: radar, lidar, sonar, camera.

[0028] In a particular embodiment of the present invention, any said method of collecting data is provided, characterized in that said environment sensor is moved in space by means of a data collection vehicle.

[0029] In a particular embodiment of the present invention, any of the aforementioned methods for collecting data is provided, characterized in that the movement is carried out at least together with the normal traffic flow and / or together with the normal pedestrian-traffic flow; wherein during the movement, the speed of movement is selected to be no higher than the maximum permissible speed of movement for a vehicle for the corresponding section of the route.

[0030] In a particular embodiment of the present invention, any of the said data collection methods is provided, characterized in that the speed of movement is changed at least depending on the distance to the obstacle.

[0031] In a particular embodiment of the present invention, any of the above mentioned data collection methods is provided, characterized in that a control signal is generated when the travel speed does not correspond to the maximum permissible travel speed for the vehicle on the corresponding section of the route, when In this case, the distance to the obstacle is chosen such that a speed of movement corresponding to the maximum permissible speed of movement for a vehicle on the corresponding section of the road is allowed.

[0032] In a particular embodiment of the present invention, any of the mentioned data collection methods is provided, characterized in that the control signal is generated when no movement is performed and / or when the movement is performed without using a motor at a speed not exceeding 1.4 m / s.

[0033] In a particular embodiment of the present invention, any of the mentioned data collection methods is provided, characterized in that the control signal is generated by means of a signaling device associated with the data collection device, or with which the data collection device is provided.

[0034] In a particular embodiment of the present invention, any of the said data collection methods is provided, characterized in that the data collection vehicle is an electric personal mobility vehicle or an electric bicycle.

[0035] In a particular embodiment of the present invention, any of the above mentioned data collection methods is provided, characterized in that the vehicle is rented for a period of time, and the possibility of using such a vehicle is ensured by a system for providing the possibility of independent use of vehicles on demand.

[0036] In another preferred embodiment of the present invention, a device for collecting data for machine learning is provided, comprising at least: one or more processors; a memory containing program code that, when executed by the processor, causes the processor to perform, in any order, at least the following processes: receiving raw data from at least one environment sensor, recording at least a portion of the received raw data on a machine-readable storage medium, receiving a control signal, and then performing at least an action on at least a portion of the collected raw data.

[0037] In a particular embodiment of the present invention, any said data collection device is provided, characterized in that the action is at least: marking and / or changing at least a portion of the raw data recorded on the machine-readable medium, and / or deleting at least a portion of the raw and / or machine learning-suitable data recorded on the machine-readable medium. data, and / or failure to record at least part of the collected raw data on a machine-readable medium.

[0038] In a particular embodiment of the present invention, any said data collection device is provided, characterized in that the environment sensor is selected from or formed by any combination of: radar, lidar, sonar, camera.

[0039] In a particular embodiment of the present invention, any said data collection device is provided, characterized in that the environment sensor is an external device connected to the data collection device by a communication line.

[0040] In a particular embodiment of the present invention, any said data collection device is provided, characterized in that the environment sensor is a vehicle environment sensor for collecting data or is attached to a vehicle for collecting data.

[0041] In a particular embodiment of the present invention, there is provided any said data collection device, characterized in that it contains a situation sensor.

[0042] In a particular embodiment of the present invention, a said data collection device is provided, characterized in that it is designed to be used in conjunction with a vehicle during its movement.

[0043] In a particular embodiment of the present invention, any said data collection device is provided, characterized in that the environment sensor moves in space by means of a vehicle.

[0044] In a particular embodiment of the present invention, a said data collection device is provided, characterized in that the environment sensor moves in space, as previously indicated with reference to the data collection method.

[0045] In a particular embodiment of the present invention, there is provided any said data collection device, characterized in that the control signal is generated as previously indicated with reference to the data collection method.

[0046] In a particular embodiment of the present invention, there is provided any said data collection device, characterized in that the machine-readable medium onto which the raw data is recorded is the memory of the data collection device.

[0047] In a particular embodiment of the present invention, there is provided any said data collection device, characterized in that the machine-readable medium on which the raw data is recorded is external and is connected to the data collection device via a data transmission interface or a wireless communication line.

[0048] In a particular embodiment of the present invention, any said data collection device is provided, characterized in that the control signal is generated by means of a signaling device.

[0049] In a particular embodiment of the present invention, there is provided any said data collection device, characterized in that the signaling device is not part of said data collection device and is connected to said data collection device via a communication line.

[0050] In a particular embodiment of the present invention, there is provided any said data collection device, characterized in that the signaling device is part of said data collection device.

[0051] In a particular embodiment of the present invention, any said data collection device is provided, characterized in that the signaling device contains a switching means, the operation of which generates a control signal for the data collection device.

[0052] In another preferred embodiment of the present invention, a data collection system for machine learning is provided, comprising at least one or more of the said data collection devices, configured to form at least one set of data for machine learning from data recorded on a machine-readable medium, and to transmit the formed set of data via a wireless communication link; a server connected via a wireless communication link to one or more of the said data collection devices, comprising at least: one or more server processors, a memory containing program code that, when executed by the server processor, ensures at least the receipt of at least one said set of data, recording the received set of data in the server memory;wherein at least one said data set includes at least one of or a combination of: raw data, marked-up raw data, modified raw data; wherein such data set does not include at least: deleted raw data, data not recorded on a machine-readable storage medium.

[0053] In another preferred embodiment of the present invention, a vehicle for collecting data for machine learning is provided, comprising at least one environment sensor associated with at least one of said data collecting devices.

[0054] In a particular embodiment of the present invention, there is provided any said data collection vehicle, characterized in that it is an electric personal mobility vehicle or an electric bicycle.

[0055] In a particular embodiment of the present invention, there is provided any said vehicle for collecting data, characterized in that it is equipped with at least one said data collection device.

[0056] In another preferred embodiment of the present invention, a method for collecting data for machine learning is provided, characterized in that, using any said vehicle, any said environment sensor is moved along a section of a path, while any said control signal is generated in any said manner for any said data collection device.

[0057] In another preferred embodiment of the present invention, a data collection system for machine learning is provided, comprising at least one or more of the said data collection devices, configured to form at least one set of data for machine learning from data recorded on a machine-readable medium, and to transmit the formed set of data via a wireless communication link; a server connected via a wireless communication link to one or more of the said data collection devices, comprising at least: one or more server processors, a memory containing program code that, when executed by the server processor, ensures at least the receipt of at least one said set of data, recording the received set of data in the server memory;wherein at least one said data set includes at least one of or a combination of: raw data, marked-up raw data, modified raw data; wherein such data set does not include at least: deleted raw data, data not recorded on a machine-readable storage medium; wherein the system includes one or more vehicles equipped with any said data collection device.

[0058] In another preferred embodiment of the present invention, a method for training a machine learning model, executable by a processor of a computer device, is provided, in which, using at least one processor of at least one computer device, a machine learning model is trained to determine the permissible speed of movement for a vehicle depending on a frame of the situation supplied to the input of the machine learning model; wherein the frame of the situation is obtained by means of a vehicle equipped with at least one sensor of the situation, configured with the possibility of obtaining a frame of the situation; wherein the training of the machine learning model is ensured using data obtained by any of the mentioned data collection methods.

[0059] In a particular embodiment of the present invention, any mentioned method for training a machine learning model is provided, in which the machine learning model is trained to solve a regression problem.

[0060] In another preferred embodiment of the present invention, a device for training a machine learning model is provided, comprising at least: one or more processors, a memory containing program code, which, when executed by the processor of the device for training a machine learning model, ensures the execution of any mentioned method for training a machine learning model.

[0061] In another preferred embodiment of the present invention, a system for training a machine learning model is provided, comprising at least: one or more devices for training a machine learning model; and one or more said vehicles for collecting data for machine learning, using which any said method for collecting data for machine learning can be performed, wherein the device is configured to use at least data obtained by such a data collection method for training a machine learning model.

[0062] In another preferred embodiment of the present invention, a machine-readable data carrier is provided, containing a program code that, when executed by at least one processor of at least one computing device, ensures the execution of any mentioned method of training a machine learning model; and / or containing a machine learning model trained by means of any mentioned method of training a machine learning model.

[0063] In another preferred embodiment of the present invention, there is provided a method executable by a processor of a computing device for determining the permissible speed of movement of a vehicle, in which, using at least one processor of at least one computer device, any said frame of the situation is fed to the input of a machine learning model trained by any said method of training the machine learning model, and the value of the permissible speed of movement of the vehicle for this frame of the situation is obtained at the output.

[0064] In another preferred embodiment of the present invention, a device for determining the permissible speed of a vehicle is provided, comprising at least: one or more processors of this device, a memory containing a program code that, when executed by the processor, causes the processor to perform any mentioned method for determining the permissible speed of a vehicle.

[0065] In another preferred embodiment of the present invention, a system for determining the permissible speed of a vehicle is provided, comprising at least: one or more of any of the mentioned devices for determining the permissible speed of a vehicle; and one or more vehicles associated with any of the mentioned device for determining the permissible speed of a vehicle and equipped with at least one situation sensor configured to obtain any of the mentioned situation frame.

[0066] In another preferred embodiment of the present invention, a machine-readable data carrier is provided containing a program code which, when executed by at least one processor of at least one computer device, ensures the execution of any of the said methods for determining the permissible speed of a vehicle.

[0067] In another preferred embodiment of the present invention, a method for controlling a vehicle, executed by a processor of a computer device, is provided, in which at least one processor of at least one computer device at least: identifies at least one frame of the situation; using the identified frame of the situation, by means of any mentioned method for determining the permissible speed of the vehicle, determines the permissible speed of the vehicle for this frame of the situation; and at least generates a control signal for the control system of the vehicle, including at least stored in the memory of the computer device of the control system vehicle setting, including the said specific permissible speed of movement of the vehicle for the said frame of the situation.

[0068] In another preferred embodiment of the present invention, a vehicle control device is provided, comprising at least: one or more control device processors, a memory containing a program code that, when executed by the control device processor, causes the processor to perform any said method of controlling the vehicle.

[0069] In another preferred embodiment of the present invention, a vehicle control system is provided, comprising at least: an engine configured to drive at least one propeller; and at least one of the said vehicle control devices, configured to at least generate a control signal for the engine and / or configured to at least generate an information signal for the vehicle information system; wherein the said control signal is generated using the said setting; wherein the said information signal is generated using the said setting.

[0070] In another preferred embodiment of the present invention, a machine-readable data carrier is provided containing a program code which, when executed by at least one processor of at least one computer device, enables the execution of any of the said methods for controlling a vehicle.

[0071] The following are embodiments of the present invention, revealing examples of its implementation in specific embodiments. However, the description itself is not intended to limit the scope of rights granted by this patent. Rather, it should be understood that the claimed invention may also be implemented in other ways that incorporate different elements and conditions, or combinations of elements and conditions similar to those described herein, in combination with other existing and future technologies.

[0072] In a preferred embodiment of the present invention, a method for collecting data for machine learning is provided, characterized at least by: obtaining raw data for machine learning using at least one environment sensor, and recording at least a portion of the obtained raw data on a machine-readable medium using a data collection device; wherein, during the data collection process, at least a control signal is generated for a data collection device for machine learning, the control signal comprising at least a command for the processor of the data collection device, causing the processor of the data collection device to execute program code, which, when executed by the processor of the data collection device, provides an effect on at least a portion of the raw data being collected. In this case, preferably, without being limited to, said effect is at least: marking and / or changing at least a portion of the raw data recorded on a machine-readable medium, and / or deleting at least a portion of the raw and / or suitable for machine learning data recorded on a machine-readable medium, and / or not recording at least a portion of the collected raw data on a machine-readable medium.For example, but not limited to, this may enable the collection of raw data, such as, but not limited to, still frames of the environment obtained by one or more environment sensors. A still frame may, but not limited to, be a static representation of the environment within the detection range of the environment sensor. For example, but not limited to, the environment sensor may be selected from or comprised of any combination of radar, lidar, sonar, or a camera. For example, but not limited to, when the environment sensor is a camera, the still frame may be an image obtained using the camera by, but not limited to, photography or video recording, including an image that is a video frame. For example, but not limited to, when the environment sensor is a lidar, the still frame may be at least a point cloud obtained using the lidar, at least suitable for creating a three-dimensional scene for subsequent analysis.For example, without limitation, when the situation sensor is a radar or sonar, then the situation frame is at least a set of reflection signals obtained using the radar, providing at least the ability to detect objects and distances to them.In this case, without being limited, it should be obvious to a person skilled in the art, possessing ordinary knowledge, for whom the present invention is intended, that any suitable combination of the mentioned variants of the environment sensor can be provided as a sensor of the environment, in which, for example, without being limited, frames of the environment of one type can be supplemented with features extracted from frames of the environment of other types, thereby ensuring the receipt of more complete and thus enriched frames of the environment; in this case, without being limited, it should mainly be assumed that at least a frame of the environment must include some reproduction of the space in the area of ​​​​the sensor, wherein the space does not necessarily include other objects. such as, but not limited to, obstacles in or near the vehicle's path of travel, such as, but not limited to, other vehicles and / or pedestrians and / or other objects typically present or encountered along the given section of the path.

[0073] Preferably, but not limited to, the data collection is carried out using a data collection device associated with at least one said environment sensor and / or containing at least one said environment sensor. Preferably, but not limited to, the said environment sensor is moved in space by means of a data collection vehicle; wherein, preferably, but not limited to, the movement is carried out in a predetermined manner along a predetermined route consisting of at least one path section. Preferably, but not limited to, the data collection vehicle is a vehicle containing at least one environment sensor associated with at least one of said data collection devices.Moreover, without being limited, it should be obvious to a person skilled in the art, possessing ordinary knowledge, for whom the present invention is intended, that any vehicle suitable for use in conjunction with the aforementioned environment sensor and / or data collection device may be selected as the vehicle, including, without being limited to, such a vehicle may be either manned or unmanned. For example, without being limited to, the vehicle may be an electric personal mobility vehicle (EPMV) or an electric bicycle; while, for example, without being limited to, an electric personal mobility vehicle may be an electric scooter, an electric skateboard, a hoverboard, a Segway, a unicycle, and the like.

[0074] For example, but not limited to, said situation sensor is moved at least together with the normal traffic flow and / or together with the normal pedestrian traffic flow. In this case, without limitation, the speed of movement is selected to be no higher than the maximum permissible speed of movement for the vehicle for the corresponding section of the route. For example, without limitation, the situation sensor is attached to any suitable ESIM in such a way that multiple frames of the situation can be obtained from a perspective in the direction of movement of the ESIM, after which they move along any route within the urban environment, ensuring the collection of data by means of the situation sensor. In this case, for example, without limitation, a route can be provided that includes several connected sections characterized, for example, but not limited to, by different pedestrian and vehicle traffic density, and / or characterized, for example but not limited to, by different maximum speeds for the ESIM. For example, but not limited to, a route may be provided consisting of at least two sections, one of which lies within a residential area or courtyard area, and the other lies outside the residential area or courtyard area. In this way, for example but not limited to, the required variability of the collected data may be ensured. Alternatively, for example but not limited to, a route may be provided consisting of at least two sections, one of which includes a dedicated lane for ESIM movement, and the other does not include such a dedicated lane and requires more careful movement. In this way, for example but not limited to, a situation sensor may be placed on the electric scooter, and a corresponding route may be provided for movement, ensuring the required variability of the collected data.In this case, for example, without being limited, the speed of movement is changed at least depending on the distance to an obstacle, such as, for example, without being limited, another vehicle, and / or a pedestrian, and / or another obstacle, located on the trajectory of movement of the data collection vehicle or near it, and also, without being limited, can be changed depending on the direction of the trajectory of movement; preferably, without being limited, the change in speed is carried out in such a way as to ensure safe movement, best minimizing the risk of a road traffic accident involving the data collection vehicle, which is preferably ensured by moving in accordance with the traffic rules established for the data collection vehicle.In this case, but not limited to, they ensure predominantly straight-line movement within the corresponding section of the route, taking into account obstacles created by existing objects, such as pedestrians or other vehicles. Thus, without limitation, upon detection of an obstacle, they ensure a smooth and safe change in speed in order to prevent an accident involving the data collection vehicle. In this case, for example, but not limited to, a control signal is generated when the travel speed does not correspond to the maximum permissible travel speed for the vehicle on the corresponding section of the route, while the distance to the obstacle is selected such that a travel speed corresponding to the maximum permissible travel speed for the vehicle on the corresponding section of the route is permitted.For example, without limitation, the vehicle may be provided with any decision-making means configured to determine at least. an approximate distance to an obstacle located on or near the path of the data collection vehicle, and predict the probability of a collision depending on the current speed of the data collection vehicle when not maneuvering; wherein, without limitation, such means can be configured to identify the current speed of the data collection vehicle and compare the identified current speed with the maximum permissible speed for a given section of the path; wherein, without limitation, when the identified speed is less than the maximum permissible speed for a given section of the path, and the possibility of safe movement at the maximum permissible speed is predicted, the generation of the said control signal is ensured. In addition, without limitation, the control signal is generated when the data collection vehicle is not moving and / or,when it is moved without the use of an engine at a speed not exceeding 1.4 m / s. For example, but not limited to, during movement along the data collection route, a situation may arise in which the data collection vehicle must stop moving in accordance with the signals of the road infrastructure, such as, for example, but not limited to, a traffic light. In this case, without limitation, an expected trajectory of movement of the data collection vehicle after receiving a signal permitting movement may be determined, wherein the trajectory includes an obstacle or runs near it; however, at the same time, the distance to the obstacle is such that it allows safe movement at the maximum permissible speed, but the data collection vehicle and, consequently, the situation sensor do not move and, in the absence of a control signal, the collection of data continues, which is thus irrelevant or even garbage,which will require subsequent cleaning of the collected data, in connection with which they ensure the formation of a control signal that ensures, for example, without limitation, the marking of data in such a way that such data are placed separately from suitable data, or completely ensures the deletion of collected garbage data, or completely ensures that the data received by the situation sensor is not recorded on a machine-readable data carrier intended for the data being collected; while, without limitation, when the said enabling signal is received and thus the possibility of movement of the data collection vehicle is restored, they ensure the cessation of the formation of the control signal, including if a suitable distance to the obstacle is ensured, after the data collection vehicle has gained, the maximum permissible speed for a given section of the path; alternatively, in this case, without limitation, a situation may be ensured where the enabling signal and the section of the path will require the movement of the data collection vehicle at an approximate pedestrian flow speed, preferably not exceeding 1.4 m / s, which, for example, without limitation, may be due to the fact that at the starting point of the section of the path, before receiving the enabling signal, the data collection vehicle was on a trajectory that requires dismounting and moving the vehicle without using its engine, which, for example, without limitation, is true for the situation where the vehicle was on the sidewalk in front of a ground-controlled pedestrian crossing, and the trajectory of movement runs along such a pedestrian crossing, in this case, without limitation, the continuation of the generation of the control signal is ensured until,until the data collection vehicle can be driven using its engine,

[0075] Fig. 1 shows exemplary diagrams of systems 100, 200 for collecting data for machine learning, a system 300 for training machine learning models and a vehicle control system 400. In another preferred embodiment of the present invention, a system 100 for collecting data for machine learning is provided, comprising at least one or more of any of the mentioned data collection devices 101, configured to generate at least one set of data for machine learning from data recorded on a machine-readable medium, and to transmit the generated set of data via a wireless communication link using, for example, a transceiver 1013;a server 102 connected, for example, using a transceiver 1023 via a wireless communication link with one or more of the said data collection devices 101, comprising at least: one or more server processors 1021, a memory 1022 containing a program code that, when executed by the server processor, ensures at least the receipt of at least one of the said data set, the recording of the received data set in the memory 1022 of the server 102; wherein at least one of the said data set includes at least one of or a combination of: raw data, raw data subjected to marking, raw data subjected to modification; wherein such a data set does not include at least: deleted raw data, data not recorded on a machine-readable storage medium.

[0076] In another preferred embodiment of the present invention, a data collection system 200 for machine learning is provided, comprising at least one or more of any of the said data collection devices 201, configured to generate at least one data set for machine learning from data recorded on a machine-readable medium, and to transmit the generated data set via a wireless communication link, for example, using a transceiver 2013; a server 202 connected, for example, using a transceiver 2023 via a wireless communication link with one or more of said data collection devices 201, comprising at least: one or more server processors 2021, a memory 2022 containing program code that, when executed by the processor 2021 of the server 202, ensures at least obtaining at least one of said data set, recording the obtained data set in the server memory; wherein at least one of said data set includes at least one of or a combination of: raw data, raw data subjected to labeling, raw data subjected to modification;wherein such a set of data does not include, at least: deleted raw data, data not recorded on a machine-readable storage medium; and the system 200 includes one or more vehicles 203 equipped with any mentioned data collection device 201 and at least one environment sensor 2032 connected to the data collection device 101, 201, for example, but not limited to, by a transceiver 1013, 2013, 2031.

[0077] In another preferred embodiment of the present invention, a system 300 for training a machine learning model is provided, comprising at least: one or more devices 301 for training a machine learning model, each of which at least comprises: one or more processors 3011, a memory 3012 containing a program code that, when executed by the processor of the device 301 for training a machine learning model, ensures the execution of some method for training a machine learning model; and one or more said vehicles 203 for collecting data for machine learning, connected to said device 301, for example, by means of a transceiver 3013 and 2031, with the use of which some said method for collecting data for machine learning can be performed, wherein the device 301 is configured to use at least the data obtained by such a data collection method for training a machine learning model.

[0078] In another preferred embodiment of the present invention, a system 400 for determining the permissible speed of a vehicle 402 is provided, comprising at least: one or more devices 401 for determining the permissible speed of a vehicle, each of which comprises at least: one or more processors 4011 of the device 401, a memory 4012 containing a program code that, when executed by the processor, causes the processor 4011 to perform some method for determining the permissible speed of the vehicle 402; and one or more vehicles 402, connected, for example, by means of transceivers 4013, 4021 with some mentioned device 401 for determining the permissible speed of the vehicle and equipped with at least one situation sensor 4022, configured to receive some mentioned situation frame.

[0079] In another preferred embodiment, a vehicle control system 4023 is provided, comprising at least: an engine 40231 configured to drive at least one propeller 40232; and at least one vehicle control device 40233, each of which comprises at least: one or more processors 402331 of the control device 40233, a memory 402332 containing a program code that, when executed by the processor of the control device 402331, causes the processor to perform some method of controlling the vehicle 402; configured at least with the possibility of generating a control signal for the engine and / or configured at least with the possibility of generating an information signal for the vehicle information system 40234;wherein said control signal is generated using a setting that includes the permissible speed of movement of the vehicle for said frame of the situation; wherein said information signal is generated using said setting.

[0080] In this case, preferably, without being limited, a device 101, 201 for collecting data for machine learning is provided, comprising at least: one or more processors 1011, 2011; a memory 1012, 2012, containing a program code, which, when executed by the processor 1011, 2011, causes the processor 1011, 2011 to perform in any order at least the following processes: receiving raw data from at least one environment sensor 2032, recording at least a portion of the received raw data on a machine-readable storage medium, receiving a control signal with subsequent execution of at least an action on at least a portion of the collected raw data. In this case, without being limited, the environment sensor 2032 can be an external device connected to the device 101, 201 for collecting data via a communication line.In this case, without limitation, the environment sensor 2032 may be an environment sensor 2032 of the vehicle 203 for collecting data or may be attached to the vehicle 203 for collecting data. In this case, without limitation, the sensor. situation 2032 may be part of any device 101, 201, and, accordingly, in such a case, the vehicle 203 for collecting data may be equipped with a corresponding device 101, 201. In this case, without being limited to, the mentioned machine-readable medium on which the raw data are recorded may be either the memory 1012, 2012 of the data collection device or an external data storage device, for example, without being limited to, a database 600 connected to the data collection device 101, 201 via a data transmission interface or a wireless communication line 500. In this case, without being limited to, the control signal may be generated either automatically after receiving any recognition signal after processing the corresponding frame of the situation, or forcibly by means of a signaling device (not shown in the drawing).In this case, without limitation, the signaling device may not be part of said data collection device and may be connected to said data collection device via a communication line, for example, via wireless radio communication. In this case, preferably, without limitation, the signaling device is provided, comprising a switching means, the actuation of which generates a control signal for the data collection device, which can thus serve as an input device.

[0081] Thus, without limitation, a method for training a machine learning model implemented using the system 300 can be provided, in which, using at least one processor 3011 of at least one computer device 301, a machine learning model is trained to determine the permissible speed of movement for a vehicle 402 depending on a frame of the situation fed to the input of the machine learning model; wherein the frame of the situation is obtained by means of a vehicle 203 for collecting data, equipped with at least one situation sensor 2032, configured to receive a frame of the situation; wherein the training of the machine learning model is ensured using data obtained by any of the mentioned data collection methods.Furthermore, without being limited, a method for determining the permissible speed of a vehicle 402, implemented using the system 400, can be provided, in which, using at least one processor 4011 of at least one computer device 401, any said frame of the situation is fed to the input of a machine learning model trained using any said method of training the machine learning model, and the value of the permissible speed of the vehicle for this frame of the situation is obtained at the output. Furthermore, without being limited, a method for implementing using can be provided in this manner. system 4023 a method for controlling a vehicle 402, in which at least one processor 402331 of at least one computer device 40233 at least: identifies at least one frame of the situation; using the identified frame of the situation, by means of any said method for determining the permissible speed of movement of the vehicle, determines the permissible speed of movement of the vehicle for this frame of the situation; and at least generates a control signal for the vehicle control system 4023, including at least a setting stored in the memory of the computer device 40233 of the vehicle control system 4023, including said determined permissible speed of movement of the vehicle for said frame of the situation.

[0082] Thus, preferably, without being limited, as shown in Fig. 1, a computer device 101, 201, 301, 401, 40233 can be provided, in which, in the context of the present invention, at least one of or any combination of: a computer device for collecting data, a computer device for training a machine learning model, a computer device for determining the permissible speed of a vehicle, a computer device for controlling a vehicle can be implemented. Such a computer device most typically comprises at least: one or more processors; a memory containing the corresponding program code, as shown earlier.At the same time, such a computing device can perform the role of a corresponding server of the corresponding system, which thus also contains at least one or more processors and memory, which are thus essentially similar, respectively, to the processor and memory of the previously described computing devices.By way of example, but not limitation, memory (machine-readable storage medium) may include non-volatile memory (NVRAM); random access memory (RAM); read-only memory (ROM); electrically erasable programmable read-only memory (EEPROM); flash memory or other memory technologies; CDROM, digital versatile disc (DVD) or other optical or holographic storage media; magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices; and any other storage medium that can be used to store and encode the desired information. In this case, memory includes, without limitation, a storage medium based on a computer storage device in the form of volatile or non-volatile memory or a combination thereof. In this case, not. Exemplary hardware devices include, but are not limited to, solid-state memory, hard disk drives, optical disk drives, and the like. Furthermore, without limitation, the machine-readable storage medium (memory) is non-transient (persistent, non-transitive), so it does not include a transient (transitive) propagating signal. Furthermore, without limitation, the memory may store an exemplary environment in which, using computer commands or codes, including those stored in the server's memory, any previously described computer procedure executable by the processor of the computing device may be performed. Furthermore, without limitation, the computing device, when not a thin client, comprises one or more processors that are designed to execute computer commands or codes stored in the device's memory in order to ensure the execution of the aforementioned procedures.In this case, without limitation, the server may be essentially similar to a computer device, when it is not a thin client, and, accordingly, contain one or more processors that are designed to execute computer commands or codes stored in the server's memory for the purpose of ensuring the execution of the aforementioned procedures. In this case, without limitation, any described system may also include a database (DB) 600. Such a DB may be, but is not limited to: a hierarchical DB, a network DB, a relational DB, an object DB, an object-oriented DB, an object-relational DB, a spatial DB, a combination of two or more of the aforementioned DBs, and the like.Moreover, without limitation, the database at least stores data, parameters, raw data, labeled data, modified data, machine learning models, and other information in memory or in a suitable memory of another computing device connected to any of the aforementioned computing devices and / or to a server, which may be, but is not limited to, memory similar to any memory as previously shown, and which can be accessed via the server. Furthermore, without limitation, a server is provided that, in addition to the previously described functions, stores and facilitates the manipulation of computer commands or codes previously described in this document, which, accordingly, are not further described. Moreover, without limitation, the server, in addition to the previously described functions, can ensure the regulation of data exchange within the system.In this case, without limitation, the exchange of data within the corresponding system is carried out via one or more data transmission networks 500. In this case, without limitation, the data transmission networks 500 may include, but are not limited to, one or more local area networks (LAN) and / or wide area networks (WAN), or may represent information- a telecommunications network, such as the Internet, an Intranet, a virtual private network (VPN), or a combination thereof, and the like. The server may also, without limitation, provide a virtual computing environment to facilitate interaction between system components. Network 500, without limitation, serves to facilitate interaction between a computing device, a server, optionally a database, and optionally the previously described systems and other systems. A non-thin client computing device and / or server may be directly connected to the database using wired and wireless communication methods and techniques known in the art, which are not further described in detail, or, without limitation, the database may be implemented in the memory of a computing device, including one that serves as a server.Moreover, without limitation, a suitable non-thin client computing device can act as a system server for other thin client computing devices. Most typically, without limitation, the components of the computing devices described herein, including the server components, are interconnected, including via a data bus.

[0083] In this case, without limitation, the said vehicle 203, 402 may be rented for a period of time, and the possibility of using such a vehicle may be provided by means of some known system for providing the possibility of independent use of vehicles on demand.

[0084] This description of the claimed technical solution demonstrates only particular embodiments and does not limit other embodiments of its implementation, since possible other alternative embodiments that do not go beyond the scope of the information set out in this application should be obvious to a specialist in this field of technology with the usual qualifications for whom the claimed technical solution is intended.

Claims

Invention formula 1. A method for collecting data for machine learning using a data collection vehicle, characterized in that, using said vehicle, a situation sensor is moved along a section of a path, while a control signal is generated for a data collection device, wherein: said vehicle is equipped with a data collection device, the control signal contains at least a command for a processor of the data collection device, causing the processor of the data collection device to execute program code, which, when executed by the processor of the data collection device, provides an effect on at least a portion of the raw data being collected; wherein said data collection device contains at least: one or more processors;a memory containing program code that, when executed by a processor, causes the processor to perform, in any order, at least the following processes: receiving raw data from at least one environment sensor, recording at least a portion of the received raw data on a machine-readable storage medium, receiving a control signal, and then performing at least an action on at least a portion of the collected raw data.

2. The method according to claim 1, characterized in that the action is at least: marking and / or changing at least part of the raw data recorded on the machine-readable medium, and / or deleting at least part of the raw and / or machine learning data recorded on the machine-readable medium, and / or not recording at least part of the collected raw data on the machine-readable medium.

3. The method according to claim 2, characterized in that the data collection device is connected to at least one said environment sensor or contains at least one said environment sensor.

4. The method according to any of paragraphs 1-3, characterized in that the environment sensor is selected from or formed by any combination of: radar, lidar, sonar, camera.

5. The method according to paragraph 4, characterized in that the movement is carried out at least together with the normal traffic flow and / or together with the normal pedestrian traffic flow; while during the movement, the speed is selected movement not higher than the maximum permissible speed of movement for a vehicle for the corresponding section of the route.

6. The method according to item 5, characterized in that the speed of movement is changed at least depending on the distance to the obstacle.

7. The method according to item 1, characterized in that the control signal is generated when the speed of movement does not correspond to the maximum permissible speed of movement for the vehicle on the corresponding section of the route, while the distance to the obstacle is selected such that a speed of movement corresponding to the maximum permissible speed of movement for the vehicle on the corresponding section of the route is allowed.

8. The method according to item 1, characterized in that the control signal is generated when no movement is performed and / or when the movement is performed without using a motor at a speed not exceeding 1.4 m / s.

9. The method according to claim 1, characterized in that the control signal is generated by means of a signaling device connected to the data collection device, or with which the data collection device is equipped.

Images