WO2012152476A1 - Method for detecting gestures - Google Patents

Abstract

The invention relates to a method for detecting a gesture (18) of a living creature (10), wherein a sequence of motions of the living creature (10) is detected by at least one radar device (4) using the micro-Doppler effect and a gesture (18) of the living creature (10) is detected.

Description

 Description title

 Method for recognizing gestures

The invention relates to a method and an arrangement for detecting a gesture.

State of the art

The detection of large-scale motion profiles of objects by means of radar is known. This detection relates to a time profile of a distance to an object to be detected and the magnitude and direction of the speed of the object. Furthermore, systems for detecting gestures are known, these use optical methods, for example. Using a video camera, a.

Document DE 10 2007 054 507 A1 relates to a method for recognizing a living being in which an object is detected by means of radar beams and in which radar data about the detected object are evaluated such that at least one property of the object is classified and with characteristics for at least one known creature is determined, so that it is determined what class of living things it is the object.

Disclosure of the invention

Against this background, a method and an arrangement with the features of the independent claims are presented. Further embodiments of the invention will become apparent from the dependent claims and the description. With the invention, inter alia, recognition of gestures of a person on a radar basis using time-resolved reception information, such as the micro-Doppler effect, possible. In an implementation of the present invention, a detection of motion sequences and thus motion profiles by means of electromagnetic waves, usually by using radar wave. Furthermore, the categorization of different movement profiles into individual classes, for example for movement sequences and / or gestures, as well as the execution of specific actions in the recognition of a specific sequence of movements is provided.

Within the scope of the invention, electromagnetic waves having different wavelengths can be used. This can be used to quantify the length of a distance covered by a body part of a person and which can be identified particularly well in the process of recognizing a gesture. The same applies to a speed which the body part has during the movement. Suitable frequencies used as operating frequencies are at least 2 GHz, maximum 900 GHz. Frequencies f ~ 2.4 GHz, 3 - 11 GHz, 22 - 26 GHz, 55-65 GHz, 76 - 81 GHz, 122 - 123 GHz, 244 - 245 GHz, 488 - 490 GHz and 620 - 660 GHz are usually suitable.

In one embodiment of the present invention, a coordinated combination of one or more high-frequency transmitters and one or more high-frequency receivers as possible components of an inventive arrangement using modulation and demodulation method and a corresponding signal processing is used to at least one time-resolved the following information: to acquire micro-Doppler, distance, angular, or amplitude information for one or more persons as objects in the vicinity of the inventive arrangement. Accordingly, the arrangement has at least one high-frequency transmitter and at least one high-frequency receiver.

The information thus obtained is then suitably linked to recognize specific gestures, ie specific movement patterns deliberately induced by an individual. Depending on the class of gestures to be recognized, a suitable selection and weighting of the abovementioned reception data, operating frequencies, a bandwidth and relative positioning or spatial arrangements of the at least one radio-frequency transmitter and radio-frequency receiver and the measurement duration of the modulation and the signal evaluation must be taken ,

Accordingly, the radar principle can be used within the scope of the invention for the detection of gestures and / or in general for the detection and classification of movements of a person and therefore of movement profiles on a small scale.

It is envisaged that at least one radar device and therefore of a number of spatially distributed radar devices transmit at least one electromagnetic wave, be reflected by a person and received again by the at least one radar device. The movement of the person can be detected by means of a frequency shift, phase shift and / or amplitude change between the at least one transmitted and the at least one received electromagnetic wave.

Based on data obtained from radar signals of a person to be examined, d. H. are provided by radar signals at least one body part, for example. Finger, arm, head, etc., are initially reconstructed motion sequences and / or recognized and identified by suitable algorithms as gestures.

In this context, a gesture is to be understood as a specific course of movement that differs sufficiently from its characteristics of random movements. Depending on a motion sequence detected thereby, actions and / or functions which are assigned to a specific gesture and commanded by the recognized gesture can be executed by a device to be controlled.

In the present invention, a gesture may be defined as a series of deliberately performed specific movements having a certain characteristic movement profile and reflecting the will of the person performing the movement through the course of the movement. It is intended to convey information by the gesture and / or to achieve a specific effect, if necessary a reaction to the gesture.

The algorithms used for this purpose are divided into two classes: the first class reconstructed from the received data, i. H. the time-resolved micro-

Doppler effect, the distance, the angle, the reception amplitude, etc., as well as the knowledge of the relative positioning and / or arrangement of at least one radio-frequency receiver as a radar sensor for the specific reception data causal movement and puts this information suitably, z. B. as time-resolved 3D position information.

The second class of algorithms compares the resulting motion sequence with stored sequences of motion. The stored movements correspond to the stored gestures. From the comparison, an algorithm of the second class determines which of the stored motion sequences comes closest to the reconstructed motion sequence with respect to a suitable metric. If there is not a sufficient match, no gesture will be detected, otherwise the gesture having the highest match will be returned.

The algorithms of the second class can - in order to perform the detection of gestures even more reliable - when comparing the reconstructed movement with the stored movements additional information about the current context, eg. For example, consider the temporal coincidence with another event, such as an incoming call or spatial correlation, such as making the gesture in a specific area of space.

The database and thus the amount of the stored movement patterns can be determined by the user via a training or by the reference of further prefabricated patterns, eg. For example, by downloading from a data pool, additional movement patterns can be extended.

In addition, the algorithms themselves may be constructed to increase the likelihood of proper recognition of motion through repeated interaction with the user, e.g. B. by prolonged use and / or in Form of feedback loops can be improved by the user, which is possible, for example, by adapting internal weightings in the algorithms.

The assignment of the specific action carried out by the device after detection of a specific gesture can either be predefined or else be configured by the user. Thus, for example, in a building due to a gesture in a room, the light on or off. With a mobile phone, the ringing can be turned off or the call accepted by a recognized gesture.

The use of electromagnetic waves of suitable frequency makes it possible to install corresponding detectors, for example. In a building hidden, z. B. behind ceramics, tiles, plastics o. Ä. In the bathroom, so switches, z. As light switch, or other actuators, z. B. valves for controlling water flow rates in the bathroom to operate by gestures contactless. However, such detectors can also be arranged behind the dashboard of a motor vehicle. In this case, frequencies are used, which u of the material-specific high-frequency characteristics. a. the attenuation of a medium, which is to be penetrated by the electromagnetic waves dependent.

Moreover, by using the gesture recognition set forth above, false triggering of switches can be avoided, e.g. B. often occur in simple capacitive proximity switches. For this purpose, information is usually already acquired during the approximation phase of the hand and / or the finger to the switch information with the actual switching command, z. B. touch the button, correlated. The switching process is triggered in total only when both information components, d. H. the movement profile in the approximation phase and the actual switching command, fit together.

By selecting suitable illumination ranges, as a rule by means of transmission and reception characteristics, the arrangement of the at least one radio-frequency transmitter and receiver and / or a suitable evaluation range using distance, angle and amplitude information, a spatially limited detection range can be created which the gestures is active. Usually, specific actions can be triggered remotely with different gestures. RFI D-like transponders allow additional features, typically security features, to be tested.

Gestures thus differ from the movement patterns mentioned in the publication DE 10 2007 054 507 A1, with which a recognition of subconscious, type-typical movement patterns, the z. B. are typical for humans or cats, is provided. However, such type-typical movement patterns are not gestures that are related to a will of the respective living being currently to be expressed by the one gesture.

By performing a probabilistic approach to motion sequences, gestures can be defined as qualitative motion profiles that are targeted and / or scheduled, and their random occurrence is usually very unlikely. Since gestures are carried out purposefully, it is even possible to convey information by means of a gesture. Accordingly, gestures can also be used as physical movement profiles for non-verbal communication.

In the context of the invention, the gestures as such can be clearly recognized by the arrangement according to the invention. It can be seen for the arrangement according to the invention that a Wlle is expressed by a movement. In an embodiment, probabilities are defined for movements that are planned or not. This is a correlation of special, d. H. with high probability not randomly occurring movements, provided.

Furthermore, a specific detection area and / or spatial area can be taken into account. In addition, a correlation to a context, e.g. B. an external event such as the ringing of a mobile phone in the car as an external event or the assumption of a conversation by the driver are made. To record the conversation, the driver can perform a wiping movement and thus a sufficiently random movement, approximately 5 - 15 cm in front of the center console as a specific detection area. A scale for large and small movements is set by a size of the detected person who performs movements to be analyzed and / or gestures to be identified. In small-scale movements, usually only a small percentage of a person's body is involved in the movement. Alternatively or additionally, it may be provided that the movements take place over distances which are small compared to the size of the person. An example of a movement on a small scale is, for example, a reciprocating motion of a finger from left to right, the finger sweeping at a certain distance over a trained as a high-frequency receiver sensor element and covers each movement direction each a distance of a few cm , In contrast to this, for example, a movement on a large scale is a circular pivoting of both arms of a person.

An application of the invention is possible in the field of mobile communications, while for novel operating concepts for non-contact operation of mobile phones

Gestures are recognized. However, the invention can also be used in industrial and domestic engineering to provide secure switch, possibly using RFID-like transponders. When used in a motor vehicle can also novel operating concepts, eg. B. for navigation or multimedia systems, can be realized.

In a further embodiment, it is possible to define a scale for movements more accurately. This can be assumed by percentage values for this. Possibly also a distance of a body part, which performs the movement, to at least one designed as a high-frequency receiver sensor element and thus to observe the at least one radar device. In a detection of the movement of the finger is, for example, for the mentioned mobile radio range to assume that the finger is in the vicinity of at least one radar device. In this regard, it should be noted that the finger is detected by the at least one radar device. It may be sufficient depending on the application, only a part of the body, eg. B. the finger above the mobile device, as well as to detect its relative movement to the mobile device. In other applications, on the other hand, it may be necessary to detect both the entire body and additionally the movement of individual body parts. Viewing areas of the sensor arrangement and the signal processing are to be adjusted accordingly. The arrangement according to the invention is designed to carry out all the steps of the presented method. In this case, individual steps of this method can also be carried out by individual components of the arrangement. Furthermore, functions of the arrangement or functions of individual components of the arrangement can be implemented as steps of the method. In addition, it is possible that steps of the method are implemented as functions of at least one component of the device or the entire device.

Further advantages and embodiments of the invention will become apparent from the description and the accompanying drawings.

It is understood that the features mentioned above and those still to be explained below can be used not only in the respectively indicated combination but also in other combinations or in isolation, without departing from the scope of the present invention.

Brief description of the drawings

Figure 1 shows a schematic representation of an embodiment of an inventive arrangement when carrying out an embodiment of the method according to the invention.

FIG. 2 shows a flow chart for a second embodiment of the method according to the invention.

Embodiments of the invention

The invention is schematically illustrated by means of embodiments in the drawings and will be described in detail below with reference to the drawings.

The figures are described in a coherent and comprehensive manner, like reference numerals designate like components. The embodiment of the arrangement 2 according to the invention shown schematically in FIG. 1 comprises at least one radar device 4 with at least one radio-frequency transmitter 6 for emitting electromagnetic waves as a first signal or primary signal 8. The emitted electromagnetic waves can be detected by an object in which it is located 1 is a person 10, reflected and received as a second signal or secondary signal 12 from at least one radio-frequency receiver 14 of the at least one radar device 4 as a sensor element of the arrangement 2. Only a high-frequency transmitter 6 and a high-frequency receiver 14 are shown in FIG. With the at least one radar device 14, a distance, an angle and a

Speed of at least one person 10 are determined.

The reflected electromagnetic waves received as a secondary signal 12 are processed by a recognition module 16 as a further component of the arrangement 2 according to the invention.

It is envisaged that the person 10 moves. The person performs 10 gestures 18, which are movements of the hands of the person 10. These gestures 18 are movements that take advantage of time-resolved distance, amplitude, angle, or micro-Doppler

Information, for example, a combination thereof, to be recorded. The induced by gestures 18 micro-Doppler effect causes a frequency of the electromagnetic waves and thus the secondary signal 12 is shifted. At the same time, in the case shown, the reception amplitude and possibly angle and distance values which are detected by the at least one radar device 4 change.

For the embodiment of the inventive arrangement 2 shown in FIG. 1, only one radar device 4 is shown. In further embodiments of the invention, however, an arrangement 2 according to the invention may also have a number of spatially distributed radar devices 4. These multiple spatially distributed radar devices 4, each of these radar devices 4 comprising at least one radio-frequency transmitter 6 and at least one radio-frequency receiver 14, can spatially detect a motion sequence of the person 10 for detecting a gesture 18 in combination. The secondary signal 12 is analyzed by the recognition module 16. For recognizing the gestures 18, wherein a gesture 18 is distinguished from another movement sequence, it is provided that the recognition module 16 stores the movements recorded by the person 10 with known, predefined movement sequences which are stored in a memory 20 of the arrangement 2, compares, so that an identification of the detected movement sequences on the basis of the comparison made as gestures 18 is possible. In this case, a first gesture 18 is detected by an algorithm. To execute the method, a plurality of movement sequences are predefined and stored in the memory 20. In this case, a gesture 18 is assigned to each predefined movement sequence.

Furthermore, the at least one radar unit 4 comprises a signal processing module 15 for modulation 15 of the electromagnetic waves sent as primary signals 8 and for demodulation of the electromagnetic waves received as secondary signal 12.

With the method according to the invention, at least one kinematic variable, i. H. the distance, the speed and / or the angle of the detected movement and thus of the movement are taken into account. The dynamic movement can be detected very well via the micro-Doppler effect.

As a gesture 18, a sequence of movements on a small scale is usually detected and recognized. In this case, a movement of at least one body part, here one of the hands of the person 10 is detected, wherein the at least one body part covers a distance during the movement, which is smaller than the person 10. A gesture 18 is defined as a small-scale movement. Accordingly, it is provided in the context of the invention, at least to detect the movement of a body part but also the entire person 10, so as to be able to set the movement of the body part in a scale relationship to the person 10. With the arrangement 2 can be distinguished between the moving body part and the entire person 10 depending on the application. If a distinction of the entire body of individual body parts should be required, this may, for. B. by a sufficiently high spatial Resolution of at least one designed as a high-frequency receiver 14 radar sensor of the arrangement 2 can be achieved. Typically, the movement patterns that can be generated by a moving body part, for example, by maximum speed and / or maximum acceleration, clearly differ from the patterns that the entire person 10 generates.

Usually, gestures 18 are deliberately performed movements that may serve non-verbal communication between people 10. Accordingly, with a gesture 18, a will of the person 10 is expressed. Thus, having identified the gesture 18 with the arrangement 2, it is possible to control a device 22 based on a particular gesture 18, which can be understood here as input and / or command, so that this device 22 has a specific function, which is excited by the particular gesture 18 performs.

The arrangement 2 can distinguish intended movements of unintentional movements by algorithms, whereby a recognition probability can be increased and / or a misidentification can be reduced.

The flowchart shown in FIG. 2 is intended to describe a second embodiment of the method according to the invention.

In this case, in a first step 30, at least one radar device 4 sends a primary signal in the direction of a moving person 10. Furthermore, electromagnetic waves are reflected by the moving person 10 as secondary signals and received by the at least one radar device 4 in a second step 32. Due to the movements and thus due to the movements performed by the person 10, there may be a frequency shift for the electromagnetic waves of the secondary signal. This frequency shift, which is caused by the movements of the person 10 is identified by the at least one radar unit 4 in a third step 34 as a micro-Doppler signal, wherein in addition to the micro-Doppler effect in general all of the at least one radar device 4th taken into account. In a further fourth step, the motion sequence recorded via the micro-Doppler signal is recorded with further sequences of motion, the diagrams of which are stored in a memory. are compared. It is provided that some of the stored movements are associated with gestures 18. If the motion sequence performed by the person 10 is a gesture 18, that gesture 18 may be identified in a concluding fifth step 38.

Claims

claims 1. A method for detecting gestures (18) of a person (10), in which at least one radar device (4) using the micro-Doppler effect detects a movement sequence of the person (10) and detects a gesture (18) of the person (10). 10) is detected. 2. The method of claim 1, wherein the movement on the basis of a frequency shift, phase shift or amplitude change of at least one of the at least one radar device (4) sent, from the person (10) reflected and from the at least one radar device (4) again received electromagnetic wave is detected. 3. The method of claim 1 or 2, wherein at least a kinematic size of a detected movement is taken into account. 4. The method according to any one of the preceding claims, wherein as a gesture (18) a movement sequence is detected and detected on a small scale, wherein a gesture (18) as a movement of at least one body part of the person (10) is defined, wherein the at least one Body part covers a distance that is less than the person (10). 5. The method according to any one of the preceding claims, wherein the gesture (18) is detected by at least one algorithm. 6. Method according to one of the preceding claims, in which a plurality of movement sequences are predefined, wherein a predefined movement sequence is assigned a gesture (18), wherein a detected movement sequence of the person (10) is compared with at least one predefined movement sequence, and wherein the gesture ( 18) is detected on the basis of a comparison of the movements. 7. Method according to one of the preceding claims, in which a control command is provided on the basis of a recognized gesture (18). 8. Arrangement for recognizing gestures (18) of a person (10) having a recognition module (16) which cooperates with at least one radar device (4), which, utilizing the micro-Doppler effect, initiates a movement of the person (10). detects and recognizes a gesture (18) of the person (10). 9. Arrangement according to claim 8, comprising at least one radar device (4), the at least one high-frequency transmitter (6) for transmitting electromagnetic waves, at least one high-frequency receiver (14) for receiving electromagnetic waves and a signal processing module (15) for Has modulation and demodulation of signals. 10. Arrangement according to claim 9, which has a plurality of spatially distributed radar devices (4).