Sensor

The sensor addresses interference issues and computational inefficiencies by employing a discrete Fourier transform after quantization, ensuring reliable and efficient object detection with minimal resources and fast response times.

DE202024106255U1Active Publication Date: 2026-03-19LEUZE ELECTRONIC GMBH & CO KG
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
DE202024106255
Authority / Receiving Office
DE · DE
Patent Type
Utility models
Current Assignee / Owner
Filing Date
2024-10-31
Publication Date
2026-03-19
Estimated Expiration
2034-10-31

AI Technical Summary

Technical Problem

Existing object detection sensors are susceptible to interference from pulsed light sources, leading to false detections, and require significant computational resources and hardware costs for Fourier transform-based signal evaluation, resulting in long response times and high memory requirements.

Method used

A sensor design that transforms sensor signals into the frequency domain using a discrete Fourier transform after quantization, allowing for efficient object detection with minimal hardware and computation time by utilizing a quantization unit and a Fourier transform unit, which only requires storage for current and previous samples, and performs calculations through simple additions and phase angle conversions.

Benefits of technology

The sensor achieves reliable and interference-resistant object detection with reduced computational effort, minimal hardware requirements, and fast response times, while maintaining accuracy and reducing false detections.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 00000000_0000_ABST
    Figure 00000000_0000_ABST
Patent Text Reader

Abstract

A sensor for detecting objects (10) in a monitoring area, comprising a transmitter unit emitting a sensor signal with a periodic time profile, a receiver unit configured to receive the sensor signal, and an evaluation unit (7) generating an object detection signal based on received signals from the receiver unit, characterized in that the evaluation unit (7) comprises a quantization unit (12) and a Fourier transform unit (13), wherein in the quantization unit (12) the sensor signal is converted into a quantized signal, and wherein in the Fourier transform unit (13) the quantized signal is transformed into a frequency signal in the frequency domain using a discrete Fourier transform, from which the object detection signal is generated.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] The invention relates to a sensor for detecting objects in a monitored area.

[0002] Such sensors generally comprise a transmitter unit that emits a sensor signal with a periodic time profile, a receiver unit designed to receive these sensor signals, and an evaluation unit in which an object detection signal is generated depending on the received signals from the receiver unit.

[0003] An example of such a sensor is an optical sensor, which has a transmitter as its transmitting unit and a receiver as its receiving unit. The optical sensor emits light rays as its sensor signal in the form of a periodic sequence of light pulses.

[0004] The evaluation unit typically analyzes the received signals generated upon receiving the light pulses in the time domain. Typically, the received signal is evaluated against a threshold value, generating a binary object detection signal whose signal states indicate whether an object is within the monitored area or not.

[0005] However, such signal evaluation in the time domain is susceptible to interference; in particular, interference from light sources in the form of pulsed light can lead to false detections.

[0006] To reduce such susceptibility to interference, it is known to perform signal evaluation in the frequency domain. For this purpose, the signal sequence of sensor signals registered at the receiver is transformed into the frequency domain by means of a Fourier transform using a Fast Fourier Transform (FFT), whereby the signal evaluation for generating the object detection signal then takes place in the frequency domain.

[0007] A disadvantage of this approach is that the entire time-dependent sensor signal must be available for performing the Fourier transform or FFT, leading to undesirably long response times for generating the object detection signal. Furthermore, the Fourier transform or FFT requires significant computational time, resulting in correspondingly high hardware and software costs for the evaluation unit. The memory required to store the signal waveform is a particular disadvantage.

[0008] Generally, the use of a Goertzel algorithm or a Goertzel filter is also possible. A disadvantage here is that this requires multiplication circuits, which occupy large areas of semiconductor switching elements.

[0009] The invention is based on the objective of providing a sensor of the type mentioned above in which reliable and interference-resistant object detection is ensured with minimal design effort.

[0010] The features of claim 1 are provided to solve this problem. Advantageous embodiments and expedient further developments of the invention are described in the dependent claims.

[0011] The invention relates to a sensor for detecting objects in a monitored area. The sensor comprises a transmitter unit that emits a sensor signal with a periodic time profile, a receiver unit for receiving the sensor signal, and an evaluation unit in which an object detection signal is generated depending on the received signals from the receiver unit. The evaluation unit comprises a quantization unit and a Fourier transform unit, wherein the sensor signal is converted into a quantized signal in the quantization unit. In the Fourier transform unit, the quantized signal is transformed into a frequency signal in the frequency domain using a discrete Fourier transform, from which the object detection signal is generated.

[0012] The sensor according to the invention includes a transmitter unit that emits a sensor signal for object detection which has a periodic time profile.

[0013] The sensor can be designed as an ultrasonic sensor or a radar sensor.

[0014] This is particularly advantageous when designed as an optical sensor with a transmitter unit comprising a transmitter and transmitting optics, and with a receiver unit comprising a receiver and receiving optics.

[0015] The transmitter emits light beams in the form of a periodic sequence of light pulses. The transmitter emits sequences of rectangular light pulses particularly advantageously.

[0016] The transmitter can consist of at least one light-emitting diode, one laser diode or one VCSEL (Vertical Cavity Surface Emitting Laser).

[0017] Accordingly, the receiver can be formed by at least one PIN diode, one APD (Avalanche Photodiode) or one SPAD (Single Photon Avalanche Photodiode).

[0018] According to the invention, the temporal sequence of sensor signals is transformed into the frequency domain, whereby the object is determined based on the generated frequency signals. According to the invention, the evaluation unit of the sensor has a quantization unit and a Fourier transform unit for this purpose.

[0019] In the quantization unit, the sensor signal is sampled in predetermined time steps, and a quantized signal is obtained at each time step.

[0020] To accurately represent the sensor signal with the generated quantized signal, the sampling must be periodic, ensuring that the sensor signal is sampled at least once in each half-period. For a square wave signal, this means it must be sampled at least once each time it reaches a minimum or maximum value. More frequent sampling is possible. However, a balance must be struck between the signal's detection reliability (which increases with more frequent sampling) and the latency until the switching output changes or the switching frequency (which deteriorates with more frequent sampling).

[0021] In the simplest case, the sensor signal is quantized in two quantization stages, so that the quantized signal takes on the signal values ​​+1 or -1.

[0022] The quantized signal is particularly advantageous if it has three quantization levels with signal values ​​of -1, 0, and +1. Generally, multiple quantization levels are also possible.

[0023] The quantized signal is used in the Fourier transform unit to perform a discrete Fourier transform.

[0024] In general, the quantization levels should be chosen so that each level corresponds to a power of two, so that the multiplication required for the discrete Fourier transform can be done by a left-shift operation.

[0025] The quantized signal is advantageously optimized in such a way that it can be easily transformed into the frequency domain using the discrete Fourier transform with little computation time.

[0026] Advantageously, a 1.5-bit discrete Fourier transform is performed in the Fourier transform unit.

[0027] The quantized signal then has the signal values ​​1, 0, -1.

[0028] In general, it is also possible for a multi-bit discrete Fourier transform to be performed in the Fourier transform unit.

[0029] The discrete Fourier transform performed in the Fourier transform unit according to the invention differs from known Fourier transforms, especially FFTs, in that the entire time course of the quantized signal does not need to be available. Rather, at the time interval of the sampling of the sensor signal in the quantization unit, only the current quantized signal and the intermediate result from the immediately preceding clock cycle are used to transform this quantized signal into the frequency domain.

[0030] In the Fourier transform unit, the quantized signals x are converted into k in a manner known per se a frequency signal F m (x) according to the relationship Fm(x)=∑k=0Nxk∗e2πikmN obtained, where k is the respective time step and m is a frequency bin. The frequency signal F m (x) is calculated for each time step of the sensor signal sampling and depends only on the current quantized signal x k of the respective time step and the sum of the values ​​of the previous time steps.

[0031] Specifically, the frequency signal F calculated for the first time step depends m (x) only from the quantized signal x k of the first time step.

[0032] The mean frequency is f bin of a frequency bin m by the relationship fbin=fabatast∗mN defined where f abtast The sampling frequency is the frequency at which the sensor signal is sampled in the quantization unit.

[0033] A particular advantage here is that no additional storage is required besides the storage spaces for the current sample value and the previous samples.

[0034] The resulting discrete Fourier transform can be performed quickly and with minimal hardware and computing time. The evaluation unit can therefore have a simple design.

[0035] In particular, the evaluation unit features an ASIC in which the quantization unit and Fourier transform unit are implemented. Alternatively, an FPGA could be used.

[0036] The discrete Fourier transform is performed using the fact that in the equation for the frequency signal, the exponent of the exponential function is used as the phase angle. φk=2πmN∗k

[0037] The system is integrable, where the phase angle φ is used to calculate a frequency signal for the k-th time step. kaccording to the relationship φk=φk−1+φ1 is calculated where φ1 is a base angle according to the relationship φ1=2πmN is.

[0038] The phase angle φ k Thus, it increases by the constant base angle φ1 in each time step. In a digital circuit, a calculation of φ is therefore possible. k can be performed by simple additions, i.e., no time-consuming multiplications or divisions are required, so that the discrete Fourier transform can be performed with minimal computational effort.

[0039] After calculating the phase angles φ k Polar coordinates are converted into Cartesian coordinates.

[0040] This involves converting phase angles into Cartesian coordinates using a table of values.

[0041] After the calculations are completed, a reverse transformation into amplitude and phase values ​​is performed after the phase angles have been calculated using a table of values.

[0042] The necessary calculations for the phase angles can be performed in Cartesian coordinates with less computation time. The conversions in the tables of values ​​also require minimal computation time.

[0043] Instead of tables of values, a CORDIC algorithm can also be used for the conversions.

[0044] The object detection signal is generated by processing the frequency signal obtained with the discrete Fourier transform.

[0045] Advantageously, the evaluation unit has a threshold unit in which the object detection signal is generated by a threshold evaluation of the frequency signal, whereby the amplitude and phase of the received signal are determined beforehand using a look-up table.

[0046] In this context, a threshold value assessment with one or more threshold values ​​is possible within the threshold unit.

[0047] The advantage is that the threshold, or each threshold, is configurable.

[0048] In particular, in the case of a sensor in which the object detection signal is generated by means of a threshold unit, the sensor is designed as a binary switching sensor in which the object detection signal has two signal states, wherein a first signal state indicates object detection and a second signal state indicates an object-free signal state.

[0049] The object detection signal can be output via a digital switching output.

[0050] Alternatively or additionally, the sensor has a communication interface through which the object detection signal and / or other information can be output.

[0051] The communication interface can be a radio interface, an IO-Link interface, or a single-pair Ethernet interface.

[0052] Especially with more complex sensors such as light grids, this can have a multi-channel structure with regard to the design of the evaluation unit and / or the communication interface.

[0053] The invention will be explained below with reference to the drawings. The drawings show: Fig. 1: Exemplary embodiment of the sensor according to the invention in the form of an optical sensor. Fig. 2: Block diagram of the evaluation unit of the optical sensor according to Fig. 1. Fig. 3: Example of a circuit-based implementation of the evaluation unit according to Fig. 2.

[0054] Fig. Figure 1 shows an embodiment of the sensor according to the invention in the form of an optical sensor 1. The sensor generally serves for object detection in a monitoring area. Such optical sensors 1 can be designed as light barriers or reflective light barriers, in particular in the form of binary switching sensors.

[0055] In the present case, the optical sensor 1 is designed as a photoelectric sensor. This optical sensor 1 comprises a transmitter 3 emitting light beams 2 with associated transmitting optics 4 and a receiver 5 with integrated receiving optics 6. The transmitter 3 can be formed by at least one light-emitting diode, a laser diode, or a VCSEL (Vertical Cavity Surface Emitting Laser).

[0056] The receiver 5 can be formed by at least one PIN diode, one APD (Avalanche Photodiode) or one SAPD (Single Photon Avalanche Photodiode).

[0057] Furthermore, the optical sensor 1 has an evaluation unit 7. For example, the evaluation unit 7 has an ASIC or FPGA in which the quantization unit 12 and Fourier transform unit 13 are implemented.

[0058] The components of the optical sensor 1 are integrated in a housing 8, wherein a disc 9 transparent to the light rays 2 is integrated in a wall of the housing 8.

[0059] The light beams 2 emitted by transmitter 3 are guided through the disc 9 into the monitoring area. If an object 10 is located in the monitoring area, the light beams 2 are reflected by it and guided back through the disc 9 to receiver 5. The received signals generated in this process are evaluated to generate an object detection signal, which is output by the sensor.

[0060] Since the light switch is also a binary switching sensor, a binary object detection signal is generated in it, the switching states of which indicate whether an object 10 is in the monitoring area or not.

[0061] The transmitter unit of the sensor according to the invention generally emits a sensor signal with a periodic profile. In the present case, the transmitter 3 of the optical sensor 1 emits light beams 2 in the form of a periodic sequence of light pulses, which are advantageously designed as rectangular pulses.

[0062] Fig. Figure 2 shows a block diagram of the evaluation unit 7 of the optical sensor 1 according to Fig. 1.

[0063] The received signal from receiver 5, i.e., the sensor signal, is fed as input signal 11 to a quantization unit 12. In the quantization unit 12, the sensor signal is sampled at predetermined time steps, and a quantized signal is obtained at each time step.

[0064] In the present case, a quantized signal with the bit values ​​-1,0,1 is generated in the quantization unit 12.

[0065] The quantized signal thus generated is fed into one of the Fourier transform units 13 downstream of the quantization unit 12. There, the quantized signal is transformed into the frequency domain by means of a discrete Fourier transform.

[0066] Adapted to the quantized signal generated in the quantization unit 12, a 1.5-bit discrete Fourier transform is performed in the Fourier transform unit 13.

[0067] This involves a Discrete Fourier Transform (DFT) performed on samples with a bit width of 1.5 bits each. Due to the reduced bit width (compared to more common bit widths of 8, 12, 16, 24, 32, etc.), simplifications result, allowing for an implementation optimized in terms of digital logic resource requirements and computational complexity.

[0068] Using the discrete Fourier transform, a Fourier-transformed signal is obtained from the quantized signal, hereinafter referred to as the frequency signal.

[0069] To perform the discrete Fourier transform, the quantized signals x are converted in the Fourier transform unit 13. k a frequency signal F m(x) according to the relationship Fm(x)=∑k=0Nxk∗e2πikmN obtained, where k is the respective time step and m is a frequency bin, where the frequency signal F m (x) is calculated for the individual time steps of the sensor signal sampling and is only dependent on the current quantized signal x k of the respective time step and the frequency signal of the previous time step.

[0070] Specifically, the frequency signal F calculated for the first time step depends m (x) only from the quantized signal x k of the first time step. The average frequency is f. bin of a frequency bin m by the relationship fbin=fabatast∗mN defined where f abtast The sampling frequency is the frequency at which the sensor signal is sampled in the quantization unit 12.

[0071] In an amplitude determination unit 14 downstream of the Fourier transform unit 13, an amplitude determination is carried out for the frequency signal, using a look-up table 29 for this purpose.

[0072] Alternatively, an approximation of the Euclidean norm of a complex number in Cartesian form can be used according to the following relationship: A=‖a+ib‖2≈|max(a,b)|−12|min(a,b)|

[0073] The determined amplitude value of the frequency signal is fed to a threshold unit 15, in which the amplitude value is evaluated against a threshold value, thereby generating the binary object detection signal. This object detection signal is output via an output 16 as output signal 17.

[0074] Output 16 can be a digital switching output. Alternatively, it can be part of a communication interface.

[0075] In particular, the communication interface is a radio interface, an IO-Link interface, or a single-pair Ethernet interface.

[0076] Fig. Figure 3 shows the implementation of the evaluation unit 7 according to the invention. Fig. 1 in the form of a digital circuit.

[0077] This digital circuit implements the formula for the discrete Fourier transform. In particular, it takes into account that in the equation for the frequency signal, the exponent of the exponential function has a phase angle. φk=2πmN∗k includes, where the phase angle φ is used to calculate a frequency signal for the k-th time step. k according to the relationship φk=φk−1=φ1 is calculated where φ1 is a base angle according to the relationship φ1=2πmN is.

[0078] The calculation of the phase angle and its tracking is performed by the input section of the digital circuit.

[0079] The quantized signals obtained in the quantization unit 12 are fed in via line 18 at the time interval of the sampling carried out in the quantization unit 12.

[0080] Line 19 is used for angle reset.

[0081] In the digital module 20, to which a clock signal CLK is fed back, the base angle φ1 is specified.

[0082] A reset angle is specified via line 21.

[0083] The base angle is fed to a multiplexer 24 via a line 22 and a shift-left operator 23. A time base line 24a increases the base angle by a specific angle corresponding to the advance of time by a specific time period.

[0084] The output of multiplexer 24 is routed via a summing element 25 to another multiplexer 26, to whose output an intermediate storage device 27 is connected. The output of the intermediate storage device 27 is fed back to the summing element 25 via a line 28.

[0085] With these switching elements, the current phase angle is calculated in the quantization unit 12 at the time interval of the sampling according to equations (3)-(5).

[0086] In a look-up table 29, the phase angles are converted into Cartesian coordinates, i.e., a real and imaginary part.

[0087] The sum of the discrete Fourier transform is performed in a downstream two-channel circuit arrangement.

[0088] Each channel has a bit shift unit 30a, 30b, a summing element 31a, 31b and a storage element 32a, 32b.

[0089] The first bit shift unit 30a receives the real part of the phase angle and the quantized signal.

[0090] The second bit shift unit 30b receives the imaginary part of the phase angle and the quantized signal.

[0091] In the summing elements 31a, 31b, the values ​​of the bit shift units 30a, 30b are summed and stored in the memory elements 32a, 32b. The sums present there are converted from Cartesian coordinates into amplitude / phase values ​​in a subsequent look-up table 33, thereby obtaining the frequency signal according to equation (1) as the result of the discrete Fourier transform.

[0092] The object detection signal is generated in the signal processing unit, which has a threshold unit 15. The object detection signal is output as output signal 17 via an output 16 of the optical sensor 1. Reference symbol list 1 Optical sensor 2 light beam 3 channels 4 transmitting optics 5 receivers 6 Receiving optics 7 Evaluation unit 8 cases 9 disc 10 objects 11 Input signal 12 quantization units 13 Fourier transform unit 14 Amplitude determination unit 15 threshold unit 16 Exit 17 Output signal 18 Management 19 Management 20 digital modules 21 Management 21a Line 22 Management 23 Shift-Left Operator 24 multiplexers 24a Line 25 summing term 26 multiplexers 27 buffers 28 Management 29 Look-up table 30a, b Bit shift unit 31a, b summing term 32a, b Storage element 33 Look-up table

Claims

[1] Sensor for detecting objects (10) in a monitoring area, comprising a transmitter unit emitting a sensor signal with a periodic time profile, a receiver unit designed to receive the sensor signal, and an evaluation unit (7) in which an object detection signal is generated depending on received signals from the receiver unit. characterized by , that the evaluation unit (7) comprises a quantization unit (12) and a Fourier transform unit (13), wherein in the quantization unit (12) the sensor signal is converted into a quantized signal, and wherein in the Fourier transform unit (13) the quantized signal is transformed into a frequency signal in the frequency domain using a discrete Fourier transform, from which the object detection signal is generated. [2] Sensor according to claim 1, characterized by, that the evaluation unit (7) has a threshold unit (15) in which the object detection signal is generated by a threshold evaluation of the frequency signal. [3] Sensor according to claim 2, characterized by , that in the threshold unit (15) the frequency signal is evaluated with one or more threshold values. [4] Sensor according to claim 3, characterized by that the threshold or each threshold is configurable. [5] Sensor according to any one of claims 1 to 4, characterized by that this is a binary switching sensor. [6] Sensor according to any one of claims 1 to 5, characterized by that this is an optical sensor (1), a radar sensor or an ultrasonic sensor. [7] Sensor according to claim 6, characterized by , that this is designed as an optical sensor (1) with a transmitter unit with a transmitter (3) and a transmitting optics (4) and with a receiver unit with a receiver (5) and a receiving optics (6). [8] Sensor according to claim 7, characterized by , that the transmitter (3) is formed by at least one light-emitting diode, one laser diode or one VCSEL (Vertical Cavity Surface Emitting Laser). [9] Sensor according to one of claims 7 or 8, characterized by , that the receiver (5) is formed by at least one PIN diode, one APD (avalanche photodiode) or one SPAD (single photon avalanche photodiode). [10] Sensor according to any one of claims 1 to 9, characterized by that it has a communication interface. [11] Sensor according to claim 10, characterized by that the communication interface is a radio interface, an IO-Link interface, or a single-pair Ethernet interface. [12] Sensor according to any one of claims 1 to 11, characterized by that it has a multi-channel structure. [13] Sensor according to any one of claims 1 to 12, characterized by, that the evaluation unit (7) has an ASIC or FPGA in which the quantization unit (12) and Fourier transform unit (13) is implemented. [14] Sensor according to any one of claims 1 to 13, characterized by , that in the quantization unit (12) a quantized signal with the bit values ​​-1,0,1 is generated. [15] Sensor according to claim 14, characterized by , that a 1.5-bit discrete Fourier transform is performed in the Fourier transform unit (13). [16] Sensor according to any one of claims 1 to 14, characterized by , that a multi-bit discrete Fourier transform is performed in the Fourier transform unit (13). [17] Sensor according to any one of claims 1 to 16, characterized by , that in the quantization unit (12) the sensor signal is sampled in predetermined time steps and a quantized signal is obtained at each time step. [18] Sensor according to claim 17, characterized by, that in the Fourier transform unit (13) the quantized signals x k a frequency signal F m (x) according to the relationship Fm(x)=∑k=0Nxk∗e2πikmN is obtained, where k is the respective time step and m is a frequency bin, where the frequency signal F m (x) is calculated for the individual time steps of the sampling of the sensor signal and is only dependent on the current quantized signal x k of the respective time step and all preceding x k depends on earlier time steps. [19] Sensor according to claim 17, characterized by , that the frequency signal F calculated for the first time step m (x) only from the quantized signal x k depends on the first time step. [20] Sensor according to one of claims 18 or 19, characterized by , that the mean frequency f bin of a frequency bin m by the relationship fbin=fabatast∗mN is defined where f abtastthe sampling frequency is the frequency at which the sensor signal is sampled in the quantization unit (12). [21] Sensor according to any one of claims 18 to 20, characterized by , that in the equation for the frequency signal the exponent of the exponential function has a phase angle φk=2πmN∗k includes, where the phase angle φ is used to calculate a frequency signal for the k-th time step. k according to the relationship φk=φk−1+φ1 is calculated where φ1 is a base angle according to the relation φ1=2πmN is. [22] Sensor according to claim 21, characterized by , that the phase angles φ k calculated in polar coordinates and then converted to Cartesian coordinates. [23] Sensor according to claim 22, characterized by that the conversion of phase angles into Cartesian coordinates is done using a table of values. [24] Sensor according to one of claims 22 or 23, characterized by, that after calculating phase angles using a table of values, a reverse transformation into amplitude and phase values ​​is performed and the sum of the discrete Fourier transformation is carried out.

Citation Information

Patent Citations

  • device and method for distance measurement

    DE102016217838A1

  • Method and a device for determining a distance to a target

    US20240036200A1

  • Time of flight detection systems with efficient phase measurement

    US20240230902A9