Alert system
Patent Information
- Authority / Receiving Office
- EP · EP
- Patent Type
- Applications
- Current Assignee / Owner
- SONAIR AS
- Filing Date
- 2025-04-28
- Publication Date
- 2026-04-22
Smart Images

Figure EP2025061541_06112025_PF_FP_ABST
Abstract
Description
[0001] Alert System
[0002] TECHNICAL FIELD
[0003] This disclosure relates generally to an alert system, and more particularly to an alert system to be carried by a device and configured to signal an alert if an object is present within a hazard zone proximate to the device.
[0004] BACKGROUND OF THE INVENTION
[0005] It is known to use alert systems to detect people or objects in a predetermined area (e.g. in a vicinity of an alert system) and to provide an output in the event that a person or object is detected. For example, it is well-known for cars to include a parking sensor system to detect the presence of obstacles in proximity of the car while it is being parked, and to alert the driver of the vehicle if an object is detected.
[0006] However, existing alert systems such as parking sensors are relatively crude and can, on occasions, incorrectly signal an alert when an object is not a hazard and / or fail to detect an object that is a hazard.
[0007] An improved alert system that addresses one or more of these issues is therefore desirable.
[0008] SUMMARY OF THE INVENTION
[0009] According to a first aspect of the invention, there is provided an alert system configured to be carried by a device and configured to signal an alert if an object is present within a hazard zone proximate to the device, wherein the alert system comprises: an array of acoustic transducers; and a processing system, wherein the alert system is configured: to emit an acoustic signal; to receive, at the array of acoustic transducers, reflections of the acoustic signal from one or more surfaces in a vicinity of the alert system; to sample electrical signals output by the array of transducers in response to the received reflections; to process the sampled electrical signals to determine a respective modelled position for each of the one or more surfaces; for each of the one or more modelled positions, to model a respective acoustic signal emanating from the modelled position; to determine if the one or more modelled acoustic signals are consistent with the sampled electrical signals output by the array of transducers in response to the received reflections, and, if not, to signal an inconsistency; and to determine if any of the one or more modelled positions is within the hazard zone, and, if so, to signal an alert.
[0010] According to a second aspect of the invention, there is provided a method of operating an alert system, the method comprising: emitting an acoustic signal from the alert system; receiving reflections of the acoustic signal from one or more surfaces in a vicinity of the alert system; sampling electrical signals in response to the received reflections; processing the sampled electrical signals to determine a respective modelled position for each of the one or more surfaces; for each of the one or more modelled positions, modelling a respective acoustic signal emanating from the modelled position; determining if the one or more modelled acoustic signals are consistent with the sampled electrical signals output by the array of transducers in response to the received reflections, and, if not, signalling an inconsistency; and determining if any of the one or more modelled positions is within a hazard zone, and, if so, signalling an alert.
[0011] According to a third aspect of the invention, there is provided a device comprising the alert system of the first aspect.
[0012] Thus it will be seen that, in accordance with embodiments of the invention, an alert system uses modelled positions to determine whether a surface of an object is present in a hazard zone, which may be proximate to the alert system and / or to a device carrying the alert system, and signals an alert if so. The alert system also determines whether acoustic signals modelled by the alert system are consistent with reflections of an emitted acoustic signal received at the alert system, and signals an inconsistency if not. Thus, the alert system is able to signal both in the case that an object is present in the hazard zone (e.g. indicating that an object is proximate the device carrying the alert system), and in the case that acoustic signals modelled by the system do not sufficiently explain the acoustic signals actually received by the system.
[0013] The alert system is therefore able in some embodiments to signal an inconsistency if it is likely that there is an object in range of the alert system that is not adequately accounted for by the model, i.e. if the presence of an object in a hazard zone proximate the device cannot be ruled out. This may allow the alert system to signal that the device should switch to a safer mode of operation, e.g. by ceasing movement, if there is a possibility that an object is present in the hazard zone proximate the device, even if it is not positively identified as such by the alert system. This may result in safer operation of the device.
[0014] In some embodiments, the alert system may be operable in both an active state, in which alerts and / or inconsistencies are signalled, and an inactive state (e.g. a service state) in which alerts and / or inconsistencies are not signalled. In this way it may be possible to prevent or mute alert signals or inconsistency signals in certain circumstances, e.g. during servicing of the alert system and / or device. The alert system may be configured to switch between the active state and the inactive state in response to an input signal received by the device and / or to the alert system (e.g by a physical button or switch operated by a user).
[0015] The device may be a mobile device; for example the device may be a passenger vehicle, such as a self-driving car, or a moving robot, preferably an autonomous moving robot. In some embodiments, however, the device may be a static device, such as a robot in a fixed cell or work area, e.g. a robot having a fixed base. The device may be a static device that has a component that is moveable, e.g. using an actuator, such as a robot arm. The array of acoustic transducers may be mounted to or part of the moveable component. The hazard zone may be proximate to the moveable component and may move with the moveable component.
[0016] It will be appreciated that the alert system may determine that every modelled position is outside of the hazard zone, and not signal an alert in this case. The alert system is therefore able to model acoustic signals for surfaces at ranges beyond the hazard zone as well as at ranges within the hazard zone. Thus, the system may determine the position of a plurality of surfaces in the environment (e.g. within a detection range that is larger than the hazard zone), but only provide an output in the case of an inconsistency between the modelled acoustic signals and the sampled electrical signals, or in the case that one or more of the modelled positions is within the hazard zone.
[0017] The hazard zone proximate the device may be a three-dimensional volume proximate to the device, e.g. in front of the device, behind the device or around the device. It may be adjacent the device, e.g. abutting a surface of the device, although this is not essential. The three-dimensional volume may substantially surround the device in some embodiments. The three-dimensional volume may extend from the device in a direction of movement of the device in some embodiments. The three-dimensional volume may be defined by a set of boundary surfaces. Each of the boundary surfaces may be defined by a respective set of three-dimensional coordinates, which may be centred on the device or the alert system.
[0018] The hazard zone may be static relative to the alert system or the device, or to a surface thereof, such that the hazard zone moves with the alert system and / or the device, or at least with a surface thereof. The hazard zone may be defined by one or more parameters. The one or more parameters may comprise a plurality of distances and directions from the device and / or the alert system, e.g. a distance and direction to each of a plurality of points defining the boundary surfaces. The one or more parameters may be stored within a memory of the device and / or a memory of the alert system.
[0019] In some embodiments the hazard zone may be configurable, such that the shape or volume of the hazard zone is adjusted based on one or more variables. The one or more variables may be determined by a processor of the device and / or the alert system. The one or more variables may comprise a velocity or an acceleration of the device. For example, when a velocity of the device is increased the spatial extent of the hazard zone may also be increased. The one or more variables may comprise a direction or trajectory of the device. The one or more variables may comprise a location of the device. For example, if the device is operating in a first zone of a building where humans are likely to be present, the size of the hazard zone may be increased, while if the device is operating in a second zone of a building where humans are unlikely to be present, the size of the hazard zone may be decreased. Configuring the hazard zone may comprise increasing or decreasing the size, e.g. spatial extent, of the hazard zone in some embodiments. The spatial extent of the hazard zone may be increased or decreased uniformly in all directions, or the change in extent may vary in different directions from the device. For example, the size of the hazard zone may be increased to a greater extent in a direction in which the device is moving. Configuring the hazard zone based on a velocity or trajectory of the device may thus allow the alert system to detect the presence of objects proximate the device further in advance of a potential collision (i.e. earlier) when the device is moving at high speed.
[0020] An alert or an inconsistency may be signalled by an electronic signal and / or an audible alert and / or a visual alert. The signal may, for example, be an OSSD (output signal switching device) signal. The alert system may be configured to provide an output signal to the device (e.g. as an electrical or optical signal) when signalling an alert and / or an inconsistency in some embodiments. The output signal to the device may provide the alert or may signal the inconsistency, or it may be additional to the signalling of the alert or inconsistency. The alert system may be configured to provide an output signal to the device instructing the device to perform a safety function when signalling an alert and / or an inconsistency. In embodiments in which the device is a mobile device, the safety function may comprise the device altering a course of movement (e.g. changing direction). Alternatively, the safety function may comprise the device arresting motion (e.g. slowing or stopping). This may reduce the likelihood of a collision between the device and objects within the hazard zone.
[0021] In some embodiments, the hazard zone is a first hazard zone, and the alert system is further configured to determine if any of the one or more modelled positions is within a further hazard zone of a set of one or more further hazard zones (e.g. providing first, second, third, etc. hazard zones). It may determine this by processing reflections of the same acoustic signal for two or more, or all, of the hazard zones, or it may emit different acoustic signals for each hazard zone (from the same transducer or from different respective transducers). It may receive the reflections from two or more, or all, hazard zones at the same array of acoustic transducers, or the alert system may comprise one or more further arrays of acoustic transducers for receiving reflections, for processing, from a further hazard zone.
[0022] The one or more further hazard zones may comprise a single hazard zone or a plurality of hazard zones, and there may be any number of hazard zones, depending on the application in which the alert system is implemented. Each hazard zone may be distinct from every other hazard zone. The hazard zones which may be non-overlapping zones. The hazard zones may all be proximate to the device, although this is not essential. One or more of the hazard zones may have a shortest distance to the device that is greater than a largest distance from the device to another of the hazard zones.
[0023] At least one of the further hazard zones may abut the first hazard zone, e.g., sharing a common border surface. For example, the first hazard zone may be a three-dimensional volume adjacent a surface of the device (e.g. to the front of a mobile device) and a further hazard zone may be a more distal three-dimensional volume that borders the first hazard zone. Alternatively, the first hazard zone and one or all of the further hazard zones may be mutually non-abutting, e.g., not sharing any common border. For example, the first hazard zone may be a three- dimensional volume adjacent a front surface of the device, and a further hazard zone may be a three-dimensional volume adjacent a side surface of the device, where the first and further hazard zones do not abut.
[0024] Any of the features of, or relating to, the first hazard zone, disclosed herein, may be features of, or relating to, any of the further hazard zones.
[0025] The alert system may be configured to signal an alert if any of the one or more modelled positions is within any of the hazard zones. The alert may be a zonespecific alert. Thus, the alert system may be configured to signal a first alert if any of the modelled positions is within the first hazard zone, and to signal a second alert, different from the first alert, if any of the modelled positions is within a second hazard zone, different from the first hazard zone. The alerts may differ by being output over different respective signal paths (e.g. over different electrical lines), or by having different characteristics (e.g. by encoding different digital values).
[0026] The alert system may be configured, when signalling an alert for any of the hazard zones, to provide an output signal (e.g. to the device), which may differ depending on which hazard zone the one or more modelled positions is determined to be in. For example, a first type of output signal may be provided in response to determining that a modelled position is within the first hazard zone, and one or more further types of output signal may be provided in response to determining that the modelled signal is within one or more respective further hazard zones.
[0027] The alert system may be configured to instruct the device to perform a safety function when signalling an alert, wherein the safety function differs depending on which hazard zone the one or more modelled positions is determined to be in. The alert system may thus instruct the device to perform a first safety function (e.g. changing direction) in response to detecting a modelled signal in a first hazard zone, and to perform a second safety function (e.g. stopping) in response to detecting a modelled signal in a second hazard zone. This may allow the device to take an appropriate type or level of safety action that depends on which hazard zone the one or more modelled positions is determined to be in. For example, in embodiments in which the device is a mobile device, the alert system may instruct the device to alter its a course of movement in response to a first output signal associated with a first hazard zone that is relatively far from the device, and to arrest motion of the device in response to a second output signal associated with a second hazard zone relatively close to the device.
[0028] In some embodiments, processing the sampled electrical signals to determine a modelled position for the one or more surfaces may comprise determining a range and / or a 2D position (and optionally a phase) for each of the surfaces. Determining a range for each surface may comprise determining a time-of-flight for the reflected acoustic signal from the respective surface. In some embodiments, the processing may comprise performing receive-beamforming processing of the received signals. Receive-beamforming processing may be performed for each of a plurality of ranges from the alert system. In some embodiments, receive beamforming processing may only be performed for ranges at which one or more surfaces are detected (e.g. ranges for which a reflection of the acoustic signal above a threshold level is received).
[0029] The alert system may be configured, when processing the sampled electrical signals, to ignore contributions to the received acoustic signal from one or more predetermined (e.g. known) surfaces in the environment. For example, contributions from a lower bounding surface (such as the floor of a room in which the device is located) may be ignored. In some embodiments, reflections from a predetermined set of surfaces in the environment (which may be known in advance), or from parts of the device itself, may be ignored.
[0030] The modelled positions of the one or more surfaces may be defined by two- or three-dimensional coordinates with respect to the device or the alert system. The modelled positions may be compared to the hazard zone to determine whether any of the modelled positions is within the hazard zone. For example, the alert system may determine whether any of the modelled positions is located within a set of boundary surfaces defining the hazard zone.
[0031] In some embodiments the alert system may be configured to model acoustic signals emanating from up to a predetermined maximum number of modelled positions. In other words, the total number of modelled positions may be capped at a predetermined value. The alert system may thus cease modelling if a number of positions to be modelled is greater than the alert system is capable of modelling. In some such embodiments, if the predetermined maximum number of modelled positions is reached or exceeded, the alert system may signal an alert indicating that the presence of objects within the hazard zone cannot be ruled out. This signal may be the same as a signal of an inconsistency, or different.
[0032] The alert system may, in some embodiments, be configured to determine the modelled positions iteratively — e.g. starting from an initial number of modelled positions (e.g. one), and increasing the number of modelled positions at each iteration (e.g. by one) until the modelled acoustic signals are consistent with the sampled electrical signals and / or until a predetermined maximum number is reached. The alert system may, after each iteration, determine if the one or more modelled acoustic signals are consistent with the sampled electrical signals output by the array of transducers. It may signal in response to reaching the predetermined maximum number of modelled positions with the modelled acoustic signals still being inconsistent with the sampled electrical signals.
[0033] In some embodiments, modelling acoustic signals emanating from the modelled position(s) may comprise modelling a respective acoustic signal emanating from a point source at each modelled position. Modelling acoustic signals may comprise modelling a respective response of the array of transducers to each of the modelled acoustic signals. This response may correspond to the actual response of the array to received acoustic signals, or it may correspond to the signals output by the array after receive-beamforming has been performed, or it may correspond to the signals prior to receive-beamforming but after downmixing of the sampled signals.
[0034] In some embodiments, determining if the one or more modelled acoustic signals are consistent with the sampled electrical signals may comprise subtracting the modelled acoustic signals from the sampled acoustic signals (or the signals output by the array after receive-beamforming, or the signals prior to receive-beamforming but after downmixing of the sampled signals), and determining whether the remainder is below a threshold level. If the remainder is below the threshold level, it may be determined that the modelled acoustic signals are consistent with the sampled signals. If however, the remainder is above a threshold level, it may be determined that the modelled acoustic signals are not consistent with the sampled signals. Such a determination may cause the alert system to signal an inconsistency, or may cause the alert system to perform a next iteration of an iterative processing of the sampled electrical signals in which the number of modelled positions increases at each iteration.
[0035] In some embodiments, determining if the one or more modelled acoustic signals are consistent with the sampled electrical signals may comprise generating a two- dimensional image representing received acoustic signals from the modelled positions, and comparing the image to a two-dimensional image representing the sampled electrical acoustic signals (or the signals output by the array after receivebeamforming, or the signals prior to receive-beamforming but after downmixing of the sampled signals). In some embodiments, the acoustic signal may be an omnidirectional acoustic signal. However, in some embodiments the acoustic signal may be directional. It may be emitted from one or more transducers of the array of transducers, or from a different transducer.
[0036] Emitting the acoustic signal may comprise applying time division multiplexing or frequency division multiplexing to the emitted signal. This may allow the alert system to address issues with coexistence if multiple alert systems are operating in the same area.
[0037] In some embodiments, the emitted acoustic signal may comprise a chirp signal. The chirp signal may comprise an upward chirp or a downward chirp. In some embodiments, the chirp signal may comprise both an upward chirp and a downward chirp.
[0038] In some embodiments, the emitted acoustic signal may comprise a coded signal. In some such embodiments, emitting the acoustic signal may comprise emitting a plurality of differently coded signals in different directions from the alert system.
[0039] The emitted acoustic signals may have a wavelength between 1.7 mm and 8.5 mm. The bandwidth of the acoustic signals may be between 3 kHz and 30 kHz.
[0040] Sampling of the electrical signals may be performed at 80-400 kHz.
[0041] In some embodiments, the array of acoustic transducers may comprise a plurality of transducers arranged in a linear or two-dimensional array. The two-dimensional array may comprise a first linear array and a second linear array, the first and second linear arrays extending along perpendicular axes. The first and second linear arrays may intersect in some embodiments. The two-dimensional array may comprise at least one additional transducer located at a position that is offset from the first and second linear arrays in some embodiments.
[0042] In some embodiments, the first and second linear arrays may be arranged in a cross shape. The cross-shaped array of transducers may comprise additional transducers located at one or more diagonals of the cross-shaped array. Providing transducers on one or more diagonals of the array may allow the alert system to resolve ambiguities in surface positions in the event that two or more surfaces are located at the same range from the acoustic transducers.
[0043] In some embodiments in which the array is a linear array, the transducers of the array may be arranged in a repeating pattern. The transducers of the array may be arranged such that there is a constant pitch between the transducers. In some such embodiments, the number of transducers in the linear array may be between three and sixteen.
[0044] In some embodiments in which the array is a 2D array, the transducers of the array may be arranged in a repeating pattern along at least one axis (i.e. one dimension). The transducers of the array along the at least one axis may be arranged such that there is a constant pitch between the transducers. The number of transducers along each axis of the 2D array may be a respective number between 8 and 16. The number of transducers along each axis (i.e. each dimension) of the 2D array may be identical in some embodiments, however in some alternative embodiments the number of transducers may be different along the two axes. This may allow for greater resolution to be achieved along one axis than the other, e.g. to reduce the amount of processing required along an axis for which low angular resolution is acceptable. In some embodiments, the pitch between transducers of the 2D array may be between 0.85 mm and 25 mm, preferably between 1.8 mm and 10 mm.
[0045] In some embodiments in which the transducers are arranged in a 2D array, each of the transducers may be arranged at a respective position on a 2D grid (e.g. a rectangular grid). The spacing between each position on the grid may be identical in one or both dimensions, e.g. the transducers on the grid may have a constant pitch. Placing the transducers on a grid in this way may facilitate performing Fast Fourier Transform (FFT) processing of the received reflections of the acoustic signal. In some embodiments, the grid may include positions at which no transducer is located. In some embodiments, each axis of the grid has at least two transducers at neighbouring positions along the axis. In some embodiments, the array comprises at least one pair of transducers that are next to each other along a diagonal of the two axes. It may comprise, for each of two diagonals of the 2D array, at least one respective pair of transducers that are next to each other along the diagonal. The alert system may comprise a processing system configured to perform at least the steps of processing, modelling and determining. The processing system may comprise one or more processors and a memory storing software for execution by the one or more processors.
[0046] Features of any aspect or embodiment described herein may, wherever appropriate, be applied to any other aspect or embodiment described herein. Where reference is made to different embodiments or sets of embodiments, it should be understood that these are not necessarily distinct but may overlap.
[0047] BRIEF DESCRIPTION OF THE DRAWINGS
[0048] Certain preferred embodiments of this disclosure will now be described, by way of example only, with reference to the accompanying drawings, in which:
[0049] FIG. 1 is a schematic illustration of an alert system according to an embodiment of the present invention;
[0050] FIG. 2A is a schematic three-dimensional diagram of an array of transducers for an alert system according to an embodiment of the invention;
[0051] FIG. 2B is a schematic diagram of an array of transducers for an alert system according to an embodiment of the invention;
[0052] FIG. 3 is a flow diagram illustrating steps of a method of operating an alert system according to an embodiment of the invention;
[0053] FIG. 4 is a schematic plan view of a mobile the alert system according to an embodiment of the invention;
[0054] FIG. 5 shows three graphs of signal magnitude against angle to illustrate processing of a received acoustic signal according to an embodiment of the invention;
[0055] FIG. 6 is a schematic plan view of a mobile robot carrying the alert system;
[0056] FIG. 7 is a schematic plan view of an exemplary application of the alert system according to an embodiment of the invention at a first time; and
[0057] FIG. 8 is a schematic plan view of an exemplary application of the alert system according to an embodiment of the invention.
[0058] DETAILED DESCRIPTION
[0059] Figure 1 shows a simplified schematic block diagram of an alert system 100 according to embodiments of the invention. The alert system 100 may be implemented as a static device (e.g. a sensor package) that can be carried by a device, e.g. a mobile device such as an autonomous mobile robot, or may be integral with such a device.
[0060] The alert system 100 comprises an array 101 of acoustic transducers 103, an analogue to digital converter (ADC) 105, a processor 107, a memory 109 and a battery 110 for powering the system.
[0061] The ultrasonic transducers 103 of the array 101 each comprise a receive element for receiving acoustic signals, which in this embodiment takes the form of a microphone 104. The microphone 104 of each transducer 103 is separated from the microphone 104 of the neighbouring transducer(s) 103 by a distance d. The same transducers 103, or a different set of one or more transducers, are used to emit acoustic signals from the alert system 100, reflections of which can be received by the microphones 104. The array 101 may be used to determine angular positions of acoustical ly-reflective surfaces, using receive-beamforming processing. When combined with time-of-flight processing, this enables the positions of surfaces relative to the array 104 to be determined in terms of an angle relative to the array and a range.
[0062] Although the array 101 in Figure 1 is shown as comprising eight transducers 103 in a linear arrangement, it will be appreciated that different numbers and arrangements of transducers 103 may be provided in the array in other embodiments. Examples of this are illustrated in Figures 2A and 2B, which show alternative arrays 201, 205 suitable for use in place of the array 101 in Figure 1.
[0063] The alert system 100 may be switchable between an active mode and an inactive mode, e.g. by means of a physical mute button. It may cease acoustic transmissions in the inactive mode and / or may cease outputting alert signals. This may be useful when servicing the system 100.
[0064] Figure 2A shows a perspective view of an array 201 comprising fifteen transducers 203, each comprising an ultrasonic transmitter (not shown) and a microphone 204 for receiving acoustic signals. The transducers 203 are arranged in two linear arrays that extend in perpendicular directions in a cross-shaped arrangement. In each perpendicular direction, the microphone 204 of each transducer 203 is separated from the microphone 204 of the neighbouring transducers 203 by a distance d. The array 201 shown in Figure 2A may be used to determine the angular position of a surface along two perpendicular axes, allowing the position of surfaces relative to the array to be determined in terms of a two-dimensional angular position and a range (i.e. in three-dimensions).
[0065] Figure 2B shows a top-down view of another two-dimensional array 205 of transducers 207, having similar to the transducers 203 shown in Figure 2A, and arranged at respective positions on a grid 209 that also includes ‘empty’ positions 211 at which no transducer is present. In Figure 2B, the transducers 207 are positioned on the grid 209 such that that there are successive transducers at neighbouring positions in each of two perpendicular directions (shown as x and y in Figure 2B), i.e. arranged in a cross, as well as two further transducers that, along with the central transducer of the cross, are located along a diagonal from bottomleft to top-right. Providing transducers on one or more of the diagonals of the array allows for ambiguities in surface positions to be resolved in the event that two or more surfaces are located at the same range from the acoustic transducers 207.
[0066] Although the embodiment shown in Figure 2B includes a linear array of six transducers 207 along each of two perpendicular axes (shown as x and y axes in Figure 2B) and three transducers 207 along a diagonal, formed of thirteen transducers 207 in total, this is merely an example and is not required in all embodiments. The distribution of transducers along the perpendicular axes can be set based on the nature and distribution of objects to be detected using the alert system. In general, a greater number of transducers along a given axis allows for a greater number of objects along the perpendicular axis to be distinguished from one another. For example, vertical poles are best distinguished by a large number of transducers in the horizontal direction, whereas horizontal poles are best distinguished by a large number of transducers in the vertical direction. The transducers along the diagonals serve to resolve ambiguities by allowing pairing of x and y coordinates identified using the transducers of perpendicular arrays.
[0067] Returning to Figure 1, the alert system 100 is configured to emit an ultrasound signal, e.g. from one or more transmitter elements of the transducers 103, into an environment in the vicinity of the alert system 100. This may be an omnidirectional signal, e.g. in the form of a spherical wave. It may be a chirp signal (e.g. an upward chirp followed by a downward chirp). The emitted signal has a bandwidth of 8 kHz with a centre frequency that is selected such that the wavelength, A, of the acoustic signal is approximately equal to twice the spacing, d, between the microphones 104 of the array 101. The emitted signal may, depending on the environment, be reflected from surfaces in the environment (not shown in Figure 1) such that reflected signals 102 are received, at detectable levels, at the microphones 104 from the surface(s) within the environment. Although multiple reflections may be received from different surfaces within the environment in practice, only a single reflected wavefront is shown in Figure 1 for simplicity.
[0068] The reflections of the acoustic signal measured by the microphones 104 are sampled by the ADC 105, which provides sampled data to the processor 107. The processor 107 is arranged to execute software stored in the memory 109 to process the sampled signals to operate the alert system as explained in the following.
[0069] The alert system 100 is configured to be carried by a device such as mobile device, e.g. an autonomous mobile robot (AMR), or a static device such as a robot in a fixed cell or work area. The alert system 100 is used to determine whether any objects are present within a hazard zone proximate to the device, e.g. a three- dimensional volume adjacent the device defined by a set of boundary surfaces, and to provide an output if so. For example, the alert system 100 may be used to determine whether an autonomous mobile robot is approaching an object in an environment by detecting an object in the hazard zone, and signalling an alert when the object is detected. As explained in more detail below with reference to Figures 3-5, the alert system 100 is also configured to signal an inconsistency in the event that the presence of an object in the hazard zone cannot be ruled out.
[0070] Figure 3 is a flow diagram illustrating steps of a method of operating the alert system 100 shown in Figure 1.
[0071] In step 301, the alert system 100 is configured to emit an acoustic signal using one or more of the transducers 103. The emitted acoustic signal propagates in the environment surrounding the alert system 100, and is reflected from the surfaces of objects in the vicinity of the alert system 100 within the environment.
[0072] In step 303, reflections of the acoustic signal from surfaces in the vicinity of the alert system are received at the array 100 of acoustic transducers 103.
[0073] In step 305, the ADC 105 samples electrical signals output by the array 100 of acoustic transducers 103 in response to received reflections of the acoustic signal, and provides an output to the processor 107.
[0074] In step 307, the processor 107 processes the sampled electrical signals to determine a respective modelled position for each of the one or more surfaces. This is achieved by determining a range to each of the surfaces based on a time of flight of the received reflections of the acoustic signal, and determining a respective angular position of each of the surfaces with respect to the array 104, so as to define a modelled position of each of the one or more surfaces. If the emitted signal was a chirp signal, a de-chirp or other pulse decompression may first be performed. Determination of the angular position of the surfaces with respect to the array 101 may be achieved by analysing phase differences between the reflected acoustic signal received by different transducers 103 of the array 101. This analysis may be performed in any suitable way, however in some advantageous embodiments it comprises beamforming the received reflected signals to construct an image which reveals the angular direction to the reflecting surface from the array, which combined with a range determined from time of flight measurement provides a modelled position of the surface with respect to the array.
[0075] In some embodiments, the processing may be performed in a single processing step that allows for up to a maximum number of modelled positions (e.g. assuming up to a maximum of four or eight surfaces). In other embodiments, the processing may be performed iteratively, starting assuming only one surface is present, and then allowing a further modelled position to be added at each successive iteration, e.g. until a maximum number is reached.
[0076] In step 309, having determined a modelled position for each of the surfaces, the processor 107 models an acoustic signal emanating from each of the modelled surface positions, so as to generate a model of the environment in the vicinity of the alert system 100. In this step, the processor 107 may generate model of a point source located at each of the modelled positions, where each point source is assumed to emit an acoustic signal having a magnitude and phase set based on the strength of the reflected acoustic signal received from the surface. Once a modelled acoustic signal has been modelled for each of the surfaces, the modelled acoustic signals are analysed to determine firstly whether they are an accurate representation of the environment in the vicinity of the alert system, and secondly whether they indicate that an object is present in the hazard zone in the proximity of the alert system 100 (which may be defined by variables stored in the memory 110 of the alert system 100).
[0077] In step 311, the modelled acoustic signals are compared to the received reflections of the acoustic signal to determine whether the modelled acoustic signals are consistent with the sampled electrical signals. This step allows the processor 107 to determine whether the acoustic signals sufficiently explain the acoustic signals actually received by the system, and in doing so allow a determination to be made as to whether there are likely to be objects in the environment that are not adequately accounted for by the model. For example, a difference between the modelled acoustic signal and the received reflections of the acoustic signal may be determined, and if this difference is greater than a threshold level, it may be determined that objects are present in the environment that are not included in the model.
[0078] If this occurs, the model is unable to rule out the presence of an object in the hazard zone around the alert system 100, and the alert system 100 is configured to signal an inconsistency. In addition to signalling an inconsistency, the alert system 100 may send a safety command to the device by which it is carried. For example, in embodiments in which the device is a mobile device that is in motion at the time an inconsistency is signalled, the alert system 100 may instruct the device to reduce its speed or to stop entirely to reduce the likelihood of a collision with an object that is not accounted for by the model. In this way, the alert system 100 defaults to a safe mode of operation to reduce the potential for collisions between the device and undetected objects that may be proximate the device. In step 313, the processor 107 determines whether any of the modelled positions is within the hazard zone proximate the device. If it is determined that an object is present in the hazard zone proximate the device, the alert system 100 is configured to signal an alert. In addition to signalling an alert, the alert system 100 may send a safety command to the device by which it is carried (either by the alert signal itself or by a further output signal). For example in embodiments in which the device is a mobile device that is in motion at the time an inconsistency is signalled, the alert system 100 instruct the mobile device to reduce its speed or to stop entirely to reduce the likelihood of a collision with the object determined to be within the hazard zone.
[0079] By employing the method shown in Figure 3, the alert system 100 is able to identify the presence of objects in the environment proximate the device by which it is carried, and to generate a signal in the event that the presence of an object in a hazard zone around the alert system is either determined or cannot be ruled out. In this way, the alert system 100 is able to reduce the likelihood of collisions between a device carrying the alert system 100 and objects in the surrounding environment that may otherwise not have been detected.
[0080] The method shown in Figure 3 may be more clearly understood with reference to Figures 4-6, which show an example use case for the alert system 100, in which the alert system 100 is mounted to an autonomous mobile robot (AMR) 400. In this embodiment, the alert system 100 is a module that is arranged to interface with a control system of the AMR 400, such that alert system 100 is able to output signals (e.g. one or more OSSD signals) to the AMR 400 and vice-versa.
[0081] Figure 4 shows the alert system 100 mounted to an autonomous mobile robot (AMR) 400 at a first time ti. The AMR 400 is configured to move within an environment 420, such as a warehouse, which includes various static objects 430a- 430d, and moving people, such as a warehouse employee 440, moving in the direction indicated by the arrow adjacent the warehouse employee 440. While the locations of the static objects 430a-430d may be known in advance, e.g. they may be stored in a memory of the AMR 400, the AMR 400 needs to be able to detect the locations of moving people, such as the warehouse employee 440, in real-time, so as to prevent collisions with the AMR 400. The AMR 400 is configured to communicate with the alert system 100 for this purpose, by determining whether an obstacle (e.g. an object 430a-430d or a moving person such as warehouse employee 440) is located within a hazard zone 410 proximate the AMR 400, or if the presence of an object 430a-430d or a moving person such as warehouse employee 440 cannot be ruled out, and signalling to the AMR 400 in either case.
[0082] In the example shown in Figure 4, the AMR 400 carrying the alert system 100 is in the process of moving at low speed from its location shown in Figure 4 towards a destination zone 450. While the AMR 400 is moving at low speed, the hazard zone 410 proximate the AMR 400 is set, based on parameters saved in a memory 110 of the alert system 100 and variables received from the control system of the AMR 400 (i.e. a velocity of the AMR 400), as a cuboid extending outwards from the alert system 100 in the direction of travel of the AMR 400.
[0083] The alert system 100 is configured to periodically emit an acoustic signal into the environment 420. Reflections 431b and 431 d, from the objects 430b and 430d respectively, are received at the microphones 104 of the alert system 100. A reflected acoustic signal 441 is also received at the microphones 104 of the alert system 100 from the warehouse employee 440. The ADC 105 of the alert system 100 samples the received reflections 431b, 431 d and 441, and the processor 107 processes the sampled electrical signals to determine a modelled position for each of the obstacles from which a reflected acoustic signal is received, i.e. from objects 430b, 430d and the warehouse employee 440.
[0084] Having determined a modelled position for each of the objects 430a-430d and the warehouse employee 440, the alert system 100 models a respective acoustic signal emanating from each of the modelled positions. The alert system 100 then determines whether any of the modelled positions is in the hazard zone, and determines whether the modelled acoustic signals are consistent with the sampled electrical signals.
[0085] To determine whether the modelled acoustic signals are consistent with the sampled electrical signals, the processor 107 of the alert system 100 is configured to subtract the modelled acoustic signals from the sampled electrical signals for each of a plurality of ranges at which reflected acoustic signals are received, and to determine whether the remainder signal is greater than a threshold level for that range. If the remainder signal is greater than the threshold level at any range, it is determined that there is an inconsistency. If the remainder signal is below a threshold level for all ranges, it is determined that the modelled signal accurately represents the environment in the vicinity of the alert system and no inconsistency is signalled.
[0086] An example of this will now be described with reference to the exemplary graphs in Figure 5. Figure 5 shows magnitude against angle for a sampled electrical signal 501 generated from acoustic signals received from the alert system 100 for a distance range at which both the warehouse employee 440 and the object 430b are present. It can be seen that the sampled electrical signal 501 includes two peaks overlaid on background noise. The processor 107 processes the sampled electrical signal to determine a modelled position for each object from which a reflected acoustic signal is received, i.e. at angular positions at which the sampled electrical signal magnitude is greatest.
[0087] The alert system 100 models acoustic signals emanating from point sources at each of the modelled positions to generate a modelled acoustic signal 503. The modelled acoustic signal 503 is subtracted from the sampled electrical signal 501 to generate a remainder signal 505, which may be analysed to determine whether the modelled acoustic signal 503 is consistent with the sampled electrical signal 501. This is achieved by comparing the remainder signal 503 to a threshold signal level 505. As the remainder signal 503 is below the threshold signal level 505, it is determined that there is no inconsistency.
[0088] Other embodiments may, of course, use different methods to check for consistency.
[0089] In the example shown in Figure 4, i.e., at time ti , it is determined that no inconsistency is present and no objects are present within the hazard zone 410.
[0090] Figure 6 shows the autonomous mobile robot (AMR) 400 carrying the alert system 100 at a later time t2 than the time h shown in Figure 4. At the later time t2, the AMR 400 is moving at a greater speed towards the destination zone 450, in a direction indicated by the arrow proximate the AMR 400. In response to the increased speed of the AMR 400, the alert system 100 configures the hazard zone 410 such that its extent in the direction in which the AMR 400 is moving is increased. This change in the hazard zone boundary is made based on variables received at the alert system 100 from the AMR 400, e.g. the velocity of the AMR 400 as determined by an onboard sensor of the AMR 400 such as a wheel encoder.
[0091] By configuring the hazard zone 410 based on the speed of the AMR 400, the alert system 100 is able to detect objects or people that may be at risk of collision with the AMR 400 at a greater distance from the AMR 400 to compensate for its increased speed. This allows the alert system 100 to signal to the AMR 400 further in advance of a potential collision, such that the AMR 400 can take a safety action such as slowing or stopping.
[0092] Reflections 431b and 431 d of the acoustic signal are received at the microphones 104 of the alert system 100 from the objects 430b and 430d. A reflection 441 is also received at the microphones 104 of the alert system 100 from the warehouse employee 440. The ADC 105 of the alert system 100 samples the received reflections 431b, 431 d and 441 , and the sampled electrical signals are processed by the processor 107 to determine a modelled position for each of the objects from which a reflected acoustic signal is received, as described above.
[0093] Having determined a modelled position for the objects 430b, 430d and the warehouse employee 440, the alert system 100 models a respective acoustic signal emanating from each of the modelled positions. The alert system 100 then determines whether the modelled acoustic signals are consistent with the sampled electrical signals, and whether any of the modelled positions is in the hazard zone 410.
[0094] At the time t2 shown in Figure 6, the alert system 100 determines that one of the modelled positions corresponds to an obstacle (the warehouse employee 440) located within the hazard zone 410, causing the alert system 100 to signal an alert. In the embodiment shown in Figure 6, an alert signal is provided to the control system of the AMR 400, allowing the AMR 400 to take a safety action. Specifically, as an obstacle is identified as being in the hazard zone 410, the AMR is configured to stop in response to receiving an alert signal from the alert system 100, in order to prevent a collision with the warehouse employee 440.
[0095] Figures 7 and 8 show an example of an alert system 100' that is configured to determine if an object is present in one or more of a plurality of hazard zones proximate to an AMR 700 to which the alert system is mounted. The alert system 100' may be the same as or similar to the alert system 100, but is additionally configured to signal zone-specific alerts that differ depending on which hazard zone an object is detected as being in, as explained below.
[0096] The alert system 100' is arranged to interface with a control system of the AMR 700, such that alert system 100' is able to output signals to the AMR 700 and vice- versa. The AMR 700 is configured to move within an environment 720, such as a warehouse, which may include various static objects (not shown in Figures 7 and 8) and moving people, such as a warehouse employee 740, moving in the direction indicated by the arrow adjacent the warehouse employee 740. The alert system 100' is configured to determine whether an obstacle (e.g. a moving person such as warehouse employee 740) is located within either of a first hazard zone 710 or a second hazard zone 711 , and to signal an alert if so. As described above with reference to the alert system 100, the alert system 100' is also configured to determine or if the presence of an object or a moving person cannot be ruled out, and to signal to the AMR 700 if so.
[0097] Figure 7 shows the alert system 100' mounted to the AMR 700 at a first time ta. At this time, the AMR 700 is in the process of moving at relatively high speed from its location shown in Figure 7 towards a destination zone 750, in a forwards direction (indicated by the arrow).
[0098] The hazard zones 710 and 711 are set based on parameters saved in a memory 110 of the alert system 100' and variables received from the control system of the AMR 700 (i.e. a velocity of the AMR 700). The first hazard zone 710 may be set as a first cuboid that extends outwards from the alert system 100' and AMR 700 from a front face of the AMR 700 (i.e. in the direction of travel of the AMR 700 when the AMR 700 is travelling straight forwards). The second hazard zone 711 may be set as a second cuboid that extends outwards from the first hazard zone 710 in a forwards direction.
[0099] In Figure 7, the alert system 100' is configured to determine whether an obstacle is present in either of two contiguous (i.e. abutting) hazard zones 710, 711. However, it will be appreciated that in other embodiments, there may be a greater number of hazard zones (e.g. three, four or more), and the hazard zones may or may not each abut one or more other hazard zones, e.g. there may be a first hazard zone that extends in a forwards direction of travel and another hazard zone that extends in a different direction (e.g. in an upwards or sideways direction) and need not necessarily border the first hazard zone.
[0100] As described above in relation to Figures 4-6, the alert system 100' is configured to periodically emit an acoustic signal into the environment 720 and to receive reflections of the emitted acoustic signals from objects and people within the environment 720. In the example shown in Figure 7, a reflection 741 from the warehouse employee 740 is received at the microphones 104 of the alert system 100'. The ADC 105 of the alert system 100' samples the received reflection 741, and sampled electrical signals are processed by the processor 107 to determine one or more modelled positions for the warehouse employee 740 (e.g. one modelled position if the received reflected signal arises principally from the employee's torso, or potentially two modelled positions if strong reflections are received from each of the employee’s legs).
[0101] The alert system 100' models an acoustic signal emanating from each modelled position, and determines whether the modelled acoustic signals are consistent with the sampled electrical signals, and whether any modelled position is in either of the hazard zones 710, 711.
[0102] At the time tashown in Figure 7, the alert system 100' detects that a modelled position is located within the second hazard zone 711 , which causes the alert system 100' to signal an alert. An alert signal, identifying the second hazard zone, is provided to the control system of the AMR 700, allowing the AMR 700 to take a safety action. As an obstacle is identified as being in the second hazard zone 711 (i.e. far from the AMR 700), the AMR 700 is configured to slow down and change direction in order to prevent a collision with the warehouse employee 740. In this way, the AMR 700 may take pre-emptive action to avoid collision with obstacles far from the AMR 700, while the risk of collision is low. This allows the AMR to continue moving towards the destination zone 750, albeit at reduced speed.
[0103] Figure 8 shows the alert system 100' mounted to the AMR 700 at a second time tb, later than the time tashown in Figure 7. At the time tb, the AMR 700 has changed direction and speed, and is in the process of moving at reduced speed from its location shown in Figure 8 towards the destination zone 750. The warehouse employee 740 has also changed position and has moved closer to the destination zone 750.
[0104] In response to the reduced speed of the AMR 700, the alert system 100' reconfigures the hazard zones 710 and 711 such that their extents in the direction in which the AMR 700 is moving is reduced. This change in the hazard zone boundaries is made based on variables received at the alert system 100' from the AMR 700, e.g. the speed of the AMR 700 as determined by an on-board sensor of the AMR 700 such as a wheel encoder. By configuring the size and / or shape of each of the hazard zones 710 and 711 based on the speed of the AMR 700, the alert system 100' is able to reduce the range at which detected objects or people are determined to be at risk of collision with the AMR 700 to take account of its reduced speed and so reduce the risk of unnecessary alerts.
[0105] At the time tb shown in Figure 8, a reflection 742 is received at the microphones 104 of the alert system 100' from the warehouse employee 740. The ADC 105 of the alert system 100' samples the received reflection 742, and sampled electrical signals are processed by the processor 107 to determine one or more modelled positions for the warehouse employee 740.
[0106] The alert system 100' models an acoustic signal emanating from each modelled position, and determines whether the modelled acoustic signals are consistent with the sampled electrical signals, and whether any modelled position is in the hazard zone 710 or the hazard zone 711. At the time tb shown in Figure 8, the alert system 100' determines that a modelled position is located within the first hazard zone 710, causing the alert system 100' to signal an alert. An alert signal that identifies the first hazard zone is provided to the control system of the AMR 700, allowing the AMR 700 to take a safety action. As an obstacle is identified as being in the first hazard zone 710 (i.e. close to the AMR 700), the AMR is configured to stop in response to receiving the alert signal, in order to prevent a collision with the warehouse employee 440.
[0107] By providing multiple hazard zones and signalling different alerts that depend on the hazard zone in which an obstacle is detected, the alert system 100' is thus able to take different actions to avoid collision with obstacles based on the relative locations of the obstacle and the AMR to which the alert system 100' is mounted.
[0108] It will be appreciated by those skilled in the art that the present disclosure has been illustrated by describing one or more specific examples thereof, but is not limited to these examples; many variations and modifications are possible, within the scope of the accompanying claims.
Claims
CLAIMS1. An alert system configured to be carried by a device and configured to signal an alert if an object is present within a hazard zone proximate to the device, wherein the alert system comprises: an array of acoustic transducers; and a processing system, wherein the alert system is configured: to emit an acoustic signal; to receive, at the array of acoustic transducers, reflections of the acoustic signal from one or more surfaces in a vicinity of the alert system; to sample electrical signals output by the array of transducers in response to the received reflections; to process the sampled electrical signals to determine a respective modelled position for each of the one or more surfaces; for each of the one or more modelled positions, to model a respective acoustic signal emanating from the modelled position; to determine if the one or more modelled acoustic signals are consistent with the sampled electrical signals output by the array of transducers in response to the received reflections, and, if not, to signal an inconsistency; and to determine if any of the one or more modelled positions is within the hazard zone, and, if so, to signal an alert.
2. The alert system of claim 1, wherein the hazard zone is a three-dimensional volume proximate the device, defined by a set of boundary surfaces.
3. The alert system of claim 1 or 2, wherein the hazard zone is configurable.
4. The alert system of claim 3, wherein the hazard zone is configurable based on a speed and / or trajectory of the device.
5. The alert system of any preceding claim, wherein the alert system is configured to provide an output signal to the device when signalling an inconsistency and / or an alert.
6. The alert system of claim 5, wherein the output signal comprises an instruction to the device to perform a safety function.
7. The alert system of claim 6, wherein the safety function comprises the device altering its course or wherein the safety function comprises arresting motion of the device.
8. The alert system of any preceding claim, wherein processing the sampled electrical signals to determine a respective modelled position for each of the one or more surfaces comprises determining a range and a two-dimensional position for each surface.
9. The alert system of claim 8, wherein determining a range comprises determining a time of flight of the reflected acoustic signal.
10. The alert system of claim 8 or 9, wherein processing the sampled electrical signals further comprises applying receive-beamforming processing to the sampled electrical signals.
11. The alert system of claim 10, wherein receive-beamforming is only performed for ranges at which one or more surfaces are detected.
12. The alert system of any preceding claim, wherein each modelled acoustic signal is modelled as emanating from a point source at a respective modelled position.
13. The alert system of any preceding claim, wherein determining if the one or more modelled acoustic signals are consistent with the sampled electrical signals comprises generating a two-dimensional image representing received acoustic signals from the modelled positions, and comparing the image to a two-dimensional image representing the sampled electrical acoustic signal.
14. The alert system of any preceding claim, wherein when processing the sampled electrical signals, reflections of the acoustic signal from one or more known surfaces are ignored.
15. The alert system of any preceding claim, wherein the array of acoustic transducers comprises a plurality of transducers arranged at respective positions on a two-dimensional grid.
16. The alert system of claim 15, wherein the two-dimensional array comprises a first linear array and a second linear array, wherein the first and second linear arrays extend along perpendicular axes.
17. The alert system of claim 16, wherein the two-dimensional array comprises a transducer located at a position on the grid that is offset from both of the first and second linear arrays.
18. The alert system of any preceding claim, wherein the alert system is further configured to determine if any of the one or more modelled positions is within a further hazard zone of a set of one or more further hazard zones, and, if so, to signal an alert.
19. The alert system of claim 18, wherein the alert system is configured, when signalling an alert, to output a signal that differs depending on which hazard zone the one or more modelled positions is determined to be in.
20. A device comprising the alert system of any preceding claim, wherein the device is a robot in a fixed cell or work area, or is an autonomous mobile robot.
21. A method of operating an alert system, the method comprising: emitting an acoustic signal from the alert system; receiving reflections of the acoustic signal from one or more surfaces in a vicinity of the alert system; sampling electrical signals in response to the received reflections; processing the sampled electrical signals to determine a respective modelled position for each of the one or more surfaces; for each of the one or more modelled positions, modelling a respective acoustic signal emanating from the modelled position;determining if the one or more modelled acoustic signals are consistent with the sampled electrical signals output by the array of transducers in response to the received reflections, and, if not, signalling an inconsistency; and determining if any of the one or more modelled positions is within a hazard zone, and, if so, signalling an alert.