Hand detection and control of sawing machines

Abstract

A sawing machine comprises at least one camera, a sensor, a saw blade, a motor configured to drive the saw blade, and an electronic controller having an electronic processor and memory. The electronic controller receives an indication of the orientation of the saw blade from the sensor and determines a restricted area based on the orientation of the saw blade, wherein the restricted area corresponds to or is defined relative to the saw blade. Images captured by the at least one camera are received by the electronic controller, wherein the captured images include at least a portion of the restricted area. The electronic controller analyzes the captured images using a machine learning (ML) model to detect whether a portion of a hand is located in the restricted area and performs a safety action in response to detecting the portion of the hand in the restricted area.

Description

Cross-reference to related applications

[0001] This application claims the benefit of U.S. Provisional Patent Application No. 63 / 608,034, entitled "POWERED SAW HAND DETECTION AND CONTROL," filed December 8, 2023, and U.S. Provisional Patent Application No. 63 / 616,115, entitled "POWERED SAW HAND DETECTION AND CONTROL," filed December 29, 2023, each of which is incorporated herein by reference in its entirety. background

[0002] Power sawing tools consist of a motor that drives (e.g., rotates) a saw blade at high speed to cut through a work material (e.g., wood, plastic, metal). These tools include, for example, chop saws, table saws, circular saws, panel saws, and the like, which are widely used in construction, woodworking, and various other industries to allow the user to make controlled cuts in a range of materials. For various reasons, injuries can occur when using a power sawing tool if a powered saw blade comes into contact with the user's hand. Overview of Revelation

[0003] The present disclosure provides a sawing machine comprising a camera, an inertial measurement unit (IMU) sensor, a saw blade, a motor configured to drive the saw blade, and an electronic controller having an electronic processor and a memory. The electronic controller is configured to: receive an indication of an orientation of the saw blade from the IMU sensor; determine an exclusion zone based on the orientation of the saw blade, wherein the exclusion zone corresponds to the saw blade; receive captured images from the camera, wherein the captured images include at least a portion of the exclusion zone; analyze the camera images using a machine learning (ML) model to determine whether a portion of a hand is located in the exclusion zone; and perform a safety action in response to detecting the portion of the hand in the exclusion zone.

[0004] Another aspect of the present disclosure is to provide a method for operating a sawing machine. The method includes receiving an indication of the orientation of a saw blade from a sensor of the sawing machine; determining an exclusion zone based on the orientation of the saw blade, the exclusion zone defining a volume relative to the saw blade; receiving image data from a camera of the sawing machine, the image data representing at least a portion of the exclusion zone; analyzing the camera images using an ML model to determine whether a portion of a hand is located in the exclusion zone; and performing a safety action in response to detecting the portion of the hand in the exclusion zone.Other embodiments of this aspect include corresponding systems (e.g., computer systems), programs, algorithms, and / or modules, each configured to perform the steps of the methods. Short description of the drawings Fig. 1A shows an example of a miter saw according to some embodiments. Fig. 1B shows an example of a camera assembly coupled to a chop saw, according to some embodiments. Fig. 1C shows an example of a miter saw having two cameras coupled to it, according to some embodiments. Fig. Figure 1D shows an example of a three-dimensional exclusion zone projected onto a two-dimensional camera plane. Fig. 1E shows an example of a warning exclusion zone according to some embodiments. Fig. 1F shows an example of a hazard exclusion zone according to some embodiments. Fig. 2A and Fig. 2B show an example of a table saw according to some embodiments. Fig. 3 is a block diagram of a sawing machine tool according to some embodiments. Fig. 4 is a flowchart illustrating an example of a method for controlling operation of a sawing machine according to some embodiments. Fig. 5 is a flowchart illustrating an example method for detecting an unsafe operating condition of a sawing machine according to some embodiments. Fig. Figure 6 shows an example of images captured with cameras coupled to a cross-cut saw. Fig. 7 is a flowchart illustrating an example method for determining whether a detected hand is within a restricted area of ​​a sawing machine, according to some embodiments. Fig. 8 shows an example of estimating the pose of a miter saw according to some embodiments. Fig. Figure 9A shows an example of a detected hand object that was identified as overlapping a warning exclusion zone. Fig. Figure 9B shows an example of a detected hand object that was identified as overlapping a hazard exclusion zone. Fig. 10 is a flowchart illustrating an example method for training a machine learning model for hand detection, according to some embodiments. Fig. 11 is a flowchart illustrating an example of a method for detecting an unsafe operating state of a sawing machine based on processing image data with an optical flow detection model. Detailed description

[0005] The systems and methods provided herein are designed to detect the presence of a user's hand in one or more restricted areas (e.g., near the saw blade or the path of the saw blade) during operation of a sawing machine and then to control the sawing machine to perform a safety action (e.g., generate a warning, stop the saw blade, or both generate the warning and stop the saw blade).

[0006] In some of the embodiments described herein, a sawing machine determines a restricted area based on the output of sensors (such as inertial sensors such as an inertial measurement unit (IMU)), executes a machine learning (ML) model to analyze camera images containing the restricted area, and performs a safety action (such as generating a warning and / or stopping the motor) in response to detecting a user's hand in the restricted area.

[0007] The disclosed hand recognition techniques have robust recognition capabilities. For example, the hand recognition techniques are not limited to detecting flesh (e.g., using a capacitive sensor), specific optical characteristics (e.g., certain skin tones, colors of work gloves (e.g., green or blue work gloves), etc.), or specific lighting conditions. Rather, they can detect a hand in various scenarios (e.g., wearing a glove, partially obscuring the view of the hands by clothing, in different lighting conditions, with different skin tones, when the sawing machine is operated at different miter and bevel angles, etc.). This robustness results, for example, from training the ML model that detects the hands from images that capture these different scenarios.Furthermore, the hand detection techniques can be adapted depending on the posture or orientation of the sawing machine. For example, the hand detection techniques can define a restricted area based on the output of an IMU or other sensor attached to the saw (e.g., a saw arm), allowing the hand detection techniques to account for different bevel and skew angles. This means that the restricted area can be defined based on the output of the IMU or other sensors, allowing the restricted area to be dynamically changed or updated as the bevel and / or skew angle of the saw blade is changed by the user.

[0008] In addition to hand detection in a restricted area of ​​a sawing machine, the systems and methods described in the present disclosure can also be used for other tools where a workpiece is movable or where the user's hands may come into close contact with a moving component of the tool. In these cases, the position of the workpiece may be detected and tracked relative to a restricted area and / or the user's hands may be detected and tracked relative to a restricted area. For example, the techniques described herein can also be applied to other machine tools and industrial machines such as planers, jointers, routers, drill presses, hydraulic presses, etc.

[0009] Fig. 1A-1C illustrate a compound miter saw 100 according to some embodiments. Compound miter saw 100 is an example of a sawing machine in which the hand detection and control techniques described herein may be employed. As illustrated, compound miter saw 100 includes one or more cameras 110 coupled to compound miter saw 100. Camera(s) 110 may be coupled to the housing of compound miter saw 100, an arm of compound miter saw 100, or the like. In the embodiment shown in Fig. 1B, a camera 110 may be mounted or otherwise attached to a boom 112 coupled to the housing of the miter saw 100.

[0010] An indicator light 114 (e.g., a feedback light) may be coupled or otherwise integrated with the boom 112. As described in more detail below, the indicator light 114 may be operated to perform a safety action when a hand is detected in a restricted area of ​​the miter saw 100. For example, the indicator light 114 may be operated to produce different colored light depending on the safety action being performed. As a non-limiting example, when a warning safety action is being performed (e.g., when a hand is detected in a warning restricted area), the indicator light 114 may be operated to produce a yellow light. When a hazard safety action is being performed (e.g., when a hand is detected in a hazard restricted area), the indicator light 114 may be operated to produce a red light.When the miter saw 100 is operating under safe conditions (for example, when no hand is detected in a restricted area), the indicator light 114 may be operated to produce a green light. The indicator light 114 may be a light-emitting diode (LED), an LED strip, an LED array, or other suitable light. The camera 110, the boom 112, and the indicator light 114 may together form a camera assembly 116. One or more such camera assemblies 116 may be connected to the miter saw 100, such as one camera assembly 116 on each side of the saw blade of the miter saw 100.

[0011] As in Fig. 1C, in some examples, two cameras 110a, 110b are coupled to the compound miter saw 100. In the example shown, a first camera 110a is coupled or attached to one side of the saw blade of the compound miter saw 100, and a second camera 110b is coupled or attached to the other side of the saw blade of the compound miter saw 100. In some examples, the cameras 110a, 110b may be arranged near the saw blade of the compound miter saw 100 so that the field of view of the cameras 110a, 110b can be aligned along the length of the saw blade. Arranging the cameras 110a, 110b in close proximity to the saw blade may help reduce ambiguities along the depth direction of the cameras 110a, 110b.

[0012] An example of a restricted area 150 for a miter saw 100 is shown in Fig. 1D. In this example, the restricted area 150 is generally defined by a volume extending around the saw blade of the miter saw 100. The restricted area 150 may consist of a single zone, a single area, a single volume, or may be composed of multiple restricted areas. For example, a warning zone of the restricted area 150 (e.g., a warning restricted area 152) may be defined as a zone in which there is an increased risk of injury to the user, but no immediate danger of injury. Similarly, a danger zone of the restricted area (e.g., a danger restricted area 154) may be defined as a smaller zone in which there is an immediate risk of injury to a user.The danger zone may be at least partially contained within the warning area, such that the danger exclusion zone 154 may be smaller than and at least partially contained within the warning exclusion zone. As described in more detail below, the volume of the exclusion zone 150 (or the warning exclusion zone 152 and the danger exclusion zone 154) is projected onto a plane, such as the two-dimensional (2D) camera plane 156 associated with one or more of the cameras 110, thereby forming a projected exclusion zone 158. A detected hand object 160 is tracked, and when the detected hand object 160 intersects the projected exclusion zone 158, one or more safety actions may be executed.

[0013] As a non-limiting example, when the handheld object 160 intersects with a projected warning zone, one or more safety actions may include issuing an audible (or audible) warning to the user, issuing a visual warning to the user, or issuing both an audible and visual warning to the user. The audible warning may include an intermittent audible warning (e.g., a series of tones, beeps, etc.) played by a speaker of the miter saw 100. The visual warning may include generating a yellow light via the indicator light 114. When the handheld object 160 intersects with a projected hazard zone, one or more safety actions may be executed, such as stopping the saw blade (e.g., by mechanically braking the saw blade, electronically braking the saw blade, etc.).), issuing an audible warning to the user, issuing a visual warning to the user, or combinations thereof. The audible warning may consist of a constant tone played by a speaker on the miter saw 100. The visual warning may include generating a red light via the indicator light 114.

[0014] As in Fig. 1E, in one non-limiting example, a warning exclusion area 152 may correspond to a volume extending through the kerf plate of the miter saw 100. As shown in Fig. 1F, a danger exclusion zone 154 may, in one non-limiting example, correspond to a volume extending through the slot in the kerf plate through which the saw blade may pass when cutting a workpiece, or may otherwise be defined by a cutting plane of the saw blade. The size and shape of each exclusion zone 150 (e.g., warning exclusion zone 152, danger exclusion zone 154) may be predetermined, set by the user, and / or dynamically adjusted during operation of the compound miter saw 100. For example, the width of each exclusion zone 150 may be set by the user and / or dynamically adjusted during operation of the compound miter saw 100 (e.g., by changing the miter angle and / or the cant angle of the saw blade).

[0015] Fig. 2A and Fig. 2B illustrate a table saw 200 according to some embodiments. Like the compound miter saw 100, the table saw 200 is an example of a sawing machine in which the hand detection and control techniques described herein may be employed. The table saw 200 includes one or more cameras 210 coupled to the table saw 200. In the depicted example, two cameras 210a, 210b are coupled to the table saw 200 via a boom 212. The cameras 210a, 210b are coupled to or otherwise attached to the boom 212, which is coupled to the frame or housing of the table saw 200. The boom 212 may be a relatively thin mount for the cameras 210a, 210b. For example, the boom 212 may have a similar or smaller width than the riving knife 232 of the table saw 200 so that the material can be guided past the boom 212 when a workpiece is cut with the blade 230 of the table saw 200.The cameras 210a, 210b may be coupled or otherwise attached to the boom 212 so that they are mounted obliquely and laterally from the saw blade 230 to bypass the saw blade handguard. Although in . Fig. 2A or Fig. 2B, an indicator light may also be coupled to the boom 212, similar to the one shown in Fig. 1B, the indicator light 114.

[0016] Similar to the miter saw 100, the table saw 200 may also have one or more restricted areas with respect to the saw blade 230. For example, a warning restricted area may be defined by a volume extending through the kerf plate of the table saw 200, and a danger restricted area may be defined by a volume extending through the slot in the kerf plate or otherwise through the cutting plane of the table saw 200.

[0017] Fig. 3 shows a block diagram of an exemplary sawing machine 300. The sawing machine 300 may be a cross-cut saw (for example, the one shown in Fig. 1A), a table saw (for example the one shown in Fig. 2A) or another suitable type of sawing machine. The block diagram in Fig. 3 therefore applies to examples of the miter saw 100 and the table saw 200, as well as to other types of sawing machines. In other examples, the sawing machine 300 is embodied as a different type of sawing machine than the illustrated examples of the miter saw 100 and the table saw 200.

[0018] The sawing machine 300 includes an electronic control unit 320, a power source 352 (e.g., a battery, a portable power pack, and / or an electrical outlet), and the like. In the illustrated embodiment, the sawing machine 300 further includes a wireless communication device 360. In other embodiments, the sawing machine 300 may not include a wireless communication device 360.

[0019] The electronic control unit 320 may include an electronic processor 330 and a memory 340. The electronic processor 330 and the memory 340 may communicate via one or more control buses, data buses, etc., which may include a device communication bus 354. The control and / or data buses are Fig. 3 for illustrative purposes. The use of one or more control and / or data buses for interconnection and communication between the various modules, circuits, and components is known to those skilled in the art.

[0020] The electronic processor 330 may be configured to communicate with the memory 340 to store data and retrieve stored data. The electronic processor 330 may be configured to receive instructions 342 and data from the memory 340 and, among other things, execute the instructions 342. In particular, the electronic processor 330 executes the instructions 342 stored in the memory 340. In this way, the electronic control unit 320, coupled to the electronic processor 330 and the memory 340, may be configured to perform the methods described herein (e.g., one or more aspects of the process 00 of Fig. 4; one or more aspects of process 00 of Fig. ; one or more aspects of process 00 of Fig. ; one or more aspects of process 00 of Fig. ; and / or one or more aspects of process 00 of Fig.). 4557710101111

[0021] In some examples, electronic processor 330 includes one or more electronic processors. For example, as shown in the drawing, electronic processor 330 includes a central processor 332 and a machine learning (ML) processor 334. In other examples, the functions of central processor 332 and / or ML processor 334 are combined into a single processor or distributed among additional processors.

[0022] Additionally or alternatively, the electronic processor 330 (or the central processor 332 or ML processor 334) may include one or more AI (artificial intelligence) accelerator cores. The AI ​​accelerator core may include specialized processing units (e.g., arithmetic logic units (ALUs), floating-point units (FPUs), etc.) configured to perform the specific operations associated with training and / or inference of neural networks. As a non-limiting example, the processing units of the AI ​​accelerator core may be organized to enable parallel processing, allowing multiple computations to be executed simultaneously.

[0023] The memory 340 may include read-only memory ("ROM"), random access memory ("RAM"), other non-transferable, computer-readable media, or a combination thereof. As described above, the memory 340 may include instructions 342 that the electronic processor 330 can execute. The instructions 342 may include software that can be executed by the electronic processor 330 to enable the electronic controller 320 to, among other things, receive data and / or commands, send data, control the operation of the sawing machine 300, and the like.The instructions 342 may include, for example, software executable by the electronic processor 330 to enable the electronic controller 320 to implement, among other things, the various functions of the electronic controller 320 described herein, including performing hand detection and controlling the operation of the sawing machine 300. The software may include, for example, firmware, one or more applications, program data, filters, rules, one or more program modules, and other executable instructions.

[0024] As shown, memory 340 may also store a machine learning (ML) model 344. ML model 344 may be a pre-trained machine learning model (e.g., a neural network or other machine learning model trained for object recognition) executed by electronic processor 330. In some examples, ML processor 334 may execute ML model 344 to perform hand recognition for sawing machine 300, as described herein. In other words, ML processor 334 may serve as a dedicated processor to execute ML model 344 to recognize hands, as described herein. In such examples, central processor 332 may perform other controls for sawing machine 300, such as activating and deactivating a motor (e.g., motor 372).

[0025] The electronic processor 330 is configured to, among other things, retrieve and execute instructions 342 related to the control operations and methods described herein from the memory 340. The electronic processor 330 is also configured to store data in the memory 340, including usage data (e.g., usage data of the sawing machine 300), maintenance data (e.g., maintenance data of the sawing machine 300), feedback data, power source data, sensor data (e.g., sensor data of the sawing machine 300), environmental data, operator data, location data, and the like.

[0026] In some examples, the electronic processor 330 may receive instructions 342 from the memory 340 that include settings or configurations for the size and shape of one or more restricted areas, the type, duration, and / or volume of audible warnings, the type of visual warnings, etc. These settings for the sawing machine may be received and / or updated wirelessly via an application (such as an app), customized in the firmware (such as programmed for specific users during manufacturing or with firmware updates), customized via inputs directly at the machine tool device (such as via a button, a switch, a series of user interface actions, controls on a screen, and the like).

[0027] Power source 352 may be an AC power source or a DC power source that may be in electrical communication with one or more outlets (e.g., AC or DC outlets). Power source 352 may be an AC power source, such as a conventional wall outlet, or a DC power source, such as a battery.

[0028] In some examples, power source 352 may include a battery interface and a selectively attachable and removable power tool battery. The battery interface may include one or more power ports, and in some cases, one or more communication ports, connected to the corresponding power and / or communication ports of the power tool battery. The power tool battery may include one or more battery cells of various chemistries, such as lithium-ion (Li-ion), nickel-cadmium (Ni-Cad), and the like. The power tool battery may also be selectively locked and unlocked (e.g., with a spring-loaded locking mechanism) to the sawing machine 300 to prevent inadvertent detachment. The power tool battery may also include an electronic controller (e.g., battery controller) having a processor and memory.The battery control unit may be configured similarly to the electronic control unit 320. The battery control unit may be configured to control the charging and discharging of the battery cells and / or to communicate with the electronic control unit 320.

[0029] In other examples, the sawing machine 300 may be corded and the power source 352 may include a wired power interface to receive external power from a wall outlet or the like (e.g., AC power).

[0030] In some embodiments, the sawing machine 300 may also include a wireless communication device 360. In these embodiments, the wireless communication device 360 ​​is coupled to the electronic controller 320 (e.g., via the device communication bus 354). The wireless communication device 360 ​​may include, for example, a radio transceiver and antenna, memory, and an electronic processor. In some examples, the wireless communication device 360 ​​may also include a GNSS receiver configured to receive signals from GNSS satellites, land-based transmitters, etc.The radio transmitter / receiver and antenna cooperate to send and receive wireless messages to and from an external device (e.g., a smartphone, a tablet computer, a cell phone, a laptop, a smartwatch, a headset, a heads-up display, virtual reality ("VR") glasses, augmented reality ("AR") glasses, a security camera, a webcam, etc.), one or more additional machine tool devices (e.g., a machine tool battery charger, a machine tool battery pack, a machine tool, a work light, a machine tool battery adapter, and other devices used in conjunction with machine tool battery chargers, machine tool batteries, and / or machine tools), a server, and / or the electronic processor of the wireless communication device 360.The memory of the wireless communication device 360 ​​stores instructions executed by the electronic processor of the wireless communication device 360 ​​and / or may store data related to communication between the sawing machine 300 and the external device, one or more additional machine tool devices, and / or the server.

[0031] The electronic processor of the wireless communication device 360 ​​controls wireless communication between the sawing machine 300 and the external device, one or more additional machine tool devices, and / or the server. For example, the electronic processor of the wireless communication device 360 ​​buffers incoming and / or outgoing data, communicates with the electronic processor 330 of the sawing machine 300, and determines the communication protocol and / or wireless communication settings.

[0032] In some embodiments, the wireless communication device 360 ​​is a Bluetooth® controller. The Bluetooth® controller communicates with the external device, one or more additional machine tool devices, and / or the server using the Bluetooth® protocol. In such embodiments, the external device, one or more additional machine tool devices, and / or the server and the sawing machine 300 are within communication range (i.e., in proximity) of each other while exchanging data. In other embodiments, the wireless communication device 360 ​​communicates using other protocols (e.g., Wi-Fi®, cellular protocols, a proprietary protocol, etc.) over another type of wireless network.For example, the wireless communication device 360 ​​may be configured to communicate via Wi-Fi® over a wide area network such as the Internet or a local area network, or via a piconet (e.g., via infrared or NFC). Communication via the wireless communication device 360 ​​may be encrypted to protect the data exchanged between the sawing machine 300 and the external device, one or more additional machine tool devices, and / or the server from third parties.

[0033] In some embodiments, the wireless communication device 360 ​​exports usage data, other machine tool device data, and / or other data, as described above, from the machine tool device 102 (such as from the electronic processor 330).

[0034] In some embodiments, the wireless communication device 360 ​​may be located in a separate housing along with the electronic control unit 320 or another electronic control unit, which separate housing may be selectively attached to the machine tool 102. For example, the separate housing may be attached to an exterior surface of the machine tool 102 or inserted into a receptacle of the machine tool 102. Accordingly, the wireless communication capabilities of the machine tool 102 may be partially housed on a selectively attachable communication device rather than being integrated into the machine tool device 102.Such optionally attachable communication devices may include electrical connectors that engage mutual electrical connectors of the machine tool device 102 to enable communication between the respective devices and enable the machine tool device 102 to power the optionally attachable communication device. In other embodiments, the wireless communication device 360 ​​may be integrated into the machine tool device 102. In some embodiments, the wireless communication device 360 ​​is not included in the machine tool device 102.

[0035] The electronic components 370 include a motor 372, one or more sensors 374, one or more cameras 376, and one or more feedback devices 378.

[0036] In some examples, motor 372 is configured to rotate saw blade 380. Electronic components 370 may also include additional sensors and circuitry for controlling motor 372. For example, electronic components 370 may include an inverter bridge controlled by pulse-width modulated signals (generated by electronic controller 320) to drive motor 372. Motor 372 may, for example, be a brushed or brushless motor.

[0037] The sensor(s) 374 may comprise an accelerometer, a gyroscope, a magnetometer, an angle encoder, or other sensing device configured to provide an indication of the orientation thereof. The sensor(s) 374 may be mounted on an arm of the sawing machine 300 or otherwise connected to a movable component of the sawing machine 300 that tracks the movement of the saw blade 380 (such as the translation of the saw blade 380, the change in orientation of the saw blade 380, the rotation of the saw blade 380 during operation, etc.). Accordingly, the output of the sensor(s) 374 may indicate a position and / or orientation of the saw blade 380, including a bevel angle, a skew angle, or both. The position and / or orientation of the saw blade 380 may be referred to as a posture of the saw blade 380.The output of the sensor(s) 374 may be forwarded to the electronic controller 320 (e.g., the electronic processor 330 and / or the ML processor 334) to determine the position, orientation, and / or attitude of the saw blade 380 and / or the sawing machine 300.

[0038] In some examples, the sensor(s) 374 may additionally include one or more voltage sensors or voltage measurement circuits, current sensors or current measurement circuits, temperature sensors or temperature measurement circuits, pressure sensors or pressure measurement circuits (e.g., a barometer), or the like. The sawing machine 300 may also include connections (e.g., wired or wireless connections) for external sensors.

[0039] The cameras 376 may capture and output image data to the electronic controller 320 (e.g., the electronic processor 330 and / or ML processor 334), which may serve as input to the ML model 344 executing on the electronic processor 330. The cameras 376 may include a left and a right camera, each mounted on opposite (left and right) sides of the saw blade 380. In the Fig. 1A to 1C, the cameras 376 may, for example, comprise the cameras 110a and 110b, and in the case of the cross-cut saw 100 shown in Fig. 2A to 2B, the cameras 376 may include the cameras 210a and 210b.

[0040] The cameras 376 may be any suitable camera for recording or otherwise capturing image frames of a scene. The cameras 376 may capture individual image frames or a series of image frames, or record a video stream of the scene. To capture a wide field of view, the cameras 376 may be equipped with a wide-angle lens, a fisheye lens, or the like.

[0041] The feedback devices 378 may be controlled by the electronic control unit 320 (e.g., the electronic processor 330, the central processor 332, and / or the ML processor 334) to provide feedback to a user based on an output from the electronic control unit 320 regarding whether an unsafe operating condition exists (e.g., when a hand is detected within one or more restricted areas). The feedback devices 378 may include lights (e.g., LEDs), speakers, or both. As described above, the lights may include an indicator light that provides a visual warning to the user when the electronic control unit 320 detects an unsafe operating condition. Similarly, the speaker may provide an audible warning when the electronic control unit 320 detects an unsafe operating condition.

[0042] The electronic components 370 may also include one or more switches (e.g., for initiating and terminating operation of the sawing machine 300), for waking up the sawing machine, or the like.

[0043] In some embodiments, the sawing machine 300 may include one or more inputs 390 (e.g., one or more buttons, switches, and the like) coupled to the electronic controller 320 that allow the user to select a mode of the sawing machine 300 (e.g., to place the sawing machine 300 into a guard mode or otherwise cause the electronic controller 320 to operate the cameras 376 to begin monitoring for hands in a restricted area). In some embodiments, the input 390 includes a user interface (UI) element, such as an actuator, button, switch, knob, dial, touchscreen, and the like, that allows user interaction with the sawing machine 300. The inputs 390 may, for example, include a UI element that allows adjustment of the size and / or shape of one or more restricted areas.A UI element may, for example, be a rotary dial, a gyroscope, a touchscreen, or the like, with which the user can adjust the width of the projected area of ​​a restricted area. In some embodiments, the sawing machine 300 may provide a visual indication of the size of the projected restricted area, for example, by projecting light onto the work surface of the sawing machine indicating the size and shape of the projected restricted area, or by providing such a visual indication on a display device.

[0044] In some embodiments, the sawing machine 300 may include one or more outputs 392 that are also connected to the electronic control unit 320. The output(s) 392 may receive control signals from the electronic control unit 320 to communicate data or information to a user or to generate other visual, audible, or other outputs. For example, the output(s) 392 may generate a visual signal to provide the user with information about the operation or condition of the sawing machine 300. The output(s) 392 may, for example, include LEDs or a screen and generate various signals that indicate, for example, an operating state or mode of the sawing machine 300, an abnormal condition or event detected during operation of the sawing machine 300, and the like.The output(s) 392 may, for example, indicate the condition or status of the sawing machine 300, an operating mode of the sawing machine 300, and the like.

[0045] Referring to Figure 4, a flowchart is shown describing the steps of an exemplary method for controlling the operation of a sawing machine.

[0046] The method includes waking the sawing machine or otherwise initiating the detection of hands in one or more restricted areas of the sawing machine, as specified in step 02. As described above, waking the sawing machine may involve actuating a UI element (e.g., a button, switch, or similar) that wakes up the sawing machine's cameras to capture image data (e.g., images, videos) of the restricted areas and other work areas in the cameras' field of view.4

[0047] Using the image data captured by the cameras, the sawing machine monitors the work area to detect whether a hand is entering one or more restricted areas, as specified in step 4. As described in more detail below, the sawing machine's electronic control unit can receive the image data from the cameras and process the image data to detect whether a hand is present in the scene.4

[0048] If a hand is detected and identified in a restricted area, the sawing machine performs one or more safety actions in response to this unsafe operating condition, as specified in step 6. As described above, the safety actions may include generating an audible warning, a visual warning, stopping the saw blade, or combinations thereof.4

[0049] Referring to Fig. 5, a flowchart is shown describing the steps of an exemplary method for detecting an unsafe operating condition of a sawing machine.

[0050] The method includes receiving image data with the electronic control unit 320 (e.g., the electronic processor 330), as indicated in step 02. Generally, the image data is images or videos captured by the cameras 376. The image data may be received by the electronic processor 330 from the cameras 376. Additionally or alternatively, the image data may be received from the memory 340 by the electronic processor 330. Additionally or alternatively, receiving the image data may include obtaining such data with the cameras 376 and transmitting or otherwise communicating the image data to the electronic control unit 320.5

[0051] The images and / or videos captured by cameras 376 cover a field of view that includes at least a portion of one or more restricted areas. Thus, the image data includes at least a portion of the restricted area. An example of images captured by cameras 376 is shown in Fig. 6. In the example shown, a first image 602 is captured with a first of the cameras 376 (e.g., camera 110a) and a second image 604 is captured with a second of the cameras 376 (e.g., camera 110b). For illustration, an example of a projected hazard exclusion zone 606 is highlighted in the first image 602 and an example of a projected warning exclusion zone 608 is highlighted in the second image 604. A detected hand object 610 (e.g., a bounding box in this example) is also visible in the first image 602.

[0052] In some examples, the image data may be stored for later use as training data for retraining, fine-tuning, or otherwise updating a machine learning model. For example, the image data may be stored in memory 340 of electronic control unit 320. The image data may then be read from memory 340 via a wired or wireless connection. For example, a user may connect an external device to sawing machine 300 via a wired connection (e.g., a USB cable) and read the image data from memory 340. Additionally or alternatively, the image data may be transmitted to an external device and / or a server via wireless communication device 360.In this way, the image data collected during operation of the sawing machine 300 can be used to create a training dataset or to supplement an existing training dataset. As described further below, the image data can be annotated to identify areas of the images that contain a hand, and this annotated image data can be stored as part of the training dataset.

[0053] A trained machine learning model is then accessed by the electronic control unit 320, as indicated in step 04. Generally, the machine learning model is trained on training data or has been trained to recognize hands in image data. The machine learning model may include one or more neural networks. 5

[0054] As a non-limiting example, the machine learning model may include a one-look-only-once (YOLO) object detection model. In general, a YOLO model is a one-stage object detection model that partitions an input image into a grid and directly predicts bounding boxes and class probabilities. The YOLO model can process an entire image in a single pass. As another example, the machine learning model may include a single-shot multi-frame detector (SSD) model. In general, an SSD model is another one-stage object detection model that predicts bounding boxes and class scores at multiple scales. The SSD model can use feature maps from different levels to detect objects of different sizes.

[0055] As another example, the machine learning model may include a Faster R-CNN region-based convolutional neural network (Faster R-CNN) or a Mask R-CNN model. Generally, a Faster R-CNN model is a two-stage object detection model that uses a region proposal network to generate region proposals, followed by an object detection network. A Mask R-CNN model is an extension of Faster R-CNN that adds an additional branch for pixel-level segmentation, allowing the model to both detect objects and generate detailed masks for each object instance.

[0056] Other object detection models, including other one-stage object detection models and / or other two-stage object detection models, can also be used for hand detection.

[0057] Accessing the trained machine learning model may include accessing model parameters (e.g., weights, biases, or both) that were optimized or otherwise estimated by training the machine learning model with training data. In some cases, retrieving the machine learning model may also include retrieving, constructing, or otherwise accessing the model architecture to be implemented. For example, data about the layers in a neural network architecture (e.g., number of layers, type of layers, arrangement of layers, connections between layers, hyperparameters for the layers) may be retrieved, selected, constructed, or otherwise retrieved.

[0058] An artificial neural network generally comprises an input layer, one or more hidden layers (or nodes), and an output layer. Typically, the input layer contains as many nodes as the number of inputs provided to the artificial neural network. The number (and type) of inputs provided to the artificial neural network can vary depending on the specific task of the artificial neural network.

[0059] The input layer is connected to one or more hidden layers. The number of hidden layers varies and can depend on the specific task of the artificial neural network. Furthermore, each hidden layer can have a different number of nodes and be connected to the next layer in different ways. For example, each node of the input layer can be connected to every node of the first hidden layer. The connection between each node of the input layer and each node of the first hidden layer can be assigned a weight parameter. In addition, each node of the neural network can also be assigned a bias value. In some configurations, not every node of the first hidden layer may be connected to every node of the second hidden layer.This means that there may be some nodes in the first hidden layer that are not connected to all of the nodes in the second hidden layer. The connections between the nodes in the first hidden layer and the second hidden layer are each assigned different weight parameters. Each node in the hidden layer is generally associated with an activation function. The activation function determines how the hidden layer should process the inputs received from the input layer or a previous input or hidden layer. These activation functions can vary and depend on the type of task associated with the artificial neural network and the specific type of hidden layer implemented.

[0060] Each hidden layer can perform a different function. For example, some hidden layers may be convolutional layers, which in some cases can reduce the dimensionality of the inputs. Other hidden layers may perform statistical functions such as max-pooling, which reduces a set of inputs to the maximum value, an averaging layer, batch normalization, and other such functions. In some of the hidden layers, each node is connected to every node of the next hidden layer, which can then be called dense layers. Some neural networks that contain more than, say, three hidden layers can be called deep neural networks.

[0061] The last hidden layer in the artificial neural network is connected to the output layer. Similar to the input layer, the output layer typically has the same number of nodes as the possible outputs. As one example, the machine learning model may output the hand detection data as a bounding box (e.g., a center coordinate of the bounding box, a height of the bounding box, a width of the bounding box). As another example, the machine learning model may output the hand detection data as a group of pixels (e.g., a mask) corresponding to a hand.

[0062] The image data is then input to the machine learning model to perform hand recognition, as indicated in step 06. For example, the electronic processor 330 may execute instructions to input the image data to the ML model 344. As described above, in some embodiments, the electronic processor 330 may include a separate ML processor 334 and / or one or more dedicated AI accelerator cores that may implement the ML model 344. Generally, the machine learning model generates the detected hand data as output. As described above, the detected hand data may include a bounding box centered on each detected hand, a mask indicating a group of pixels corresponding to a detected hand, or the like.In this way, the electronic processor 330 (or the ML processor 334) analyzes the image data using the ML model 344 to determine whether a portion of a hand is present in the image data. The hand recognition data may be stored in the memory 340 or in a cache memory of the electronic processor 330. 5.

[0063] Based on the captured hand data, the electronic processor 330 determines whether an unsafe operating condition exists by determining whether a hand is located within one or more restricted areas of the sawing machine, as specified in step 8. As described above and further explained below, an unsafe operating condition can be determined when a hand (represented by the hand detection data) intersects a restricted area (represented by a projected restricted area). In this way, the electronic processor uses the ML model 344 to detect whether a portion of a hand is present in the image data and then determines whether this detected hand is located within the restricted area. 5

[0064] As described above, a two-stage object detection model, such as an optical flow detection model, may additionally or alternatively be used to detect an unsafe operating condition of the sawing machine. As a non-limiting example, the machine learning model may be used to perform hand detection in step 06, and an optical flow detection model may additionally be used to detect other movements between the image frames of the image data received in step 02. For example, the machine learning model may be used to generate hand detection data in step 06, which is then used to determine whether the user's hand intersects a restricted area, while the optical flow detection model may be used to generally detect fast-moving objects within the image data.In this way, the optical flow model can be used to detect rapid movements anywhere in the image frame, thereby covering other unsafe operating conditions that may cause injury, such as when the hand is drawn into the sawblade or otherwise into the restricted area, when material and / or dust is violently ejected from the sawing machine, and / or when environmental factors are present.555.

[0065] Thus, in some embodiments, an optical flow model may be used to provide a second level of unsafe operating condition detection, which may additionally be used to prevent user injury. The image data is input to the optical flow model to generate optical flow detection data as output. The optical flow detection data may include an optical flow field, as described below, or additional data calculated, derived, estimated, or otherwise generated from the optical flow field.Examples of additional data that can be generated from the optical flow field include images or parameters generated from the optical flow field, such as images indicating motion regions, motion masks, images classifying different motion regions based on displacement and / or velocity, bounding boxes encompassing regions of detected motion, and so on.

[0066] If an unsafe operating condition is detected, the electronic processor 330 may generate an unsafe operating condition signal, as indicated in step 10. The unsafe operating condition output may include a signal generated by the electronic processor 330 to instruct the electronic controller 320 to control the operation of one or more feedback devices 378, the motor 372 of the sawing machine 300, or other such signals for controlling the electronic controller 320 to perform a safety action. As previously mentioned, in some cases, the unsafe operating condition may be determined by detecting whether the user's hand moves into a restricted area.Additionally or alternatively, the unsafe operating condition can be identified based on the output of an optical flow model indicating object movement within the image frame that could lead to potential injury to the user (e.g., unsafe movement towards the saw blade, unsafe movement of an object ejected from the sawing machine, unsafe environmental conditions).5.

[0067] Fig. 7 shows a flowchart describing the steps of an exemplary method for determining whether a detected hand is located in a restricted area of ​​a sawing machine.

[0068] The method includes receiving sensor data with the electronic processor 330, as indicated in step 02. Generally, the sensor data includes data acquired with one or more of the sensors 374. The sensor data may include inertial sensor data such as accelerometer data, gyroscope data, magnetometer data, or the like. The sensor data may be received by the electronic processor 330 from the sensors 374. Additionally or alternatively, the sensor data may be received by the electronic processor 330 from the memory 340. Additionally or alternatively, receiving the sensor data may include obtaining such data with the sensors 374 and transmitting or otherwise communicating the sensor data to the electronic control unit 320.7

[0069] The sensor data is processed by the electronic processor 330 to determine the position of the sawing machine, as indicated in step 04. The position of the sawing machine may include a position of the sawing machine 300, an orientation of the sawing machine 300, or both. As shown in Fig. As shown in Figure 8, the position of the sawing machine 300 can be estimated, for example, based on the sensor data and the known geometry of the saw. In this way, the electronic controller 310 can receive an indication of the orientation of the saw blade of the sawing machine 300 from the sensor data received from the sensors 374.7

[0070] Based on the estimated position of the sawing machine 300, the electronic processor 330 generates restricted area data defining one or more restricted areas of the sawing machine 300, as indicated in step 06. The restricted area data may include one or more volumes defining one or more restricted areas of the sawing machine 300. For example, the restricted area data may include a first volume defining a warning restricted area and a second volume defining a hazard restricted area. As described above, in one non-limiting example, the warning restricted area may include a volume extending through a kerf plate of the sawing machine 300 (e.g., as in the example of Fig. 1E). Similarly, in one non-limiting example, the danger exclusion zone may comprise a volume extending through the cutting plane of the saw blade of the sawing machine 300 (e.g., as in the example of Fig. 1F). The data of the restricted area can be adjusted by the user via one or more UI elements of the sawing machine 300, as described above. 7

[0071] The electronic processor 330 then projects the restricted areas in the restricted area data onto a 2D plane (e.g., the 2D camera plane), as indicated in step 08. Projecting a restricted area onto a 2D plane may involve processing intrinsic and extrinsic parameters of the cameras 376. Processing the intrinsic parameters may involve, for example, removing image distortions from the image data (e.g., when using a fisheye lens, rectifying the image field onto a 2D plane) so that the hand detection data is mapped onto an undistorted camera plane. Processing the extrinsic parameters may involve using the estimated pose of the sawing machine 300 to determine the position and orientation of the cameras 376, thus facilitating the projection of the volume associated with each restricted area onto the undistorted 2D camera plane.7

[0072] The electronic processor 330 then determines whether a detected hand in the hand detection data intersects one or more of the projected exclusion zones, as indicated in step 10. If a detected hand object (e.g., a bounding box, a mask, etc.) in the hand detection data intersects a projected exclusion zone, the processor 330 determines that an unsafe operating condition exists and generates the appropriate output for a less safe operating condition to execute the associated safety action. As indicated in Fig. 9A, for example, if a detected hand-held object 902 is determined to be intersecting with the warning area 952, the electronic processor 330 may control the electronic control unit 320 to execute safety actions associated with a warning condition (e.g., generating a corresponding audible or visual warning, or both). On the other hand, as shown in Fig. 9B, control the electronic control unit 320 to execute safety actions associated with a hazardous condition (e.g., stopping the saw blade in addition to generating an appropriate audible warning, visual warning, or both) when a detected hand-held object 902 is identified as intersecting a hazardous exclusion zone 954. 7

[0073] Fig. 10 shows a flowchart describing the steps of an exemplary method for training a machine learning model for hand recognition.

[0074] In general, the machine learning model can implement any number of different model architectures suitable for performing object detection. A non-limiting example is an artificial neural network that enables machine learning. The artificial neural network can be implemented as a convolutional neural network, a residual neural network, or the like.

[0075] The method includes accessing training data with a computer system, as indicated in step 02. Accessing the training data may include retrieving such data from a memory or other suitable data storage device or medium. Alternatively, accessing the training data may also include obtaining such data and transmitting or otherwise communicating it to the computer system. The training data may, for example, include images captured with one or more cameras. As described above, in some examples, the training data may also include image data captured during operation of a sawing machine 300.10

[0076] In general, the training data may include images, each depicting one or more hands. In some embodiments, the training data may include images that have been annotated or tagged (e.g., with patterns, features, or characteristics indicative of one or more hands in the images, and the like). For example, the training data may include annotated images of hands from videos, such as the 100 Days of Hands (100DOH) dataset. Additionally or alternatively, the training data may include annotated images of first-person perspectives of the hands of two interacting individuals, such as the EgoHands dataset. Additionally or alternatively, the training data may include annotated images of hands captured during operation of a sawing machine under safe and unsafe operating conditions.As described above, in some cases the training data may include images captured during operation of the sawing machine 300, which may be stored in the memory 340 for later use as training data.

[0077] The method may include compiling training data from image data using a computer system. This step may include compiling the image data into a suitable data structure on which the machine learning model can be trained. Compiling the training data may include compiling image data, segmented image data, and other relevant data. For example, compiling the training data may include generating labeled data and incorporating the labeled data into the training data. The labeled data may include image data, segmented image data, or other relevant data that has been labeled as belonging to one or more different classifications or categories.The labeled data may include, for example, image data and / or segmented image data in which one or more regions of the images have been labeled as containing or otherwise depicting a hand.

[0078] A machine learning model is then trained on the training data, as specified in step 4. In general, the machine learning model can be trained by optimizing the model parameters (e.g., weights, biases, or both) based on minimizing a loss function. As a non-limiting example, the loss function can be a mean square error loss function.10

[0079] Training a neural network may involve initializing the neural network, for example, by calculating, estimating, or otherwise selecting the initial network parameters (e.g., weights, biases, or both). During training, an artificial neural network receives the inputs for a training example and generates an output using the bias for each node and the connections between each node and the corresponding weights. For example, training data may be input to the initialized neural network, which generates the output as recognized hand data. The artificial neural network then compares the generated output with the actual output of the training example to evaluate the quality of the hand recognition data. The hand recognition data may, for example, be passed to a loss function to calculate an error.The current neural network can then be updated based on the calculated error (e.g., using backpropagation techniques based on the calculated error). The current neural network can be updated, for example, by updating the network parameters (e.g., weights, biases, or both) to minimize the loss according to the loss function. Training continues until a training condition is met. The training condition can be, for example, that a predetermined number of training examples are used, that a minimum accuracy threshold is reached during training and validation, that a predetermined number of validation iterations are completed, and the like.If the training condition is met (for example, by determining whether an error threshold or other stopping criterion has been met), the current neural network and its associated network parameters represent the trained neural network. Various types of training procedures can be used to adjust the bias values ​​and the weights of the node connections based on the training examples. Training procedures include, for example, gradient descent, Newton's method, conjugate gradient, quasi-Newton, Levenberg-Marquardt, and others.

[0080] The artificial neural network can be constructed or otherwise trained based on training data using one or more different learning techniques, such as supervised learning, unsupervised learning, reinforcement learning, ensemble learning, active learning, transfer learning, or other suitable learning techniques for neural networks. In supervised learning, for example, a computer system is presented with sample inputs and their actual outputs (e.g., categorizations). In these cases, the artificial neural network is configured to learn a general rule or model that maps the inputs to the outputs based on the provided sample input-output pairs.

[0081] As another example, the machine learning model may be a pre-trained machine learning model, and training the pre-trained machine learning model on the training data may include retraining, fine-tuning, or otherwise updating the pre-trained machine learning model using the training data. For example, the machine learning model may be a pre-trained neural network that is retrained to recognize hands using transfer learning. In such cases, one or more of the last layers of the neural network may be removed and the neural network retrained using the training data. Using this approach, a pre-trained object detection model (e.g., a YOLO object detection model) can be retrained specifically to recognize hands.

[0082] The trained machine learning model is then stored for later use, as specified in step 6. Storing the machine learning model may include storing model parameters (e.g., weights, biases, or both) calculated or otherwise estimated by training the machine learning model on the training data. If the machine learning model is a neural network, storing the trained machine learning model may also include storing the specific neural network architecture to be implemented. For example, data related to the layers in the neural network architecture (e.g., number of layers, type of layers, arrangement of layers, connections between layers, hyperparameters for the layers) may be stored. 10

[0083] As described above, the trained machine learning model may be stored in the memory 340 of the electronic control unit 320 of the sawing machine 300. For example, the trained machine learning model may be stored as an ML model 344 in the memory 340. The trained machine learning model may be communicated to the memory 340 via a wired or wireless connection. In some cases, the trained machine learning model may be stored in the memory 340 when the sawing machine 300 is manufactured. In other cases, the trained machine learning model may be communicated to the sawing machine 300 at a time after the sawing machine 300 has been manufactured. For example, the trained machine learning model may be communicated to the sawing machine 300 as part of a firmware or other update to the sawing machine 300.The ML model 344 may therefore be transmitted to the electronic control unit 320 (e.g., via the wireless communication device 360, via a wired connection) and stored in the memory 340. Additionally or alternatively, an updated ML model 344 may be transmitted to the electronic control unit 320 (e.g., via the wireless communication device 360, via a wired connection) and stored in the memory 340. The updated ML model 344 may include updated model parameters, an updated model architecture, or combinations thereof.

[0084] Referring to Fig. 11, a flowchart is shown describing the steps of an exemplary method for identifying an unsafe operating condition of a sawing machine based on processing image data using an optical flow recognition model.

[0085] The method includes receiving image data with the electronic control unit 320 (e.g., the electronic processor 330), as indicated in step 02. In general, the image data received in step 02 may be the same image data that the electronic control unit 320 received in step 02 of the Fig.02. Alternatively, the image data may also include additional images or videos captured by one or more cameras, such as camera 376 or other cameras in the work environment. The image data may therefore be received by the electronic processor 330 of the electronic control unit 320 from the cameras 376, additional cameras in the work environment, or additionally or alternatively from the memory 340 by the electronic processor 330. Additionally or alternatively, receiving the image data may include capturing such data with the cameras 376 and transmitting or otherwise communicating the image data to the electronic control unit 320.111155

[0086] An optical flow recognition model is then accessed by the electronic control unit 320 (e.g., the electronic processor 330), as indicated in step 04. As described, the optical flow recognition model processes the image data to generate an optical flow field from which the motion of objects within the imaged field of view can be determined. The optical flow model may implement any number of suitable optical flow algorithms, operations, or methods. As a non-limiting example, the optical flow model may employ a Lucas-Kanade method, a Horn-Shunck method, or a Farnebäck method. In further examples, the optical flow model may employ a dense optical flow method, a variational method, a sparse optical flow method, a pyramidal optical flow method, or the like.In some applications, the optical flow recognition model may be implemented using a machine learning model that is different from the machine learning model used for hand recognition. For example, the optical flow recognition model may include a machine learning model that implements an optical flow algorithm, such as a PWC-Net model, a FlowNet model, or the like. In use, a PWC-Net model receives consecutive frames of image data as input and generates a corresponding optical flow field as output.11.

[0087] The image data is then processed by the electronic processor 330 using the optical flow recognition model to generate an optical flow field, as indicated in step 06. In general, optical flow is the displacement of pixels between two consecutive image frames. Optical flow can be measured, for example, as the apparent motion of an object between two consecutive image frames, caused by the movements of the object itself, other objects in the image frames, and / or the camera. An optical flow recognition model is used to calculate the optical flow field, which is a two-dimensional vector field in which each vector is a motion vector indicating the movement of points from the first image frame to the second image frame. These motion vectors provide information about the displacement and speed of movement in the images.Using this optical flow field, objects moving between image frames can be detected.11.

[0088] The electronic processor 330 then determines whether an unsafe operating condition exists based on the optical flow field, as indicated in step 08. As a non-limiting example, detecting an unsafe operating condition based on the optical flow field may include processing the optical flow field with the electronic processor 330 to determine whether motion occurs between frames of the image data that could lead to an unsafe operating condition. For example, the optical flow field may be processed by the electronic processor 330 to determine whether motion vectors in the optical flow field indicate that an object is moving toward one of the restricted areas of the sawing machine. If the processor 330 determines that an unsafe operating condition exists, it generates the appropriate unsafe operating condition output to execute the associated safety action.For example, if the optical flow field indicates that an object is moving toward the warning exclusion zone 952, the electronic processor 330 may control the electronic controller 320 to perform safety actions (e.g., generating an appropriate audible warning, a visual warning, or both) associated with a warning condition. On the other hand, if the optical flow field indicates that an object is moving toward the hazard exclusion zone 954, the electronic processor 330 may control the electronic controller 320 to perform safety actions (e.g., stopping the saw blade in addition to generating an appropriate audible or visual warning, or both) associated with a hazardous condition. These actions may be performed in addition to, or alternatively to, the actions performed based on the hand detection-based model in the method of FIG.117 are executed.

[0089] As another example, the optical flow field may be processed by electronic processor 330 to determine whether motion vectors in the optical flow field indicate that an object is moving away from the saw at high speed (e.g., a piece of material is being ejected from the saw by kickback or the like). In these cases, the motion vectors may be analyzed to determine whether the direction of movement is toward an area where a user is likely to be located (e.g., an area behind or in line with the saw blade of the saw) and whether the speed of movement is above a safety threshold.If an object is moving toward the user at a low speed, an unsafe operating condition may not exist, but if the object is moving at a higher speed where impact could injure the user, an unsafe operating condition may exist. The electronic processor 330 can then control the electronic controller 320 to perform safety actions related to the unsafe operating condition (e.g., stopping the saw blade and / or generating an appropriate audible or visual warning, or both).

[0090] In further examples, the optical flow field may be processed by the electronic processor 330 to determine whether motion vectors in the optical flow field indicate an object moving in the environment of the sawing machine that could indicate an unsafe operating condition. For example, if the motion vectors in the optical flow field indicate significant movement near the area where the user can safely operate the sawing machine, then an unsafe operating condition may be detected. In this case, the unsafe operating condition may indicate that the environment around the user may not permit safe operation of the sawing machine.The electronic processor 330 may then control the electronic controller 320 to perform safety actions related to the unsafe operating condition (e.g., stopping the saw blade and / or generating an appropriate audible or visual warning, or both).

[0091] It is to be understood that the disclosure is not limited in its application to the details of construction and arrangement of components set forth in the following description or illustrated in the following drawings. The disclosure is capable of being practiced or carried out in other embodiments and in various ways. It is also to be understood that the phraseology and terminology used herein is for the purpose of description and should not be considered limiting. The use of "including," "comprising," or "having," and variations thereof, is intended to include the terms listed below and their equivalents, as well as additional terms. Unless otherwise specified or limited, the terms "attached," "connected," "held," and "coupled," and variations thereof, are broadly defined to include both direct and indirect attachments, connections, supports, and couplings.Furthermore, the terms “connected” and “coupled” are not limited to physical or mechanical connections or couplings.

[0092] Unless otherwise limited or defined, particular directions are described herein only by way of example with respect to particular embodiments or relevant illustrations. For example, references to "top," "front," or "back" are generally intended only to describe the orientation of such features relative to a frame of reference of a particular example or illustration. Accordingly, for example, an "upper" feature may be located below a "lower" feature in some arrangements or embodiments (and so on). Furthermore, references to particular rotational or other movements (e.g., counterclockwise rotation) are generally intended only to describe movement relative to a frame of reference of a particular example or illustration.

[0093] In some embodiments, including computer-based applications of methods according to the disclosure, these may be implemented as a system, method, apparatus, or article of manufacture using standard programming or design techniques to generate software, firmware, hardware, or any combination thereof to control a processor device (e.g., a serial or parallel processor chip, a single or multi-core chip, a microprocessor, a field-programmable gate array, a variety of combinations of a control unit, an arithmetic logic unit, and a processor register, etc.), a computer (e.g., a processor device operatively connected to a memory), or other electronically operated control device to perform the aspects described herein.Accordingly, for example, embodiments of the disclosure may be embodied as a series of instructions tangibly embodied on a non-transitory, computer-readable medium such that a processor device may execute the instructions based on reading the instructions from the computer-readable medium. Some embodiments of the disclosure may include (or utilize) a control device, such as an automation device, a computer including various computer hardware, software, firmware, etc., as described below. As specific examples, a control device may include a processor, a microcontroller, a field-programmable gate array, a programmable logic controller, logic gates, etc.and other typical components known in the art for performing suitable functions (e.g., memory, communication systems, power sources, user interfaces and other inputs, etc.). Also, functions performed by multiple components may be consolidated and performed by a single component. Likewise, functions described herein that are performed by one component may be performed by multiple components in a distributed manner. Furthermore, a component that performs a particular function may also perform additional functions not described herein. A device or structure that is "configured" in a particular way is at least configured in that way, but may also be configured in other ways not listed here.

[0094] The term "article of manufacture" as used herein is intended to include a computer program accessible from any computer-readable device, carrier (e.g., non-transitory signals), or media (e.g., non-transitory media). Examples of computer-readable media include, but are not limited to, magnetic storage devices (e.g., hard disk, floppy disk, magnetic stripe, etc.), optical disks (e.g., compact disk (CD), digital versatile disk (DVD), etc.), smart cards, and flash memory devices (e.g., card, thumb drive, etc.). In addition, a carrier wave may also be used to carry computer-readable electronic data, such as in sending and receiving electronic mail or in accessing a network such as the Internet or a local area network (LAN).Those skilled in the art will recognize that numerous modifications can be made to these configurations without departing from the scope or spirit of the claimed subject matter.

[0095] Certain operations of methods according to the disclosure, or of systems performing those methods, may be schematically illustrated in the drawings or otherwise discussed herein. Unless otherwise indicated or limited, the depiction of certain operations in the drawings in a particular spatial order does not necessarily require that those operations be performed in a particular order corresponding to the particular spatial order. Accordingly, certain operations illustrated in the figures or otherwise disclosed herein may be performed in a different order than expressly illustrated or described, as appropriate for particular embodiments of the disclosure.Furthermore, in some embodiments, certain activities may be performed in parallel, for example, by dedicated parallel processing devices or by separate computers configured to work together as part of a larger system.

[0096] As used herein in the context of computer use, the terms "component," "system," "module," and the like, unless otherwise specified or limited, are intended to include some or all of a computer-related system that includes hardware, software, a combination of hardware and software, or software in execution. A component may be, for example, but is not limited to, a processor device, an operation performed (or executable) by a processor device, an object, an executable file, a thread of execution, a computer program, or a computer. For illustrative purposes, both an application running on a computer and the computer may be a component. One or more components (or systems, modules, etc.)) may be located within a process or thread of execution, localized on one computer, distributed across two or more computers or other processing devices, or contained within another component (or system, module, etc.).

[0097] In some embodiments, the devices or systems disclosed herein may be used or installed using methods embodying aspects of the disclosure. Accordingly, the description contained herein of particular features, capabilities, or intended purposes of a device or system is generally intended to include the disclosure of a method for utilizing such features for the intended purposes, a method for implementing such capabilities, and a method for installing disclosed (or otherwise known) components to support those purposes or capabilities.Similarly, unless otherwise stated or limited, discussion of a method of making or using a particular device or system, including installation of the device or system, is intended by its nature to include disclosure of the employed features and implemented capabilities of such device or system as embodiments of the disclosure.

[0098] Unless otherwise defined or limited, ordinal numbers are used herein, which are generally based on the order in which particular components are presented for the relevant portion of the disclosure. For example, in this context, designations such as "first," "second," etc., generally indicate only the order in which the component in question is presented for discussion and generally do not denote or require any particular spatial arrangement, functional or structural precedence, or order.

[0099] Unless otherwise defined or restricted, directional terms are used herein to facilitate discussion of specific drawings or examples. For example, references to downward (or other) directions or upper (or other) positions may be used to discuss aspects of a particular example or drawing, but do not necessarily require a similar orientation or geometry in all installations or configurations.

[0100] Unless otherwise defined or limited, the term "and / or" as used herein refers to two or more elements, both individually and collectively. For example, a device with "a and / or b" is intended to include: a device with a (but not b); a device with b (but not a); and a device with both a and b.

[0101] This discussion is intended to enable those skilled in the art to make and use embodiments of the disclosure. Various modifications to the illustrated examples will be readily apparent to those skilled in the art, and the general principles set forth herein may be applied to other examples and applications without departing from the principles disclosed herein. Therefore, the embodiments of the disclosure are not limited to the illustrated embodiments, but are to be given the widest possible applicability consistent with the principles and features disclosed herein and the following claims. The following detailed description should be read with reference to the drawings, in which the same elements have the same reference numerals in different drawings. The drawings, which are not necessarily to scale, illustrate selected examples and are not intended to limit the scope of the disclosure.Those of skill in the art will recognize that the examples presented here offer many useful alternatives and fall within the scope of the disclosure. QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature

[0000] US 63 / 608,034

[0001] US 63 / 616,115

[0001]

Claims

[1] Sawing machine, comprising: at least one camera; an inertial measurement unit (IMU) sensor; a saw blade; a motor adapted to drive the saw blade; an electronic control unit comprising an electronic processor and a memory, the electronic control unit being configured to: to receive an indication of the saw blade orientation from the IMU sensor; to determine a restricted area based on the orientation of the saw blade, the restricted area corresponding to the saw blade; receive images captured by the at least one camera, the captured images comprising at least a portion of the restricted area; analyze the captured images using a machine learning (ML) model to determine whether a portion of a hand is located in the restricted area; and to perform a security action in response to the detection of the portion of the hand in the restricted area. [2] Sawing machine according to claim 1, wherein the sawing machine is a cross-cut saw. [3] Sawing machine according to claim 1, wherein the sawing machine is a table saw. [4] A sawing machine according to any one of claims 1 to 3, wherein the safety measure comprises at least one of the following measures: controlling a feedback light to generate a visual warning, controlling a feedback speaker to generate an audible warning, and controlling the motor to stop. [5] Sawing machine according to one of claims 1 to 4, wherein the restricted area comprises a warning zone and a danger zone and the danger zone is smaller than the warning zone. [6] Sawing machine according to claim 5, wherein the danger zone lies at least partially within the warning zone. [7] Sawing machine according to claim 5, wherein the electronic control unit is arranged, in response to detecting the portion of the hand in the warning zone, to generate a warning to perform a first safety measure; and the electronic control unit is arranged to carry out a second safety measure in response to the detection of the portion of the hand in the danger zone. [8] Sawing machine according to claim 7, wherein the first safety measure comprises at least one of the following measures: controlling a feedback light to generate a visual warning or controlling a feedback loudspeaker to generate an audible warning. [9] A sawing machine according to claim 7, wherein the second safety measure comprises stopping the motor. [10] A sawing machine according to claim 9, wherein the second safety measure further comprises controlling a feedback light to generate a visual warning and / or controlling a feedback speaker to generate an audible warning. [11] Sawing machine according to claim 1, wherein the at least one camera comprises two cameras. [12] Sawing machine according to claim 11, wherein the two cameras comprise a first camera arranged on one side of the saw blade and a second camera arranged on an opposite side of the saw blade. [13] Sawing machine according to claim 12, wherein the first camera and the second camera are arranged near the saw blade. [14] Sawing machine according to claim 1, wherein the electronic control device is further arranged to: generate optical flow field data by analyzing the acquired images with an optical flow recognition model; to determine the presence of an unsafe operating condition based on the motion vectors contained in the optical flow field data; and to implement an additional safety measure in response to the detection of the unsafe operating condition. [15] A sawing machine according to claim 14, wherein the additional safety measure comprises at least one of the following measures: controlling a feedback light to generate a visual warning, controlling a feedback speaker to generate an audible warning and controlling the motor to stop. [16] A sawing machine according to claim 14, wherein the presence of the unsafe operating condition is determined when at least one motion vector in the optical flow field data indicates the movement of an object towards the restricted area. [17] A sawing machine according to claim 14, wherein the presence of the unsafe operating condition is determined when at least one motion vector in the optical flow field data indicates the movement of an object from the saw blade at a speed above a safety threshold. [18] A sawing machine according to claim 14, wherein the presence of the unsafe operating condition is determined when at least one motion vector in the optical flow field data indicates an unsafe environmental condition around the sawing machine. [19] A method for operating a sawing machine, comprising: Receiving an indication of the orientation of a saw blade from a sensor of the sawing machine; Determining a restricted area based on the orientation of the saw blade, the restricted area defining a volume relative to the saw blade; Receiving image data captured by a camera of the sawing machine, the image data depicting at least a portion of the restricted area; Analyzing the image data using a machine learning (ML) model to determine whether a portion of a hand is located in the restricted area; and Performing a security action in response to detecting the portion of the hand in the restricted area. [20] The method of claim 19, wherein the alignment comprises at least one skew angle and one miter angle. [21] A method according to claim 19, wherein the sawing machine is a cross-cut saw or a table saw. [22] The method of claim 19, wherein the safety measure comprises at least one of the following measures: controlling a feedback light to generate a visual warning, controlling a feedback speaker to generate an audible warning, and controlling a motor of the sawing machine to stop driving the saw blade. [23] A method according to claim 19, wherein the restricted area comprises a warning zone and a danger zone, the danger zone being smaller than and located within the warning zone. [24] The method of claim 23 further comprising: in response to detecting the portion of the hand in the warning zone, executing the safety measure by controlling a feedback device to generate a warning; and in response to detecting the portion of the hand in the danger zone, controlling the motor to stop. [25] The method of claim 19, wherein analyzing the image data using the ML model comprises: Inputting the image data into the ML model using the electronic processor, generating hand recognition data as output, wherein the hand recognition data indicates the portion of the hand detected in the image data; Creating a projected exclusion zone by projecting the exclusion zone onto a two-dimensional (2D) plane; and Determine whether the portion of the hand is in the exclusion area by determining whether the hand detection data overlaps with the projected exclusion area.

Images