Danger prediction system

Abstract

This invention provides a hazard prediction system that can perform high-precision hazard prediction regardless of the user's knowledge and experience. The hazard prediction system (100) includes an information acquisition unit (101), a plan acquisition unit (102), and a hazard prediction information acquisition unit (106). The information acquisition unit (101) acquires site information related to the work site (WS). The plan acquisition unit (102) acquires the work plan in the work site (WS). The hazard prediction information acquisition unit (106) acquires hazard prediction information as a result of having a language model (LM) interpret the site information acquired by the information acquisition unit (101) and the work plan acquired by the plan acquisition unit (102).

Description

Technical Field

[0001] This invention relates to a hazard prediction system. Background Technology

[0002] Previously, a procedure was known that could appropriately support hazard prediction activities (see Patent Document 1 below).

[0003] The procedure described in Patent Document 1 involves displaying a dialog box with continuously configured input and output text on a display unit, and accepting input text from the operator indicating potential hazards at the work site. Furthermore, the computer performs the following processing: upon receiving the input text indicating a hazard in the dialog box, it displays an output text indicating countermeasures against that hazard in the dialog box.

[0004] Existing technical documents Patent documents Patent Document 1: Japanese Patent Application Publication No. 2021-018751 Summary of the Invention

[0005] The technical problem to be solved by the invention However, in the technology described in Patent Document 1, the user needs to consider the hazards of the operation and select from multiple options or input them directly. Therefore, there is a problem that the accuracy of the hazard prediction for the operation depends on the user's knowledge and experience.

[0006] This invention provides a hazard prediction system that can perform high-precision hazard prediction regardless of the user's knowledge or experience.

[0007] means for solving technical problems One aspect of the present invention provides a hazard prediction system comprising: an information acquisition unit for acquiring site information related to a work site; a plan acquisition unit for acquiring a work plan at the work site; and a hazard prediction information acquisition unit for acquiring hazard prediction information as a result of a language model interpreting the site information acquired by the information acquisition unit and the work plan acquired by the plan acquisition unit.

[0008] Invention Effects According to the above-described manner of the present invention, a hazard prediction system can be provided that can perform high-precision hazard prediction regardless of the user's knowledge or experience. Attached Figure Description

[0009] Figure 1 This is a conceptual diagram illustrating an example of an implementation of the hazard prediction system according to the present invention.

[0010] Figure 2 It shows the composition Figure 1 A diagram illustrating the configuration of an information terminal in a hazard prediction system.

[0011] Figure 3 Is with Figure 1 The functional block diagram related to hazard prediction in the hazard prediction system.

[0012] Figure 4 It is shown Figure 1 A schematic diagram of a hazard prediction system as an example of an object's work site.

[0013] Figure 5 It is shown Figure 1 A flowchart illustrating an example of hazard prediction performed by a hazard prediction system. Detailed Implementation

[0014] Hereinafter, embodiments of the hazard prediction system according to the present invention will be described with reference to the accompanying drawings. The embodiments described below are illustrative and do not limit the invention. All features and combinations thereof in the embodiments of the present invention are not necessarily the essence of the invention. Furthermore, in the accompanying drawings, the same or corresponding components are sometimes labeled with the same or corresponding symbols, and repeated descriptions are omitted.

[0015] Figure 1 This is a conceptual diagram illustrating an example of an implementation of the hazard prediction system according to the present invention. Figure 2 It shows the composition Figure 1 A diagram illustrating the configuration of an information terminal IT of a danger prediction system 100. Figure 3 Is with Figure 1 The functional block diagram related to the hazard prediction system 100. Figure 4 It is shown Figure 1 A schematic diagram of a hazard prediction system 100 as an example of a work site WS.

[0016] The hazard prediction system 100 of this embodiment, for example, predicts... Figure 4 This refers to identifying potential hazards in work sites such as construction sites and building sites, and supporting hazard prediction training implemented in these work sites. Hazard prediction training, also known as hazard prediction activities or toolbox meetings, is training for workers engaged in construction, manufacturing, and other operations to anticipate potential hazards in their work and point them out to each other in order to prevent accidents and disasters.

[0017] Details will be described later, but the hazard prediction system 100 of this embodiment is characterized by, for example Figure 3As shown, it mainly includes an information acquisition unit 101, a planning acquisition unit 102, and a predictive information acquisition unit 106. The information acquisition unit 101 acquires site information related to the work site (WS). The planning acquisition unit 102 acquires the work plan from the work site (WS). The predictive information acquisition unit 106, as a result of the language model (LM) interpreting the site information acquired by the information acquisition unit 101 and the work plan acquired by the planning acquisition unit 102, acquires, for example, hazard prediction information including suggestions for hazard prediction training.

[0018] The configuration of the hazard prediction system 100 of this embodiment will be described in detail below.

[0019] like Figure 1 As shown, the hazard prediction system 100 of this embodiment is composed of one or more information terminals IT connected to a communication line NW. Furthermore, the hazard prediction system 100 may, for example, have one or more servers SV connected to the communication line NW.

[0020] The information terminal IT and server SV constituting the hazard prediction system 100 are connected to one or more external servers ESV via communication lines NW. Furthermore, the information terminal IT and server SV constituting the hazard prediction system 100 are configured to acquire detection results from sensors SN installed at the work site WS via communication lines NW.

[0021] Communication lines NW may include, for example, a local area network (LAN) of the work site WS. Furthermore, communication lines NW may also include wide area networks (WANs). Wide area networks include, for example, mobile communication networks with base stations as terminals, satellite communication networks utilizing communication satellites, and the Internet. Furthermore, communication lines NW may also include, for example, short-range communication lines based on wireless communication standards such as WiFi and Bluetooth (registered trademark).

[0022] Information terminals (IT) include, for example, mobile information terminals (PDAs) such as smartphones and tablet PCs, desktop PCs (DPCs), and laptop computers (LPCs). Servers (SVs) are computers connected to information terminals (IT) via communication lines (NW).

[0023] For example, such as Figure 2 As shown, the information terminal IT includes an output device OD, an input device ID, a communication interface CI, a central processing unit (CPU), an auxiliary storage device AS, and a memory device MD, all connected via a bus. Furthermore, the server SV can also have... Figure 2 The information terminal shown has the same structure as IT.

[0024] Output devices (ODs) include, for example, those that output various information to the user. Examples of output devices (ODs) include display devices, sound output devices, printers, vibration generating devices, etc. Display devices include, for example, liquid crystal displays (LCDs), organic EL (Electro-Luminescence) displays, and indicator lights. Sound output devices include, for example, speakers and buzzers.

[0025] The input device ID receives various inputs from the user. The input device ID includes, for example, input devices (mechanical input devices) that accept mechanical operation inputs from the user. Mechanical input devices include, for example, buttons, toggle switches, keyboards, mice, touch panels mounted on the output device OD, and touchpads separate from the output device OD.

[0026] Furthermore, the input device ID may include a voice input device capable of accepting voice input from a user. The voice input device may include, for example, a microphone capable of collecting the user's voice. The input device ID may also include a biometric input device capable of accepting biometric input from a user. The biometric input device may include, for example, a camera capable of acquiring image data containing information related to the user's fingerprint or iris.

[0027] The Communication Interface (CI) is used as an interface for communicating with external devices. Thus, the Information Terminal (IT) can communicate with other Information Terminals (IT), Servers (SVs), External Servers (ESVs), Sensors (SNs), and other external devices through the Communication Interface (CI). Furthermore, the Communication Interface (CI) can have multiple types of communication interfaces depending on the communication method with the connected device.

[0028] For example, the CPU executes programs loaded into the memory device MD and performs various functions of the information terminal IT according to the program's commands.

[0029] Auxiliary storage device (AS) is a non-volatile storage device that stores installed programs and necessary files, data, etc. Examples of auxiliary storage devices (AS) include EEPROM (Electrically Erasable Programmable Read-Only Memory) and flash memory.

[0030] Memory device MD, for example, loads a program from secondary storage device AS in a manner that the CPU can read, provided a program start instruction is present. Memory device MD is, for example, SRAM (Static Random Access Memory).

[0031] Figure 1The external server ESV shown includes, for example, an external server providing meteorological information, an external server providing a construction management system, and an external server providing a database of hazardous incidents occurring at the work site. The construction management system is, for example, a cloud service that streamlines the business processes of construction management, including the management of photos, construction drawings, and work procedures from various work sites' work windows (WS).

[0032] Users of the Hazard Prediction System 100 can, for example, obtain meteorological information for the work site WS at a specified date and time from an external server ESV that provides meteorological information via an information terminal IT. The meteorological information includes, for example, information on the weather, temperature, humidity, air pressure, precipitation, sunshine, snow cover, wind speed, and wind direction at the work site WS.

[0033] Users of the Hazard Prediction System 100 can, for example, pre-register photos, construction drawings, and work order sheets of the work site (WS) in the external server (ESV) providing the construction management system via the information terminal (IT). Furthermore, users of the Hazard Prediction System 100 can, for example, retrieve the pre-registered photos, construction drawings, and work order sheets of the work site (WS) from the external server (ESV) providing the construction management system via the information terminal (IT).

[0034] Furthermore, the server SV constituting the hazard prediction system 100 can obtain meteorological information for the work site WS at a specified date and time from the external server ESV that provides meteorological information and store it in a storage device. The server SV can also obtain photos, construction drawings, work order sheets, etc., of the work site WS from the external server ESV that provides the construction management system and store them in a storage device. Additionally, users of the hazard prediction system 100 can, for example, pre-register photos, construction drawings, work order sheets, etc., of the work site WS in the server SV via an information terminal IT.

[0035] For example, such as Figure 4 As shown, the sensor SN includes a camera device SN1, a temperature and humidity sensor SN2, a wind direction and speed sensor SN3, a rain gauge SN4, etc., installed at the work site WS. Furthermore, the sensor SN may include, for example, a LiDAR (Light Detection and Ranging) ranging sensor, a noise sensor, a vibration sensor, an oxygen content sensor, a hydrogen sulfide content sensor, a dust sensor, a rain sensor, a soil moisture sensor, etc. Moreover, the sensor SN may include sensors for detecting the status of construction machinery such as hydraulic excavators HS, and sensors for detecting the physical condition of workers operating at the work site WS.

[0036] The camera device SN1 is, for example, a monocular camera. Furthermore, the camera device SN1 can also be, for example, a stereo camera, a ToF (Time of Flight) camera, a depth camera, etc., capable of acquiring not only two-dimensional image information but also three-dimensional information including information related to the distance to objects projected in the image and the depth of the image; it is a 3D camera. The camera device SN1 is, for example, installed at a high point such as the roof or pillar of the field office SO at the work site WS, providing a panoramic view of the entire work site WS.

[0037] The camera device SN1 is connected to the communication line NW and sends the captured images to the server SV and the information terminal IT. Furthermore, the camera device SN1, capable of acquiring 3D information, and ranging sensors such as LiDAR detect the position, size, and shape of objects present at the work site WS and send this information to the server SV and the information terminal IT. Objects present at the work site WS include, for example, personnel P, hydraulic excavators HS, construction machinery such as cranes, vehicles such as dump trucks DT, construction materials CM such as piping, safety fences SF, and traffic cones PY.

[0038] As described above, the hazard prediction system 100 of this embodiment includes Figure 3 The information acquisition unit 101, the planning acquisition unit 102, and the predictive information acquisition unit 106 are shown. Furthermore, in... Figure 3 In the example shown, the danger prediction system 100 also includes a speech unit 103, a prompt word generation unit 104, and a calling unit 105.

[0039] Figure 3 The various parts of the hazard prediction system 100 shown represent, for example, the functions of the hazard prediction system 100. These functions are achieved by loading a program stored in the auxiliary storage device AS into the memory device MD and executing it via the CPU, from the information terminal IT or the server SV. That is, Figure 3 Each part of the hazard prediction system 100 shown is implemented in at least one of the information terminal IT and the server SV connected via the communication line NW.

[0040] Language models (LMs) are installed on information terminals (ITs), servers (SVs), or external servers (ESVs) depending on capacity. Language model LMs can be large-scale language models (LLMs). For example, GPT-4 can be used as a language model LM.

[0041] The information acquisition unit 101 acquires, for example, on-site information related to the work site WS, and outputs the acquired on-site information to at least one of the speech recognition unit 103 and the prompt generation unit 104. The on-site information includes, for example, at least one of the construction drawings of the work site WS, the detection results of the sensors SN installed on the work site WS, or the meteorological information (WEB information) of the work site WS acquired via the communication line NW. That is, the information acquisition unit 101 is configured to acquire on-site information including at least one of the construction drawings of the work site WS, the detection results of the sensors SN installed on the work site WS, or the meteorological information of the work site WS.

[0042] Specifically, the information acquisition unit 101 may, for example, access an external server ESV that provides the construction management system via a communication interface CI and a communication line NW, thereby obtaining the construction drawings of the work site WS from the external server ESV. Furthermore, the information acquisition unit 101 may, for example, obtain the construction drawings of the work site WS from an information terminal IT or a server SV that has pre-stored the construction drawings of the work site WS in an auxiliary storage device AS.

[0043] Furthermore, the information acquisition unit 101 can acquire the detection results of the sensor SN by directly accessing the sensor SN installed at the work site WS via the communication interface CI and the communication line NW. Alternatively, the information acquisition unit 101 can acquire the detection results of the sensor SN by accessing the information terminal IT or server SV, which have pre-stored the detection results of the sensor SN in the auxiliary storage device AS.

[0044] The information acquisition unit 101 acquires information about objects at the work site WS, for example, based on the detection results of a camera device SN1 or a sensor SN such as a ranging sensor. Specifically, the information acquisition unit 101 detects objects of the monitored target from the images of the camera device SN1 by applying known image processing techniques such as semantic segmentation and machine learning. Objects of the monitored target include, for example, personnel P such as workers. Furthermore, objects of the monitored target may also include other objects present at the work site WS. Other objects include, for example, hydraulic excavators HS, cranes, and other construction machinery, dump trucks DT, and other vehicles. Other objects may also include specific fixed objects present at the work site WS, such as piled-up construction materials CM, safety fences SF, and traffic cones PY. Furthermore, other objects may include specific terrain shapes present at the work site WS, such as trenches, holes, and sand piles.

[0045] Furthermore, the information acquisition unit 101 acquires, for example, on-site information of the work site WS detected by sensors SN such as temperature and humidity sensor SN2, wind direction and speed sensor SN3, and rain gauge SN4, including current temperature, humidity, wind direction, wind speed, and rainfall. Additionally, the information acquisition unit 101 acquires, for example, on-site information of the work site WS detected by sensors SN such as noise sensor, vibration sensor, oxygen content sensor, hydrogen sulfide content sensor, dust sensor, rain sensor, and soil moisture sensor, including current noise, oxygen concentration, hydrogen sulfide concentration, dust concentration, rainfall, and soil moisture content. Furthermore, the information acquisition unit 101 acquires, for example, on-site information such as the posture of the construction machinery, the forces acting on the construction machinery, and the physical condition of the workers operating at the work site WS, detected by sensors SN such as those detecting the status of construction machinery such as hydraulic excavators HS and those detecting the physical condition of workers operating at the work site WS.

[0046] Furthermore, the information acquisition unit 101 may, for example, access an external server ESV that provides meteorological information via a communication interface CI and a communication line NW, thereby obtaining the meteorological information of the work site WS for the day of operation from the external server ESV. Alternatively, the information acquisition unit 101 may also obtain the meteorological information of the work site WS for the day of operation from an information terminal IT or a server SV that has pre-stored the meteorological information of the work site WS in an auxiliary storage device AS.

[0047] Furthermore, the information acquisition unit 101 can also acquire, for example, past hazardous incidents that occurred at other work sites. Specifically, the information acquisition unit 101 can acquire past hazardous incidents that occurred at other work sites by accessing an external server ESV, which stores a database of past hazardous incidents that occurred at other work sites, via a communication interface CI and a communication line NW. Alternatively, the information acquisition unit 101 can also acquire past hazardous incidents that occurred at other work sites from an information terminal IT or a server SV that pre-stores a database of past hazardous incidents that occurred at other work sites.

[0048] The planning acquisition unit 102 acquires, for example, the work plan in the work site WS, and outputs the acquired site information to at least one of the language processing unit 103 and the prompt generation unit 104. The work plan may include, for example, natural language-based text input by the operators of the work site WS into the input device ID of the information terminal IT, or at least one of the work schedule for the day in the work site WS.

[0049] The information terminal (IT) may have an input device ID capable of inputting work plans in natural language. Specifically, the input device ID of the information terminal (IT) may include, for example, the aforementioned mechanical input device. At this time, the operator at the work site (WS) inputs the work plan for the day in natural language text, for example, using the keyboard of a desktop personal computer (DPC) located in the field office (SO), a laptop computer (LPC) they carry, or the touch panel of a PDA. The plan acquisition unit 102 acquires the natural language text-based work plan input into the mechanical input device of the information terminal (IT).

[0050] Furthermore, the input device ID of the information terminal IT includes, for example, the aforementioned voice input device. At this time, the operator at the work site WS speaks the work plan for the day in natural language and inputs it into the voice input device of the desktop personal computer (DPC) located in the field office SO or their own PDA or laptop computer (LPC). The plan acquisition unit 102, for example, uses known speech recognition technology to transcribe the speech input into the voice input device of the information terminal IT, thereby acquiring a text-based work plan in natural language.

[0051] Furthermore, the planning acquisition unit 102 is configured to acquire a work plan that includes the work schedule of the work site WS. Specifically, the planning acquisition unit 102, for example, accesses an external server ESV that provides the construction management system via a communication interface CI and a communication line NW, thereby acquiring the work schedule of the work site WS for the current work day from the external server ESV. Additionally, the planning acquisition unit 102 can, for example, acquire the work schedule of the work site WS for the current work day from an information terminal IT or a server SV that has pre-stored the work schedule of the work site WS in an auxiliary storage device AS.

[0052] The languageization unit 103, for example, uses natural language to translate the on-site information and hazard examples acquired by the information acquisition unit 101, as well as the process schedule acquired by the planning acquisition unit 102. Specifically, the languageization unit 103 translates the information acquired by the information acquisition unit 101 into a predefined text template, for example.

[0053] The template for the text indicating the current weather conditions at the work site (WS) is as follows: for example, "aaa" represents the weather, "bbb" represents the temperature, "ccc" represents the humidity, "ddd" represents the wind direction, and "eee" represents the wind speed. It can be specified in the form of "The current weather at the work site is aaa, the temperature is bbb degrees, the humidity is ccc degrees, the wind is ddd, and the wind speed is eee meters." The template for the text indicating the configuration of monitored objects in the work site (WS) is as follows: for example, "fff" represents the location information of the monitored object, and "ggg" represents the type or name of the monitored object. It can be specified in the form of "ggg is present at fff." or "ggg exists at fff."

[0054] Alternatively, the language generation unit 103 can be omitted. In this case, for example, the work plan, including the on-site information and hazard examples acquired by the information acquisition unit 101 and the process table acquired by the plan acquisition unit 102, is input into the prompt word generation unit 104.

[0055] The prompt word generation unit 104 generates prompt words input to the language model LM, for example, based on the site information acquired by the information acquisition unit 101 and the work plan acquired by the plan acquisition unit 102. Furthermore, the site information and work plan used by the prompt word generation unit 104 in prompt word generation can be language-translated by the language translation unit 103. The prompt word generation unit 104 generates, for example, prompt words for outputting hazard prediction information corresponding to a combination of untranslated or translated site information and untranslated or translated work plans.

[0056] The prompt generation unit 104 generates multiple example questions, which are pre-assigned to the language model LM via the calling unit 105. The example questions are defined by a combination of on-site information of the work site WS as prerequisites (i.e., constraints), the work plan, and the correct answer to be output. Thus, the language model LM can understand (learn) the output format indicated by the prompt words.

[0057] Examples of questions assigned to the language model LM include, for example, on-site information such as two-dimensional and three-dimensional construction drawings of the work site (WS), long-term and short-term meteorological information acquired via communication lines (NW), images of the work site (WS), and detection results from sensors (SN) measuring environmental values. Furthermore, examples of questions as work plans include various tasks that workers may perform, tasks for the current day and several days later as interpreted from construction drawings and work schedules, and the time allocation for each task. Additionally, examples of questions as correct answers include hazard prediction information such as precautions based on buried objects, numerical values, and terrain as interpreted from construction drawings.

[0058] More specifically, the on-site information assigned to the example questions in the language model LM includes, for example, the number of personnel P based on the detection results of camera device SN1, the ground condition of the work site WS, the working conditions of personnel P, the state of dust, the state of accumulated construction materials CM, and the state of the work area. In this case, the correct answer to the example question includes hazard prediction information related to preventing personnel P from coming into contact with each other and preventing falls.

[0059] Furthermore, when the on-site information assigned to the language model LM includes environmental measurements based on sensors SN installed at the work site WS, the answers to the examples include hazard prediction information related to worker health damage. Also, when the on-site information assigned to the language model LM includes meteorological information obtained via communication line NW, the answers to the examples include hazard prediction information related to slipping due to weather, disruption of work procedures, and hazard prediction information related to worker health damage.

[0060] The calling unit 105 calls the language model LM through a specified API (Application Programming Interface) for example, and inputs the prompt words generated by the prompt word generation unit 104 into the language model LM to obtain its output (answer).

[0061] The predictive information acquisition unit 106 acquires hazard prediction information as the output of the language model LM via the invocation unit 105. Specifically, the predictive information acquisition unit 106 acquires hazard prediction information as a result of the language model LM interpreting the site information acquired by the information acquisition unit 101 and the work plan acquired by the plan acquisition unit 102. More specifically, the predictive information acquisition unit 106 acquires hazard prediction information, for example, including suggestions for hazard prediction training.

[0062] Furthermore, as described above, the information acquisition unit 101 can acquire hazardous incidents corresponding to the site information and work plan of the work site WS from an external server ESV that provides a database of hazardous incidents occurring at other sites. At this time, the predictive information acquisition unit 106 can also acquire hazard prediction information as a result of the language model LM interpreting hazardous incidents occurring at other work sites in addition to the site information and work plan of the work site WS.

[0063] Furthermore, the hazard prediction system 100 of this embodiment includes an information terminal IT with an output device OD. At this time, the prediction information acquisition unit 106 generates a control signal for outputting the acquired hazard prediction information and outputs it to the output device OD, which includes a display device. As a result, hazard prediction information corresponding to the site information and work plan of the work site WS is output from the output device OD of the information terminal IT, and the hazard prediction information is displayed on the display device.

[0064] Next, refer to Figure 5 The operation of the hazard prediction system 100 in this embodiment will be explained. Figure 5 It is shown Figure 1 A flowchart illustrating an example of a hazard prediction system 100 performing hazard prediction.

[0065] If start Figure 5 The processing flow shown is such that the hazard prediction system 100 performs the process P1 of acquiring site information. In this process P1, the information acquisition unit 101 acquires the site information of the work site WS where the operator is performing the work from the information terminal IT, server SV, or external server ESV, and outputs the acquired information to the speech unit 103 and the prompt word generation unit 104.

[0066] Next, the hazard prediction system 100 executes the work plan acquisition process P2. In this process P2, the plan acquisition unit 102 acquires the work plan of the operation performed by the operator at the work site WS from the information terminal IT, server SV or external server ESV, and outputs the acquired information to the speech unit 103 and the prompt word generation unit 104.

[0067] Next, the hazard prediction system 100 performs, for example, a process P3 that translates the site information and work plan into language. In this process P3, the language translation unit 103, for example, translates the site information and work plan input from the information acquisition unit 101 and the plan acquisition unit 102 into natural language and outputs it to the prompt word generation unit 104. Alternatively, if the hazard prediction system 100 does not have a language translation unit 103 and there is no need to translate the site information and work plan into language, process P3 can be omitted.

[0068] Next, the hazard prediction system 100 performs, for example, a prompt word generation process P4. In this process P4, the prompt word generation unit 104 generates prompt words to be assigned to the language model LM based on the site information and work plan input from the information acquisition unit 101 and the plan acquisition unit 102, and the already verbalized site information and work plan input from the languageization unit 103. The prompt word generation unit 104 outputs the generated prompt words to the calling unit 105, for example.

[0069] Next, the hazard prediction system 100 executes, for example, the process P5 of calling the language model LM. In this process P5, the calling unit 105 calls the language model LM and assigns the prompt words generated by the prompt word generation unit 104 in the previous process P4 to the language model LM.

[0070] Next, the hazard prediction system 100 executes the process P6 for acquiring hazard prediction information. In this process P6, the prediction information acquisition unit 106 acquires, for example, the response of the language model LM to the prompt words assigned in the previous process P5, i.e., the hazard prediction information, via the calling unit 105.

[0071] Next, the hazard prediction system 100 executes, for example, the process P7 of outputting hazard prediction information. In this process P7, the prediction information acquisition unit 106 generates a control signal to output the hazard prediction information acquired in the previous process P6 to the output device OD of the information terminal IT, and outputs it to the output device OD. Then, the hazard prediction system 100 terminates. Figure 5 The processing flow is shown below.

[0072] The hazard prediction system 100 of this embodiment executes... Figure 5 The processing flow shown can be used in the following scenarios.

[0073] Suppose that the first scenario of using the hazard prediction system 100 is, for example, the scenario in which the workers use it on-site before each operation such as excavation or loading in the work site WS.

[0074] Specifically, workers input the work content to the hazard prediction system 100 in natural language using the touch panel or voice input device of their own PDAs. Workers can also input additional information such as emergency situations, collaboration with hydraulic excavators (HS), or whether they are beginners. The hazard prediction system 100's planning acquisition unit 102 then acquires the input work content and additional information as a work plan and outputs the plan to the speech recognition unit 103 and the prompt generation unit 104.

[0075] Furthermore, the information acquisition unit 101 of the hazard prediction system 100 acquires, for example, on-site information, including environmental measurement results detected by sensors SN installed at the work site WS where the operator performs the work, construction drawings of the work site WS, and meteorological information of the work site WS via communication line NW. The information acquisition unit 101 outputs the acquired on-site information to the speech recognition unit 103 and the prompt word generation unit 104.

[0076] The prompt word generation unit 104 generates prompt words based on the on-site information and work plan input from the information acquisition unit 101 and the plan acquisition unit 102, and the already verbalized on-site information and work plan input from the language processing unit 103, and outputs them to the calling unit 105. The calling unit 105 calls the language model LM and assigns prompt words to the language model LM. The predictive information acquisition unit 106 acquires the hazard prediction information obtained by the calling unit 105 as the response of the language model LM, and outputs the acquired hazard prediction information to the display device and sound output device of the PDA.

[0077] As a result, for example, the PDA's display device shows hazard warning information, including indications of dangerous areas related to the work site (WS) and situations where people are likely to fall in, and the PDA's voice output device outputs voice information related to the hazard warning information. Furthermore, the hazard warning system 100 can also communicate with workers via the input device ID of an information terminal (IT) such as the PDA's voice input device or touch panel.

[0078] Therefore, for example, it is possible to prevent dangers to workers caused by schedule deviations from the initial plan at the start of the work, and to reduce worker anxiety. Specifically, for example, it can prevent workers from taking shortcuts on walking routes, or approaching heavy machinery such as hydraulic excavators (HS) before they have come to a complete stop. Thus, it can reduce the burden on construction managers at the work site for safety inspections.

[0079] Suppose that the second scenario using the hazard prediction system 100 is, for example, a scenario in which a worker performing work at the work site WS is notified of a hazard.

[0080] At this time, the information acquisition unit 101 of the hazard prediction system 100 acquires, for example, on-site information including environmental measurement results of the work site WS detected by the sensor SN, measurement results of the workers' physical condition, and meteorological information provided by the external server ESV. Furthermore, the information acquisition unit 101 acquires, for example, hazardous incidents occurring in other work sites from the information terminal IT, server SV, or external server ESV. Additionally, the planning acquisition unit 102 acquires, for example, work plans such as the work schedule of the work site WS where workers are performing their tasks from the information terminal IT, server SV, or external server ESV.

[0081] As a result, the predictive information acquisition unit 106, as a result of the language model LM interpreting on-site information, hazardous events, and work plans, can acquire hazard prediction information, such as situations that could lead to accidents and detection reports of poor physical condition of workers. Furthermore, the predictive information acquisition unit 106 notifies the workers of the acquired hazard prediction information via the output device OD of the information terminal IT. Thus, the hazard prediction system 100 can notify workers of predicted hazards in the work site WS, thereby providing hazard prediction training and operational assistance to less experienced workers.

[0082] Suppose that the third scenario of using the hazard prediction system 100 is, for example, a scenario in which recommendations related to safety during the construction phase are made when design drawings and construction drawings are produced.

[0083] At this time, the person in charge of creating construction drawings, etc., inputs the construction purpose, intent, number of planned workers, time period, work plan, and other necessary construction information into the hazard prediction system 100 via the input device ID of the information terminal IT. The information acquisition unit 101 of the hazard prediction system 100 acquires the information input by the person in charge, construction drawings, meteorological information and geological information provided by the external server ESV, etc., as on-site information. Furthermore, the information acquisition unit 101 acquires, for example, hazardous incidents that have occurred at other work sites from the information terminal IT, server SV, or external server ESV. The planning acquisition unit 102 acquires, for example, the work plan input by the person in charge.

[0084] As a result, the predictive information acquisition unit 106 can, for example, acquire hazard prediction information, including past incident examples and indications of hazards caused by time constraints, based on the interpretation of site information, hazardous incidents, and work plans by the language model LM. The predictive information acquisition unit 106 outputs the acquired hazard prediction information to the output device OD of the information terminal IT. Therefore, when construction is carried out at the work site WS, the frequency of encountering hazardous situations can be reduced. Furthermore, the hazard prediction system 100 can be used when creating construction drawings and when using drawings.

[0085] As explained above, the hazard prediction system 100 according to this embodiment includes an information acquisition unit 101, a plan acquisition unit 102, and a hazard prediction information acquisition unit 106. The information acquisition unit 101 acquires site information related to the work site WS. The plan acquisition unit 102 acquires the work plan in the work site WS. The hazard prediction information acquisition unit 106 acquires hazard prediction information as a result of the language model LM interpreting the site information acquired by the information acquisition unit 101 and the work plan acquired by the plan acquisition unit 102.

[0086] With this configuration, the hazard prediction system 100 of this embodiment can perform high-precision hazard prediction regardless of the knowledge and experience of users such as operators and design managers at the work site WS. That is, the hazard prediction system 100 of this embodiment, by assigning prompts based on site information and work plans specific to the work site WS to the language model LM, can utilize the vast knowledge of the language model LM to perform high-precision hazard prediction corresponding to each work site WS.

[0087] Furthermore, the hazard prediction system 100 of this embodiment includes an information terminal IT, which has an output device OD for outputting hazard prediction information. With this configuration, the hazard prediction system 100 can notify users such as operators and design managers of hazard prediction information via the output device OD of the information terminal IT.

[0088] Furthermore, in the hazard prediction system 100 of this embodiment, the information terminal IT has an input device ID that can input work plans in natural language. With this configuration, the user of the hazard prediction system 100 can input work plans into the hazard prediction system 100 via the input device ID of the information terminal IT, through natural language-based speech or text input.

[0089] Furthermore, in the hazard prediction system 100 of this embodiment, the planning acquisition unit 102 is configured to acquire a work plan that includes the work schedule of the work site WS. With this configuration, the user of the hazard prediction system 100 can confirm the hazard prediction information corresponding to the work of the day without inputting the work plan of the work site WS before the work begins.

[0090] Furthermore, in the hazard prediction system 100 of this embodiment, the information acquisition unit 101 is configured to acquire site information including construction drawings of the work site WS.

[0091] With this configuration, the hazard prediction system 100 of this embodiment can obtain hazard prediction information based on the construction drawings by having the language model LM interpret site information including the construction drawings. Specifically, for example, when the construction drawings include information about buried objects, the operator inputs instructions in natural language via the input device ID of the information terminal IT, which outputs the key points of the work to be performed at the work site WS. As a result, the operator can, for example, obtain hazard prediction information via the output device OD of the information terminal IT, including suggestions such as "Buried object display tape is attached at the location of the buried object. If there is tape, please perform manual digging to confirm the buried object," and "It is predicted that a large amount of excavated soil and sand will accumulate on the construction road at the end of the work. Please pay attention to the stopping position of the hydraulic excavator."

[0092] Furthermore, in the hazard prediction system 100 of this embodiment, the information acquisition unit 101 is configured to acquire field information including the detection results of the sensor SN installed at the work site WS. With this configuration, the hazard prediction system 100 of this embodiment can acquire hazard prediction information corresponding to the real-time field information of the work site WS including the detection results of the sensor SN.

[0093] Furthermore, in the hazard prediction system 100 of this embodiment, the information acquisition unit 101 is configured to acquire site information including meteorological information of the work site WS. With this configuration, the hazard prediction system 100 of this embodiment can acquire hazard prediction information corresponding to site information specific to the work site WS, which includes meteorological information of the work site WS.

[0094] Furthermore, in the hazard prediction system 100 of this embodiment, the prediction information acquisition unit 106 acquires hazard prediction information including suggestions for hazard prediction training. With this configuration, even less experienced operators can perform highly accurate hazard prediction training corresponding to the site information and work plan of the work site (WS). Moreover, since suggestions for hazard prediction training specific to the site information and work plan unique to the work site (WS) can be acquired, even skilled operators can perform hazard prediction training without becoming simplistic.

[0095] Furthermore, in the hazard prediction system 100 of this embodiment, the prediction information acquisition unit 106 acquires hazard prediction information as a result of the language model LM interpreting hazardous events occurring in other work sites in addition to on-site information and work plans. With this configuration, the user of the hazard prediction system 100 can obtain hazard prediction information based on hazardous events occurring in other work sites similar to the on-site information and work plans of the work site WS where the work is being carried out.

[0096] As explained above, according to this embodiment, a hazard prediction system 100 can be provided that can perform high-precision hazard prediction regardless of the user's knowledge or experience.

[0097] The preferred embodiments of the present invention have been described above. However, the invention is not limited to the embodiments described above. Various modifications and substitutions can be applied to the above embodiments without departing from the scope of the invention. Furthermore, the features described with reference to the above embodiments can be appropriately combined as long as they are not technically contradictory.

[0098] This application claims priority based on Japanese Patent Application No. 2024-073737, filed on April 30, 2024, the entire contents of which are incorporated herein by reference.

[0099] Symbol Explanation 100 - Hazard prediction system, 101 - Information acquisition unit, 102 - Planning acquisition unit, 106 - Predictive information acquisition unit, ID - Input device, IT - Information terminal, LM - Language model, OD - Output device (display device), SN - Sensor, WS - Work site. Claims (as amended under Article 19 of the Treaty) 1. A hazard prediction system, comprising: The Information Acquisition Department acquires on-site information related to the work site. The planning acquisition department acquires the work plan from the work site; and The predictive information acquisition unit acquires hazard prediction information as a result of the language model interpreting the site information acquired by the information acquisition unit and the work plan acquired by the plan acquisition unit. 2. The hazard prediction system according to claim 1, wherein, The system is equipped with an information terminal, which has an output device for outputting the danger prediction information. 3. The hazard prediction system according to claim 2, wherein, The information terminal has an input device that allows the user to input the job plan in natural language. 4. The hazard prediction system according to claim 1, wherein, The plan acquisition unit is configured to acquire the work plan, which includes the work schedule of the work site. 5. The hazard prediction system according to claim 1, wherein, The information acquisition unit is configured to acquire the site information, including the construction drawings of the work site. 6. The hazard prediction system according to claim 1, wherein, The information acquisition unit is configured to acquire the site information, including the detection results of sensors installed at the work site. 7. The hazard prediction system according to claim 1, wherein, The information acquisition unit is configured to acquire the site information, including meteorological information of the work site. 8. The hazard prediction system according to claim 1, wherein, The predictive information acquisition unit acquires the hazard prediction information, including suggestions for hazard prediction training. 9. The hazard prediction system according to claim 1, wherein, The predictive information acquisition unit acquires the hazard prediction information as a result of the language model interpreting hazardous events occurring in other work sites, in addition to the on-site information and the work plan. 10. (Additionally) The hazard prediction system according to claim 1, wherein, The language model described is a large-scale language model. 11. (Additionally) The hazard prediction system according to claim 10, wherein, It also includes a prompt word generation unit, which generates prompt words that are input into the large-scale language model based on the site information obtained by the information acquisition unit and the work plan obtained by the plan acquisition unit. 12. (Additionally) The hazard prediction system according to claim 11, wherein, It also includes a language processing unit, which processes the on-site information acquired by the information acquisition unit and the process schedule acquired by the planning acquisition unit into natural language and outputs it to the prompt word generation unit. 13. (Additionally) The hazard prediction system according to claim 11 or 12, wherein, It also includes a calling unit, which calls the large-scale language model, inputs the prompt words generated by the prompt word generation unit into the large-scale language model, and obtains the output of the large-scale language model. 14. (Additionally) The hazard prediction system according to claim 4, wherein, The planning and acquisition department obtains the work schedule for the current day from an external server. 15. (Additionally) The hazard prediction system according to claim 1, wherein, The information acquisition unit acquires the site information from an external server or from sensors installed at the work site.

Claims

1. A hazard prediction system, comprising: The Information Acquisition Department acquires on-site information related to the work site. The planning acquisition department acquires the work plan from the work site; and The predictive information acquisition unit acquires hazard prediction information as a result of the language model interpreting the site information acquired by the information acquisition unit and the work plan acquired by the plan acquisition unit.

2. The hazard prediction system according to claim 1, wherein, The system is equipped with an information terminal, which has an output device for outputting the danger prediction information.

3. The hazard prediction system according to claim 2, wherein, The information terminal has an input device that allows the user to input the job plan in natural language.

4. The hazard prediction system according to claim 1, wherein, The plan acquisition unit is configured to acquire the work plan, which includes the work schedule of the work site.

5. The hazard prediction system according to claim 1, wherein, The information acquisition unit is configured to acquire the site information, including the construction drawings of the work site.

6. The hazard prediction system according to claim 1, wherein, The information acquisition unit is configured to acquire the site information, including the detection results of sensors installed at the work site.

7. The hazard prediction system according to claim 1, wherein, The information acquisition unit is configured to acquire the site information, including meteorological information of the work site.

8. The hazard prediction system according to claim 1, wherein, The predictive information acquisition unit acquires the hazard prediction information, including suggestions for hazard prediction training.

9. The hazard prediction system according to claim 1, wherein, The predictive information acquisition unit acquires the hazard prediction information as a result of the language model interpreting hazardous events occurring in other work sites, in addition to the on-site information and the work plan.

Images