Method for generating warning sound for indoor / outdoor mobile robot and system therefor

Abstract

A method and system for generating an alarm sound for an indoor / outdoor driving robot are provided. A method for generating an alarm sound for an indoor / outdoor driving robot according to some embodiments may include the steps of: acquiring driving environment data regarding the driving environment of the indoor / outdoor driving robot; calculating object information including object type information and object state information for each of a plurality of objects identified from the acquired driving environment data; determining an object event based on the calculated object information and calculating the urgency of the determined object event; and generating a robot alarm sound considering the calculated urgency. According to this method, an alarm sound capable of responding to unexpected situations occurring during the driving of the indoor / outdoor driving robot can be generated.

Description

Technology Field

[0001] The present disclosure relates to a technology for generating an alarm sound for an indoor / outdoor driving robot based on type and state information of an object within a driving environment. Background Technology

[0002] Recently, the use of autonomous robots navigating indoor and outdoor environments has been rapidly increasing in various industrial sectors, including logistics, services, and healthcare. As these robots often share the same paths as spaces where humans are active, concerns regarding collisions and safety accidents between robots and people are continuously being raised. In particular, in areas with limited visibility, such as indoor hallways, around elevators, and stairwells, situations where robots and people suddenly encounter each other can occur frequently, further increasing the risk of collision accidents.

[0003] As a solution to this problem, a technology has been proposed in which the robot emits a constant sound while driving to alert people in the vicinity of its presence. However, this method of simply outputting driving sounds has limitations, as it fails to provide a sufficient warning effect in situations where people cannot perceive the robot's sound or are distracted. Furthermore, if an alarm sound is emitted at a constant volume and pattern regardless of ambient noise levels, there is a problem in that the sound may be drowned out by surrounding noise in noisy environments, or conversely, perceived as unnecessary noise in quiet environments, causing discomfort to those nearby.

[0004] In particular, there is a problem in that the risk of collision increases because it is difficult to respond in advance to sudden situations where a person or obstacle suddenly appears on the robot's path. Furthermore, there is a fundamental limitation in that it is impossible to provide sufficient advance response with a simple detection-based reactive alarm output method for situations where a currently stationary object suddenly starts moving, or where a dynamic object appears from a non-visible area around entrance structures such as doors or elevators.

[0005] Therefore, there is a need for technology that goes beyond simply outputting driving sounds and synthesizes and outputs warning sounds in real time alongside the driving sounds to respond to specific events recognized by the robot—such as the sudden appearance of people or obstacles, detection of collision risks, or the presence of stationary objects capable of movement. Furthermore, there is a need for technology that quantitatively calculates the risk level of detected events and dynamically reflects this in the acoustic parameters of the warning sounds, thereby enabling surrounding personnel to more intuitively perceive the robot's current status and dangerous situations, and enhancing safety in the operating environment of mobile robots.

[0006] The aforementioned background technology is one that the inventor possessed or acquired in the process of deriving the content of the present disclosure, and it cannot be considered as prior art disclosed to the general public prior to the filing of this application. The problem to be solved

[0007] The technical problem to be solved through some embodiments of the present disclosure is to provide a method and system for generating a situation-adaptive alarm sound that enables surrounding people and moving objects to more intuitively recognize the presence of the robot and dangerous situations by classifying objects within the driving environment of an indoor or outdoor driving robot by type and calculating event urgency based on the state information of each object.

[0008] The technical problems of the present disclosure are not limited to those mentioned above, and other unmentioned technical problems will be clearly understood by a person skilled in the art of the present disclosure from the description below. means of solving the problem

[0009] A method for generating an alarm sound for an indoor / outdoor driving robot according to some embodiments of the present disclosure for solving the aforementioned technical problem is a method performed by at least one processor, and may include the steps of: acquiring driving environment data regarding the driving environment of an indoor / outdoor driving robot; calculating object information including object type information and object state information for each of a plurality of objects identified from the acquired driving environment data; determining an object event based on the calculated object information and calculating the urgency of the determined object event; and generating a robot alarm sound considering the calculated urgency.

[0010] In some embodiments, the step of acquiring the driving environment data may include the step of acquiring driving environment data including at least one of image data and distance data by using at least one sensor among a camera, a depth camera, and a lidar equipped in the indoor / outdoor driving robot.

[0011] In some embodiments, the step of calculating object information may include the step of identifying a plurality of objects from the acquired driving environment data, the step of calculating object type information for each of the identified plurality of objects, one of a dynamic object, a semi-dynamic object, an entry / exit structure object, and a static object, and the step of calculating object state information for each of the plurality of objects according to the calculated object type information.

[0012] In some embodiments, the step of calculating any one of the object type information comprises: determining whether the identified object is a person; if the identified object is determined to be a person, calculating the object type information of the identified object as a person among dynamic objects; if the identified object is determined not to be a person, calculating the speed of the identified object and, if the calculated speed is greater than 0, calculating the identified object type information as a moving body among dynamic objects; if the calculated speed is 0, determining whether the identified object is a movable object; if the identified object is determined to be a movable object, calculating the object type information of the identified object as a quasi-dynamic object; if the identified object is determined not to be a movable object, determining whether the identified object is a structure through which the dynamic object enters and exits; if the identified object is determined to be a structure through which the dynamic object enters and exits, calculating the object type information of the identified object as an entry / exit structure object; and if the identified object is determined not to be a structure through which the dynamic object enters and exits, the identified It may include a step of producing object type information of an object as a static object.

[0013] In some embodiments, the step of calculating the object state information may include: a step of calculating object state information including the distance between the person and the indoor / outdoor driving robot, the movement speed of the person, the movement direction of the person, and the robot recognition rate of the person when the calculated object type information is a person among dynamic objects; a step of calculating object state information including the distance between the moving object and the indoor / outdoor driving robot, the movement speed of the moving object, the movement direction of the moving object, and the robot recognition rate of the moving object when the calculated object type information is a moving object among dynamic objects; a step of calculating object state information including the distance between the moving object and the indoor / outdoor driving robot and the possibility of movement of the moving object when the calculated object type information is a semi-dynamic object; a step of calculating object state information including the distance between the moving object and the indoor / outdoor driving robot and the possibility of appearance of a dynamic object around the moving object when the calculated object type information is an access structure object; and a step of calculating the distance between the static object and the indoor / outdoor driving robot as object state information when the calculated object type information is a static object.

[0014] In some embodiments, the step of determining the object event and calculating the urgency of the determined object event may include the step of determining the object event as a dynamic object detection event and calculating the urgency of the dynamic object detection event based on object state information for the dynamic object when the calculated object type information is a dynamic object, and the step of determining the object event as a dynamic object prediction event and calculating the urgency of the dynamic object prediction event based on object state information for the dynamic object when the calculated object type information is a semi-dynamic object or an access structure object.

[0015] In some embodiments, the step of calculating the urgency of the dynamic object detection event may include: determining the dynamic object detection event as a person detection event when the dynamic object is a person, and calculating the urgency of the person detection event based on the distance between the person and the indoor / outdoor driving robot, the driving speed of the indoor / outdoor driving robot, the movement speed of the person, the direction of movement of the person, and the robot recognition rate of the person; and determining the dynamic object detection event as a moving object detection event when the dynamic object is a moving object, and calculating the urgency of the moving object detection event based on the distance between the moving object and the indoor / outdoor driving robot, the driving speed of the indoor / outdoor driving robot, the movement speed of the moving object, the direction of movement of the moving object, and the robot recognition rate of the moving object.

[0016] In some embodiments, the step of calculating the urgency of the dynamic object prediction event may include, when the object type information is a quasi-dynamic object, determining the dynamic object prediction event as a quasi-dynamic object movement possibility event and calculating the urgency of the quasi-dynamic object movement possibility event based on the distance between the quasi-dynamic object and the indoor / outdoor driving robot, the driving speed of the indoor / outdoor driving robot, and the movement possibility of the quasi-dynamic object; and when the object type information is an access structure object, determining the dynamic object prediction event as a dynamic object appearance possibility event and calculating the urgency of the dynamic object appearance possibility event based on the distance between the access structure object and the indoor / outdoor driving robot, the driving speed of the indoor / outdoor driving robot, and the possibility of a dynamic object appearing around the access structure object.

[0017] In some embodiments, the step of generating the robot alarm sound may include: analyzing the environmental sound for the driving path of the indoor / outdoor driving robot and selecting an environmental sound source from a plurality of pre-stored candidate environmental sound sources; generating a robot driving alarm sound by adjusting the acoustic parameters of the selected environmental sound source according to the driving speed of the indoor / outdoor driving robot; generating an event alarm sound by adjusting the acoustic parameters of a pre-defined event-specific notification sound source corresponding to the determined object event according to the calculated urgency; and generating a robot alarm sound draft by synthesizing the generated robot driving alarm sound and the generated event alarm sound, and generating a final robot alarm sound by adjusting the acoustic parameters of the generated robot alarm sound draft according to the detected noise level.

[0018] In some embodiments, the step of selecting the environmental sound source comprises: extracting environmental sound characteristics including at least one of a frequency band, volume, and timbre from the environmental sound for the driving path; calculating a suitability score and a distinction score for each of a plurality of previously stored candidate environmental sound sources based on at least one of frequency band similarity, volume difference, timbre similarity, possibility of avoiding masking by ambient noise, transmission suitability, and estimated user discomfort with respect to the extracted environmental sound characteristics; and selecting at least one of an assimilation type environmental sound source and a contrast type environmental sound source based on the calculated suitability score, wherein the assimilation type environmental sound source is a candidate environmental sound source having a suitability score with respect to the extracted environmental sound characteristics of at least a first score, and the contrast type environmental sound source is a candidate environmental sound source having a distinction score with respect to the extracted environmental sound characteristics of at least a second score.

[0019] In some embodiments, the step of generating the robot driving warning sound may include at least one of the following steps: generating the robot driving warning sound by increasing the volume of the selected environment sound source as the driving speed of the indoor / outdoor driving robot increases; generating the robot driving warning sound by shortening the repetition cycle of the selected environment sound source or increasing the tempo of the selected environment sound source as the driving speed of the indoor / outdoor driving robot increases; and generating the robot driving warning sound by increasing the pitch of the selected environment sound source as the driving speed of the indoor / outdoor driving robot increases.

[0020] In some embodiments, the step of generating the event alarm sound includes selecting a predefined event-specific notification sound source corresponding to the determined object event and generating an event alarm sound in which at least one acoustic parameter among the frequency, volume, repetition period, tempo, pitch, and timbre of the selected event-specific notification sound source is adjusted according to the calculated urgency, and the step of generating the event alarm sound with adjusted acoustic parameters may include increasing the volume of the selected event-specific notification sound source, shortening the repetition period, increasing the pitch, or enhancing the timbre that is distinct from the ambient sound as the calculated urgency increases.

[0021] In some embodiments, the step of generating the event alarm sound may further include the step of calculating the expected collision time based on the driving trajectory of the indoor / outdoor driving robot and the predicted movement trajectory of the dynamic object, the step of calculating a correction value for the calculated urgency considering the calculated expected collision time, and the step of generating a final event alarm sound by applying the calculated correction value to the event alarm sound with the acoustic parameters adjusted.

[0022] An alarm sound generation system for an indoor / outdoor driving robot according to some embodiments of the present disclosure for solving the technical problem described above comprises one or more processors and a memory for storing a computer program executed by said one or more processors, and said computer program may include instructions for an operation of acquiring driving environment data regarding the driving environment of the indoor / outdoor driving robot, an operation of calculating object information including object type information and object state information for each of a plurality of objects identified from said acquired driving environment data, an operation of determining an object event based on said calculated object information and calculating the urgency of said determined object event, and an operation of generating a robot alarm sound considering said calculated urgency.

[0023] A computer program according to some embodiments of the present disclosure for solving the aforementioned technical problem may be stored in a computer-readable recording medium to execute the steps of: acquiring driving environment data regarding the driving environment of an indoor / outdoor driving robot, acquiring driving environment data; calculating object information including object type information and object state information for each of a plurality of objects identified from the acquired driving environment data; determining an object event based on the calculated object information and calculating the urgency of the determined object event; and generating a robot alarm sound considering the calculated urgency.

[0024] Other specific details of the present disclosure are included in the detailed description and drawings. Effects of the invention

[0025] According to some embodiments of the present disclosure, by classifying objects within the driving environment of an indoor / outdoor driving robot by type and quantitatively calculating the event urgency, an alarm sound capable of preemptively responding to unexpected situations can be generated.

[0026] In addition, by learning the ambient sounds for each section of the driving path and adaptively adjusting the acoustic parameters of the alarm sound according to real-time ambient noise levels, the transmission effect of the alarm sound can be maximized in various indoor and outdoor environments, and safety in the driving robot operation environment can be improved.

[0027] The effects according to the technical concept of the present disclosure are not limited to those mentioned above, and other unmentioned effects will be clearly understood by a person skilled in the art from the description below. Brief explanation of the drawing

[0028] The attached drawings are examples to aid in understanding the present disclosure, and the present disclosure is not limited to the matters described in the drawings and various modifications are possible. FIG. 1 is an exemplary drawing for explaining the operation of an indoor / outdoor driving robot alarm sound generation system according to some embodiments of the present disclosure at the system level. FIG. 2 is an exemplary drawing for further explaining the operation of an indoor and outdoor driving robot alarm sound generation system according to some embodiments of the present disclosure. FIG. 3 is an exemplary drawing for explaining the functional module configuration of a computing device that performs an indoor / outdoor driving robot alarm sound generation operation according to some embodiments of the present disclosure. FIG. 4 is an exemplary flowchart illustrating a method for generating an indoor and outdoor driving robot alarm sound according to some embodiments of the present disclosure. FIG. 5 is an exemplary drawing for explaining a method for calculating object information according to some embodiments of the present disclosure. FIG. 6 is an exemplary drawing for explaining a method for determining an object event based on object type information according to some embodiments of the present disclosure. FIG. 7 is an exemplary drawing for explaining a method for generating a robot alarm sound according to some embodiments of the present disclosure. FIG. 8 is an exemplary flowchart illustrating a method for generating a robot alarm sound with a collision prediction time taken into account according to some embodiments of the present disclosure. FIG. 9 is an exemplary drawing for illustrating a method for generating a robot alarm sound that takes into account the expected collision time according to some embodiments of the present disclosure. FIG. 10 illustrates an exemplary computing device capable of implementing an alarm sound generation system for an indoor / outdoor driving robot according to some embodiments of the present disclosure. Specific details for implementing the invention

[0029] Hereinafter, various embodiments of the present disclosure will be described in detail with reference to the attached drawings. The advantages and features of the present disclosure and the methods for achieving them will become clear by referring to the embodiments described below in detail together with the attached drawings. However, the technical concept of the present disclosure is not limited to the following embodiments but can be implemented in various different forms. The following embodiments are provided merely to complete the technical concept of the present disclosure and to fully inform those skilled in the art of the scope of the present disclosure, and the technical concept of the present disclosure is defined only by the scope of the claims.

[0030] In describing the various embodiments of the present disclosure, if it is determined that a detailed description of related known configurations or functions could obscure the essence of the present disclosure, such detailed description is omitted.

[0031] Unless otherwise defined, terms used in the following embodiments (including technical and scientific terms) may be used in a meaning commonly understood by those skilled in the art to which this disclosure pertains, but this may vary depending on the intent of those skilled in the art, case law, the emergence of new technology, etc. The terms used in this disclosure are for describing the embodiments and are not intended to limit the scope of this disclosure.

[0032] In the following embodiments, singular expressions include plural concepts unless the context clearly specifies them as singular. Additionally, plural expressions include singular concepts unless the context clearly specifies them as plural.

[0033] In addition, terms such as first, second, A, B, (a), (b), etc. used in the following embodiments are used merely to distinguish one component from another, and the essence, order, or sequence of the said component is not limited by such terms.

[0034] In the following embodiments, the components described with reference to terms such as ~part or unit, module, block, ~or, ~er, and the functional blocks illustrated in the drawings may be implemented in the form of software, hardware, or a combination thereof. Software may be, for example, machine code, firmware, embedded code, and application software. Additionally, hardware may include, for example, electrical circuits, electronic circuits, processors, computers, integrated circuits, integrated circuit cores, passive components, or a combination thereof.

[0035] Hereinafter, various embodiments of the present disclosure will be described in detail with reference to the attached drawings.

[0037] FIG. 1 is an exemplary drawing for explaining the operation of an indoor / outdoor driving robot alarm sound generation system according to some embodiments of the present disclosure at the system level.

[0038] Referring to FIG. 1, the indoor / outdoor driving robot alarm sound generation system (10) according to the embodiments is a computing device / system having a function to generate a robot alarm sound (14) based on driving environment data (13) regarding the driving environment of the indoor / outdoor driving robot (11). For example, the indoor / outdoor driving robot alarm sound generation system (10) can receive driving environment data (13) from the indoor / outdoor driving robot (11), process the data, and provide the generated robot alarm sound (14) to the indoor / outdoor driving robot (11).

[0039] This indoor / outdoor driving robot alarm sound generation system (10) may be named, depending on the case, as an 'alarm sound generation device / system', 'robot alarm sound generation device / system', 'situation-adaptive alarm sound generation device / system', etc. For convenience of explanation, the indoor / outdoor driving robot alarm sound generation system (10) will be abbreviated as 'alarm sound generation system (10)'.

[0040] An indoor / outdoor driving robot (11) is a robot that autonomously drives in indoor and outdoor environments in various industrial fields such as logistics, service, and medical, and may be equipped with one or more sensors (12) for detecting the driving environment. The sensors (12) may include, for example, a camera, a depth camera, LiDAR, etc., and through this, driving environment data (13) including image data, distance data, etc. regarding the driving environment can be acquired. The indoor / outdoor driving robot (11) may transmit the acquired driving environment data (13) to an alarm sound generation system (10) via a network. However, the scope of the present disclosure is not limited thereto, and the driving environment data (13) may further include noise data acquired through a microphone, etc., equipped in the indoor / outdoor driving robot (11).

[0041] Driving environment data (13) is used as input data for the alarm sound generation system (10) to analyze the surrounding environment of the indoor / outdoor driving robot (11) and to generate a robot alarm sound (14). The alarm sound generation system (10) can identify multiple objects from the received driving environment data (13), calculate the type and state information of each identified object, determine object events and urgency based thereon, and generate a robot alarm sound (14) by considering the determined urgency.

[0042] The robot alarm sound (14) is a final alarm sound generated by the alarm sound generation system (10) and can be transmitted to and output to the indoor / outdoor driving robot (11) via a network. The robot alarm sound (14) may be a composite form of a robot driving alarm sound indicating the driving status of the indoor / outdoor driving robot (11) and an event alarm sound reflecting the urgency of a detected or predicted object event. Accordingly, people and moving objects around the indoor / outdoor driving robot (11) can more intuitively perceive the presence of the robot and the dangerous situation.

[0043] As described, the alarm sound generating system (10) and the indoor / outdoor driving robot (11) can communicate through a network. Here, the network can be implemented as any type of wired / wireless network, such as a Local Area Network (LAN), a Wide Area Network (WAN), a mobile radio communication network, a Wireless LAN, etc.

[0044] For reference, in FIG. 1, the alarm sound generation system (10) is depicted as being composed of an indoor / outdoor driving robot (11) and a separate external device, but it is not limited thereto. For example, the function of the alarm sound generation system (10) may be implemented in an on-device form mounted on the indoor / outdoor driving robot (11). Alternatively, some functions of the alarm sound generation system (10) may be performed on the indoor / outdoor driving robot (11) side, and the remaining functions may be configured to be performed on an external server side.

[0045] The above-described alarm sound generation system (10) may be implemented in at least one computing device. For example, all functions of the alarm sound generation system (10) may be implemented in a single computing device, or the first function of the alarm sound generation system (10) may be implemented in a first computing device and the second function may be implemented in a second computing device. Alternatively, specific functions of the alarm sound generation system (10) may be implemented in multiple computing devices. A computing device may encompass any device equipped with computing (processing) functions, and for an example of such a device, refer to FIG. 10.

[0046] FIG. 2 is an exemplary drawing for further explaining the operation of an indoor and outdoor driving robot alarm sound generation system according to some embodiments of the present disclosure.

[0047] Referring to FIG. 2, the alarm sound generation system (10) receives and processes driving environment data (21) as input and finally outputs a robot alarm sound (25). More specifically, the alarm sound generation system (10) can operate in the order of object analysis (22), object event determination (23), and urgency calculation (24).

[0048] In the object analysis (22) operation, the alarm sound generation system (10) identifies multiple objects from the received driving environment data (21) and calculates object type and object state information for each identified object. For example, the alarm sound generation system (10) can analyze image data and distance data, etc. included in the driving environment data (21) to classify objects within the driving environment into one of the following types: people, moving objects, semi-dynamic objects, entry / exit structure objects, and static objects. Additionally, for each classified object, object state information including distance, movement speed, direction of movement, robot recognition rate, possibility of movement, or possibility of dynamic object appearance can be calculated.

[0049] In the object event judgment (23) operation, the alarm sound generation system (10) determines an object event based on the object type and object state information calculated in the object analysis (22) step. The events subject to judgment can be broadly classified into dynamic object detection events and dynamic object prediction events. A dynamic object detection event is an event that occurs when an object that is currently moving or has a possibility of collision is detected, and a dynamic object prediction event is an event that occurs when there is a possibility that an object will start moving in the future or that a dynamic object will appear from a non-visible area.

[0050] In the operation of calculating the urgency (24), the alarm sound generation system (10) quantitatively calculates the urgency of the determined object event and generates a robot alarm sound (25) by taking into account the calculated urgency. The urgency can be calculated by comprehensively considering, for example, the distance between the object and the indoor / outdoor driving robot, the driving speed of the indoor / outdoor driving robot, the movement speed of the object, the direction of movement of the object, and the robot recognition rate of the object. The calculated urgency is dynamically reflected in the acoustic parameters (e.g., volume, pitch, repetition period, tempo, etc.) of the robot alarm sound (25).

[0051] FIG. 3 is an exemplary drawing for explaining the functional module configuration of a computing device that performs an indoor / outdoor driving robot alarm sound generation operation according to some embodiments of the present disclosure.

[0052] Referring to FIG. 3, the alarm sound generation system (10) may include, in terms of function, an object information calculation unit (31), an event detection unit (32), a robot alarm sound generation unit (33), and a sound source selection unit (34).

[0053] The alarm sound generation system (10) receives driving environment data obtained from a sensor (12) equipped in an indoor / outdoor driving robot (11) as input. The driving environment data may include image data and distance data obtained through, for example, a camera, a depth camera, or a lidar, and may further include noise data obtained through a microphone, etc.

[0054] The object information calculation unit (31) is a module that identifies multiple objects from the received driving environment data and calculates object type and object state information for each identified object. The object information calculation unit (31) may include, for example, a segmentation model and a classifier.

[0055] A segmentation model is a model that separates different object regions from driving environment data into independent object units. For example, a segmentation model can be implemented based on the Segment Anything Model (SAM) and can separate all regions within the input data into independent object units without relying on specific pre-trained classes.

[0056] Specifically, the segmentation model receives image data acquired through a camera, distance data acquired through a depth camera or LiDAR as input, and can generate pixel segments that distinguish which object each region belongs to at the pixel or point level within the input data. For example, the segmentation model can separate regions such as people, bicycles, carts, doors, and walls within an image into distinct object candidate regions.

[0057] In this way, the segmentation model is not dependent on a specific class and independently separates all regions within the input data, thereby having the advantage of being able to separate new types of objects that are not previously defined into distinct object candidate regions even if they appear in the driving environment. However, the scope of the present disclosure is not limited thereto, and the segmentation model may be implemented based on various segmentation algorithms or deep learning models.

[0058] A classifier is a model that determines the object type for each object candidate region separated by a segmentation model. For example, the classifier can operate connected to the end of the segmentation model and can determine the type of the corresponding object by receiving pixel segments of each object candidate region and corresponding distance data as input.

[0059] Specifically, the classifier first determines whether the object is a person, and if it is determined not to be a person, calculates the speed of the object and classifies it as a moving object if the speed is greater than 0, as a movable object if the speed is 0, as a semi-dynamic object or an access structure object if it is not movable, and as a static object if it is not movable.

[0060] For example, objects currently in motion, such as electric scooters, bicycles, carts, forklifts, wheelchairs, and strollers, can be classified as moving objects; objects that are movable but currently stationary, such as parked vehicles, stationary electric scooters, and waiting carts, can be classified as semi-dynamic objects; structures through which dynamic objects pass, such as stairs, doors, and elevators, can be classified as entrance structures; and structures with fixed positions, such as walls and floor ledges, can be classified as static objects.

[0061] In addition, the classifier can perform additional learning depending on changes in the operating environment or detection targets. Specifically, the additional learning of the classifier can be performed through the following process.

[0062] First, the classifier can identify unclassified objects among the object candidate regions separated by the segmentation model from the driving environment data that are not classified into existing defined object classes. When an unclassified object is identified, the alarm sound generation system (10) can collect driving environment data (e.g., image data, distance data, etc.) for the unclassified object and store the collected data as training data for a new object class.

[0063] Next, the alarm sound generation system (10) may assign a label for a new object class to the stored training data and retrain the classifier by adding the labeled training data to the existing training dataset. At this time, the retraining may be performed using a full retraining method using the entire training dataset, or it may be performed using a fine-tuning method that additionally trains the classifier using only the new training data while maintaining the parameters of the existing classifier.

[0064] In addition, if the classification criteria of an existing object class are to be changed, the alarm sound generation system (10) can retrain the classifier by updating existing training data for the object class or supplementing additional training data. For example, if a history has accumulated where an object classified as a semi-dynamic object in a specific operating environment frequently switches to a moving object, the classification criteria for the object can be updated to retrain the classifier so that a more accurate classification is achieved.

[0065] In this way, the classifier can add new object classes or change the classification criteria of existing object classes through additional learning, and accordingly, the alarm sound generation system (10) can flexibly respond to changes in various operating environments and requirements. The object type and object state information calculated by the object information calculation unit (31) is transmitted to the event detection unit (32).

[0066] The event detection unit (32) is a module that determines an object event based on object type and object state information received from the object information calculation unit (31) and calculates the urgency of the determined object event. The object event and urgency information determined by the event detection unit (32) are transmitted to the robot alarm sound generation unit (33).

[0067] The robot alarm sound generation unit (33) is a module that generates and outputs a robot alarm sound based on object event and urgency information received from the event detection unit (32). The robot alarm sound generation unit (33) can generate a robot driving alarm sound based on environmental sounds for each section of the driving path and an event alarm sound based on event urgency, respectively, and synthesize them to output a final robot alarm sound. In addition, the robot alarm sound generation unit (33) can correct the acoustic parameters of the final robot alarm sound according to the ambient noise level detected in real time.

[0068] The sound source selection unit (34) is a module that interacts with the robot alarm sound generation unit (33) to support the selection and updating of environmental sound sources. The sound source selection unit (34) can be used to select assimilated environmental sound sources and contrasting environmental sound sources suitable for the environmental sound characteristics of each section of the driving path from a plurality of pre-stored candidate environmental sound sources. Here, candidate environmental sound sources refer to sound sources that are pre-produced or secured and stored from an external sound source database, and may be named as a 'candidate sound source library', 'environmental sound source candidate group', etc.

[0069] Specifically, the environment sound source selection process of the sound source selection unit (34) can be performed as follows.

[0070] First, the sound source selection unit (34) can utilize ambient sound data collected by section through sensors such as microphones while the indoor / outdoor driving robot (11) is driving along the driving path as a subject for analysis. At this time, the ambient sound data is collected and stored by section along the driving path, and as the amount of ambient sound data accumulated for the same section increases, the acoustic characteristics of that section can be reflected more accurately. In addition, the ambient sound data can be collected and stored separately by time period, and accordingly, even for the same section, changes in acoustic characteristics according to time period can be reflected in the selection of the ambient sound source.

[0071] Next, the sound source selection unit (34) can extract acoustic characteristics such as the main frequency band, volume, and timbre of the corresponding section from the collected environmental sound data. For example, in a quiet space such as a hallway or office, a low frequency band (100~500Hz) can be extracted as the main frequency band, and in a noisy space such as a factory or outdoors, a mid-to-high frequency band (1kHz~4kHz) can be extracted as the main frequency band.

[0072] Based on the extracted acoustic characteristics, the sound source selection unit (34) calculates a suitability score and a distinguishability score for each of the multiple candidate environmental sound sources, and can select an assimilated environmental sound source and a contrasting environmental sound source, respectively, according to the calculated suitability score and the calculated distinguishability score. The assimilated environmental sound source is a candidate sound source having a frequency band, volume, and timbre similar to the extracted acoustic characteristics, and can be used to announce the presence of the robot in a way that naturally blends with the atmosphere of the space.

[0073] On the other hand, contrasting environmental sound sources are candidate sound sources possessing frequency bands, volume, and timbres that are clearly distinct from the extracted acoustic characteristics, and can be utilized to strongly induce auditory attention so that surrounding people and moving objects clearly perceive the robot's approach. For example, in a quiet space, a candidate sound source with a volume higher than a certain decibel (dB) and short, regular beats compared to the ambient sound can be selected as a contrasting environmental sound source, while in a noisy space, a candidate sound source in the high-frequency range (4 kHz or higher) that is distinct from the main frequency band of the surrounding noise can be selected as a contrasting environmental sound source.

[0074] Additionally, the sound source selection unit (34) can re-select and update the corresponding animated and contrasting environmental sound sources from the candidate sound source library whenever environmental sound data for a new section is collected or environmental sound data for an existing section is updated. Accordingly, the alarm sound generation system (10) can continuously provide environmental sound sources suitable for the environment even when the indoor / outdoor driving robot (11) travels a new driving path or when the acoustic environment of the existing driving path changes.

[0075] To reduce the burden of on-device real-time computation, the animated and contrasting environmental sound sources selected by the sound source selection unit (34) can be stored in advance for each section. During actual driving, the stored environmental sound sources can be recalled, and the system can be operated by adjusting only acoustic parameters such as volume, pitch, and tempo in real time according to the robot's current driving speed, event urgency, or the level of ambient noise detected in real time.

[0076] For example, when the driving speed of a robot increases, the volume of the environmental sound source may increase or the tempo may increase, and when the level of ambient noise is high, the volume of the environmental sound source may increase or the proportion of contrasting environmental sound sources may increase. In this way, by adopting a method of selecting and storing environmental sound sources from a candidate sound source library, there is an advantage of being able to efficiently generate and output situation-appropriate alarm sounds in real time, even in an on-device environment.

[0077] Up to now, the operation of an alarm sound generating system (10) according to some embodiments of the present disclosure has been described schematically with reference to FIGS. 1 to 3. Hereinafter, various methods that can be performed in the alarm sound generating system (10) will be described in more detail with reference to FIGS. 4 and subsequent drawings.

[0078] For the sake of convenience of understanding, the following description will continue under the assumption that all steps / operations of the methods described below are performed in an alarm sound generation system (10, e.g., at least one processor). Therefore, if the subject of a specific step / operation is omitted, it can be understood that the step / operation is performed by the alarm sound generation system (10). However, in an actual environment, some steps / operations of the methods described below may be performed on other computing devices.

[0080] FIG. 4 is an exemplary flowchart illustrating a method for generating an indoor / outdoor driving robot alarm sound according to some embodiments of the present disclosure. However, this is merely an exemplary embodiment for achieving the purpose of the present disclosure, and it is understood that some steps may be added or deleted as necessary.

[0081] As illustrated in FIG. 4, the method for generating an alarm sound according to the embodiments may begin at step S41, which involves acquiring driving environment data regarding the driving environment of an indoor / outdoor driving robot. For example, the alarm sound generation system (10) may acquire driving environment data regarding the driving environment through one or more sensors (12) provided in the indoor / outdoor driving robot (11). As described above, the driving environment data may include at least one of image data, distance data, and noise data.

[0082] In some embodiments, the alarm sound generating system (10) may acquire driving environment data including at least one of image data and distance data by using at least one sensor among a camera, a depth camera, and a lidar equipped on an indoor / outdoor driving robot (11). For example, image data regarding the driving environment may be acquired through a camera, and distance data with respect to an object within the driving environment may be acquired through a depth camera or lidar.

[0083] In addition, noise data regarding environmental sounds for each section of the driving path can be acquired through a microphone. At this time, the driving environment data acquired through multiple sensors can be integrated and processed using a sensor fusion technique, thereby enabling more accurate recognition of the driving environment.

[0084] In some other embodiments, the alarm sound generating system (10) may additionally receive driving environment data from external sensors (e.g., CCTV, fixed LiDAR, etc.) installed in the driving environment, in addition to the sensor (12) provided in the indoor / outdoor driving robot (11). In this case, the driving environment data obtained from the sensor (12) provided in the indoor / outdoor driving robot (11) and the driving environment data received from the external sensors can be integrated and processed through a sensor fusion technique.

[0085] Specifically, sensor fusion can be performed through the following process. First, the alarm sound generation system (10) can perform coordinate system alignment on driving environment data received from each of the sensors (12) equipped in the indoor / outdoor driving robot (11) and the external sensors. For example, since the external sensors (e.g., fixed LiDAR, CCTV, etc.) are installed at different locations and directions from the indoor / outdoor driving robot (11), coordinate transformation processing can be performed to convert the driving environment data received from each sensor into a common coordinate system. At this time, the coordinate transformation can be performed based on the location and direction information of each sensor (e.g., installation location of the external sensor, current location and direction of the indoor / outdoor driving robot (11), etc.).

[0086] Next, the alarm sound generation system (10) can perform time synchronization on multiple driving environment data in which coordinate systems are aligned. For example, since the sensor (12) equipped in the indoor / outdoor driving robot (11) and the external sensor can each collect data at different sampling periods, a process can be performed to align the data of each sensor based on the same time standard according to the timestamp.

[0087] Next, the alarm sound generation system (10) can generate integrated driving environment data by fusing multiple time-synchronized driving environment data. The data fusion method may vary depending on the embodiment. For example, in the initial fusion method, raw data collected from each sensor can be directly combined to create a single integrated data and then used for object identification. In the later fusion method, objects can be identified independently from each sensor data, and then the identified object information can be combined to generate final object information. For example, integrated object information can be generated by comparing object information identified from the sensor (12) of the indoor / outdoor driving robot (11) with object information identified from an external sensor, weighting the average of the information regarding the same object based on reliability, and adding object information identified from only one sensor complementarily. In the intermediate fusion method, features extracted from each sensor data can be combined and used for object identification.

[0088] In this way, by integrating and processing driving environment data obtained from a sensor (12) equipped in an indoor / outdoor driving robot (11) and driving environment data received from an external sensor through a sensor fusion technique, it is possible to supplement driving environment data for non-visible areas (e.g., beyond corners, around entrance / exit structures, etc.) that are difficult to detect with only the sensor (12) of the indoor / outdoor driving robot (11), and there is an advantage that more accurate object identification and urgency calculation are possible.

[0089] In step S42, object information including object type information and object state information is calculated for each of the plurality of objects identified from the acquired driving environment data. For example, the alarm sound generation system (10) identifies each of the plurality of objects from the acquired driving environment data, calculates one of the object type information (e.g., one of the object type information of a dynamic object, a semi-dynamic object, an entry / exit structure object, and a static object) for each of the identified plurality of objects, and calculates object state information for each of the plurality of objects according to the calculated object type information.

[0090] In some embodiments, a segmentation model and a classifier may be used to identify multiple objects from driving environment data and to produce one of the object type information. For example, as illustrated in FIG. 5, an alarm sound generation system (10) inputs driving environment data (51) into a SAM (Segment Anything Model, 52) to generate pixel segments separated by object units, and inputs the generated pixel segments into a classifier (53) to determine the type of each object. The classifier (53) may classify each object into one of the following types: a dynamic object (54), a semi-dynamic object (55), an entry / exit structure object (56), and a static object (57). At this time, the dynamic object (54) may be subdivided into a person (54-1) and a moving object (54-2).

[0091] Specifically, the classifier (53) first determines whether the identified object is a person, and if it is determined to be a person, it classifies the object as a person (54-1) among dynamic objects. If the identified object is determined not to be a person, it calculates the speed of the identified object, and if the calculated speed is greater than 0, it classifies it as a moving object (54-2). If the calculated speed is 0, it further determines whether the identified object is a movable object, and if the identified object is determined to be a movable object, it classifies it as a semi-dynamic object (55). If the identified object is not movable, it determines whether it is a structure that a dynamic object enters and exits, and classifies it as an entry structure object (56) or a static object (57), respectively.

[0092] More specifically, the classifier (53) can determine the object type for each object candidate region separated by the segmentation model through the following process.

[0093] First, the classifier (53) determines whether the identified object is a person. For example, the classifier (53) can extract morphological features such as the silhouette, skeleton structure, and body proportions of the object from image data acquired through a camera, and input the extracted features into a pre-trained human classification model (e.g., pose estimation model, pedestrian detection model, etc.) to determine whether the object is a person.

[0094] For example, if a skeletal structure corresponding to a body part such as the head, shoulders, arms, and legs is detected, the object may be identified as a person. If it is identified as a person, the classifier (53) calculates the type of the object as a person (54-1) among dynamic objects.

[0095] If it is determined that it is not a person, the classifier (53) calculates the speed of the object to determine whether it is a moving object. The speed can be calculated by dividing the change in position of the object between consecutive frames by time, based on distance data obtained through a depth camera or lidar. If the calculated speed is greater than 0, the classifier (53) determines the type of the object as a moving object (54-2). For example, an object currently in motion, such as an electric scooter, bicycle, cart, forklift, wheelchair, stroller, etc., can be classified as a moving object (54-2).

[0096] When the calculated speed is 0, the classifier (53) determines whether the object is a movable object. For example, the classifier (53) can determine whether the object is movable by comprehensively considering the object's morphological features (e.g., presence or absence of wheels, presence or absence of handles, etc.), material characteristics (e.g., metal, plastic, etc.) and similarity to a pre-learned list of movable object classes (e.g., vehicle, scooter, cart, wheelchair, stroller, etc.).

[0097] Additionally, the classifier (53) may determine whether movement is possible by additionally referring to movement history information of the object or section learned from past driving data. If it is determined to be a movable object, the classifier (53) calculates the type of the object as a semi-dynamic object (55). For example, a parked vehicle, a stationary electric scooter, a waiting cart, etc., may be classified as a semi-dynamic object (55).

[0098] If it is determined that the object is not a movable object, the classifier (53) determines whether the object is a structure through which a dynamic object enters and exits. For example, the classifier (53) can determine whether the object is a structure through which an object exits and exits based on the morphological features of the object (e.g., door frame structure, whether it is open or closed, elevator structure, etc.) and similarity with a pre-learned list of entry structure object classes (e.g., door, stairs, elevator, etc.).

[0099] Additionally, the classifier (53) may determine whether it is an access structure by additionally referring to whether there is a history of a dynamic object appearing in the vicinity of the object in the past. If it is determined to be an access structure, the classifier (53) calculates the type of the object as an access structure object (56). For example, stairs, doors, elevators, etc., may be classified as access structure objects (56).

[0100] If it is determined that the object is not an access structure, the classifier (53) calculates the type of the object as a static object (57). A static object (57) is a structure that has a fixed position and cannot be moved, such as a wall, a floor ledge, or a fixed fixture.

[0101] In this way, the classifier (53) can calculate the type of each object through a hierarchical classification process that sequentially determines whether the identified object is a person, speed, mobility, and whether it is an access structure. Through this hierarchical classification process, the classifier (53) can classify various objects within the driving environment more accurately, and accordingly, improve the accuracy of subsequent event judgment and urgency calculation.

[0102] When the type of each object is determined by the classifier (53), the alarm sound generation system (10) can calculate the object state information of each object according to the calculated object type information.

[0103] When the calculated object type information is a person (54-1) among dynamic objects, the alarm sound generation system (10) can calculate object state information including the distance between the person and the indoor / outdoor driving robot (11), the person's movement speed, movement direction, and robot recognition rate. Specifically, the distance can be calculated by extracting the depth value of the pixel segment corresponding to the person object from distance data obtained through a depth camera or LiDAR. The movement speed can be calculated by dividing the amount of change in the position of the person object between consecutive frames by time, and the movement direction can be calculated from the vector of the position change of the person object between consecutive frames. The robot recognition rate can be calculated by analyzing the person's gaze direction, head movement, change in movement direction, and change in movement speed from video data obtained through a camera. For example, the head direction and gaze vector of the person can be extracted through a pose estimation model, and if the extracted gaze vector is directed toward the indoor / outdoor driving robot (11), the robot recognition rate can be calculated as high. Conversely, if the gaze vector is directed in the opposite direction to the indoor / outdoor driving robot (11) or if no avoidance movement is detected, the robot recognition rate may be calculated as low. Additionally, if the direction of movement of the person is away from the indoor / outdoor driving robot (11), the robot recognition rate may be calculated as high, and if the direction is approaching the indoor / outdoor driving robot (11), the robot recognition rate may be calculated as low.

[0104] When the calculated object type information is a moving object (54-2), the alarm sound generation system (10) can calculate object state information including the distance between the moving object and the indoor / outdoor driving robot (11), the moving speed of the moving object, the direction of movement, and the robot recognition rate. Specifically, the distance, moving speed, and direction of movement can be calculated from distance data and position changes between consecutive frames, similar to the case of a human. Since it is difficult to directly analyze the gaze or head movements of the moving object unlike a human, the robot recognition rate can be calculated by analyzing the speed change of the moving object (e.g., whether it decelerates), the direction change of movement (e.g., whether it changes direction), and the relative movement trajectory with the indoor / outdoor driving robot (11). For example, if the moving object decelerates or changes direction in response to the approach of the indoor / outdoor driving robot (11), the robot recognition rate can be calculated as high, and if it continuously approaches the indoor / outdoor driving robot (11) without changes in speed and direction, the robot recognition rate can be calculated as low.

[0105] When the calculated object type information is a semi-dynamic object (55), the alarm sound generation system (10) can calculate object state information including the distance and possibility of movement between the semi-dynamic object and the indoor / outdoor driving robot (11). Specifically, the distance can be calculated from distance data. The possibility of movement can be calculated by comprehensively considering the type of the semi-dynamic object (e.g., vehicle, kickboard, cart, etc.), the past movement history of the object or section (e.g., frequency of starting movement at a specific time), and current time information. For example, if a cart that was stopped at a specific location has a history of frequently starting movement between 9:00 AM and 10:00 AM and the current time corresponds to that time period, the possibility of movement of the semi-dynamic object can be calculated as high. In addition, if an object identified as a semi-dynamic object subsequently converts into a moving object, the alarm sound generation system (10) can store the location and time information of the conversion event as training data and reflect this in the calculation of the possibility of movement in the same or similar situation in the future.

[0106] When the calculated object type information is an access structure object (56), the alarm sound generation system (10) can calculate object state information including the distance between the access structure object and the indoor / outdoor driving robot (11) and the possibility of dynamic objects appearing around the access structure object. Specifically, the distance can be calculated from distance data. The possibility of dynamic objects appearing can be calculated by comprehensively considering the type of the access structure object (e.g., door, stairs, elevator, etc.), the history of dynamic objects appearing around the access structure object in the past (e.g., frequency of dynamic objects appearing at a specific time), current time information, and changes around the access structure object between the previous frame and the current frame (e.g., whether the door is open or closed, whether the elevator door is open or closed, etc.). For example, if the opening of the elevator door is detected or there is a history of dynamic objects appearing frequently around the access structure object at a specific time, the possibility of dynamic objects appearing around the access structure object can be calculated as high. Additionally, whenever a dynamic object appearance event occurs, the alarm sound generation system (10) can perform learning by updating the probability of dynamic objects appearing at different times for the access structure object.

[0107] When the calculated object type information is a static object (57), the alarm sound generation system (10) can calculate the distance between the static object and the indoor / outdoor driving robot (11) as object state information. The distance can be calculated by extracting the depth value of the pixel segment corresponding to the static object from the distance data. Since the static object (57) is a structure that cannot move, additional state information other than the distance may not be calculated.

[0108] In some other embodiments, instead of using a segmentation model and a classifier, an object detection model may be used to identify multiple objects from driving environment data and generate object type information.

[0109] An object detection model refers to a deep learning model capable of simultaneously outputting the location (e.g., bounding box) and type (e.g., class label) of an object from input data (e.g., image data). Examples of object detection models include YOLO (You Only Look Once), Faster R-CNN (Region-based Convolutional Neural Network), SSD (Single Shot MultiBox Detector), and DETR (Detection Transformer).

[0110] For example, YOLO-based models have the advantage of being suitable for real-time processing as they operate in a single-stage manner, dividing input images into grid units and simultaneously predicting the presence of an object, bounding box coordinates, and class labels in each grid cell. Faster R-CNN-based models have the advantage of high detection accuracy as they operate in a two-stage manner, first generating object candidate regions through a Region Proposal Network (RPN) and then performing class classification and bounding box regression for each candidate region. However, the scope of this disclosure is not limited thereto, and various object detection models may be applied.

[0111] The process for calculating object type information using an object detection model can be performed as follows. First, the alarm sound generation system (10) inputs image data acquired through a camera into the object detection model. The object detection model outputs the bounding box coordinates and class labels (e.g., person, bicycle, cart, door, elevator, wall, etc.) of each object from the input image data. At this time, since the object detection model determines the type of each object based on a pre-trained class list, the class list can be designed to correspond to object types defined in the present disclosure (e.g., person, moving object, semi-dynamic object, access structure object, static object).

[0112] Next, the alarm sound generation system (10) can calculate the distance between the object and the indoor / outdoor driving robot (11) by mapping the distance data corresponding to the bounding box of each object from the output result of the object detection model with the distance data obtained from the depth camera or lidar. In addition, the alarm sound generation system (10) can calculate the movement speed of the object based on the change in position and distance of the center point of the bounding box of each object between consecutive frames.

[0113] Next, the alarm sound generation system (10) can finally calculate object type information for each object based on the output class label of the object detection model and the calculated movement speed. For example, if the class label is person, it can be calculated as a person (54-1) among dynamic objects; if the class label is bicycle, cart, kickboard, etc. and the movement speed is greater than 0, it can be calculated as a moving body (54-2); and if the movement speed is 0, it can be calculated as a semi-dynamic object (55). Additionally, if the class label is door, stairs, elevator, etc., it can be calculated as an entrance structure object (56); and if the class label is wall, floor ledge, etc., it can be calculated as a static object (57).

[0114] Meanwhile, the object detection model may perform additional learning (e.g., fine-tuning) in accordance with changes in the operating environment or detection targets, thereby adding new object classes or improving the detection accuracy of existing object classes. However, the scope of the present disclosure is not limited thereto, and various methods for identifying multiple objects from driving environment data and calculating object type information may be applied.

[0115] In step S43, an object event is determined based on the calculated object information, and the urgency of the determined object event is calculated. For example, if the calculated object type information is a dynamic object, the alarm sound generation system (10) determines the object event as a dynamic object detection event and calculates the urgency of the dynamic object detection event based on the object state information for the dynamic object, and if the calculated object type information is a semi-dynamic object or an access structure object, determines the object event as a dynamic object prediction event and calculates the urgency of the dynamic object prediction event based on the object state information for the semi-dynamic object or the access structure object.

[0116] Here, the object events subject to judgment can be broadly classified into dynamic object detection events and dynamic object prediction events. Dynamic object detection events occur when an object that is currently moving or has a potential for collision is detected, while dynamic object prediction events occur when there is a possibility that an object will begin moving or a dynamic object will emerge from an invisible area in the future, even if a direct collision object is not currently detected.

[0117] In some embodiments, an object event is determined based on object type information calculated by a classifier (61), and the urgency of the determined object event can be calculated. For example, as shown in FIG. 6, the alarm sound generation system (10) first determines whether the object is a dynamic object based on object type information calculated from the classifier (61) (62-1). If the calculated object type information is a dynamic object, the alarm sound generation system (10) further determines whether the dynamic object is a person (62-2), and if it is a person, determines the object event as a person detection event (65-1), and if it is a non-person moving object, determines it as a moving object detection event (65-2).

[0118] If the generated object type information is not a dynamic object, the alarm sound generation system (10) determines whether the object is a semi-dynamic object (63), and if it is a semi-dynamic object, determines the object event as a semi-dynamic object movement possibility event (66).

[0119] If the generated object type information is not a quasi-dynamic object, the alarm sound generation system (10) determines whether the object is an access structure object (64), and if it is an access structure object, determines the object event as a dynamic object appearance possibility event (67). An urgency calculation (68) is performed according to the determined object event, and the calculated urgency is used to generate a robot alarm sound (69).

[0120] The process for calculating the urgency for each object event can be performed as follows.

[0121] The urgency of a person detection event (65-1) can be calculated by comprehensively considering the distance between the person and the indoor / outdoor driving robot (11), the current driving speed of the indoor / outdoor driving robot (11), the movement speed of the person, the direction of movement, and the robot recognition rate. Specifically, each parameter can be calculated as follows.

[0122] The distance (d) can be calculated by extracting the depth value of the pixel segment corresponding to the person object from the distance data obtained through the depth camera or lidar. The driving speed (vr) of the indoor / outdoor driving robot (11) can be calculated using the value obtained from the speed sensor (e.g., encoder, IMU (Inertial Measurement Unit), etc.) equipped in the indoor / outdoor driving robot (11).

[0123] The movement speed (vp) of the person can be calculated by dividing the change in position of the person object between consecutive frames by the time interval between frames. The direction of movement (θp) of the person can be calculated from the vector of the change in position of the person object between consecutive frames and can be expressed as the angle difference with the direction of movement vector of the indoor / outdoor driving robot (11).

[0124] The robot recognition rate (a) can be calculated as a value greater than 0 and less than 1 by using a pose estimation model to extract the gaze vector and head direction of the person in question from image data obtained through a camera, and by comprehensively analyzing the angle difference between the extracted gaze vector and the direction vector of the indoor / outdoor driving robot (11), the rate of change in the direction of movement of the person in question, and the rate of change in the speed of movement.

[0125] Based on each calculated parameter, the urgency (U_person) of a person detection event can be calculated, for example, by the following formula: U_person = w1 × (1 / d) + w2 × vr + w3 × vp + w4 × cos(θp) + w5 × (1 - a). Here, the urgency can be calculated as high if the distance (d) is less than a preset threshold distance (d_th), if the direction of movement (θp) is less than or equal to a preset threshold angle (θ_th) (i.e., if the direction of movement of the person is toward the indoor / outdoor driving robot (11)), or if the robot recognition rate (a) is less than a preset threshold recognition rate (a_th). Each weight (w1 to w5) can be adjusted according to the operating environment and requirements.

[0126] The urgency of the moving object detection event (65-2) can be calculated by comprehensively considering the distance between the moving object and the indoor / outdoor driving robot (11), the current driving speed of the indoor / outdoor driving robot (11), the moving speed of the moving object, the direction of movement, and the robot recognition rate.

[0127] Specifically, the distance (d) and the driving speed (vr) of the indoor / outdoor driving robot (11) can be calculated in the same way as in the case of a person detection event. The movement speed (vo) and movement direction (θo) of the corresponding moving body can be calculated from the amount of position change and the position change vector of the corresponding moving body between consecutive frames, respectively.

[0128] Since it is difficult to directly analyze gaze or head movements of a moving object unlike humans, the robot recognition rate (a) can be calculated by analyzing the rate of change of velocity (dvo / dt) and the rate of change of direction of movement (dθo / dt) of the moving object between consecutive frames.

[0129] For example, if the rate of change in speed decreases above a preset threshold (dvo_th / dt) or the rate of change in the direction of movement changes above a preset threshold (dθo_th / dt), it is determined that the moving body recognizes and avoids the indoor / outdoor driving robot (11), and the robot recognition rate can be calculated as high. The urgency (U_moving) of the moving body detection event can be calculated using a formula such as, for example, U_moving = w1 × (1 / d) + w2 × vr + w3 × vo + w4 × cos(θo) + w5 × (1 - a).

[0130] The urgency of the semi-dynamic object movement possibility event (66) can be calculated by comprehensively considering the movement possibility of the semi-dynamic object, the distance between the semi-dynamic object and the indoor / outdoor driving robot (11), and the current driving speed of the indoor / outdoor driving robot (11).

[0131] Specifically, the distance (d) and the driving speed (vr) of the indoor / outdoor driving robot (11) can be calculated in the same way as described above. The possibility of movement (p_move) can be calculated as a probability value between 0 and 1 based on the type of the semi-dynamic object, the object, or the past movement history in the section (e.g., the number of times movement started at a specific time / the total number of observations).

[0132] In addition, as the history of an object identified as a semi-dynamic object subsequently actually being converted into a moving object accumulates, the likelihood of movement can be updated to a higher level. The urgency (U_semi) of a semi-dynamic object movement likelihood event can be calculated using a formula such as, for example, U_semi = w1 × (1 / d) + w2 × vr + w3 × p_move, and the event can be determined to have occurred if the likelihood of movement (p_move) is greater than or equal to a preset threshold probability (p_th) and the distance (d) is less than a preset threshold distance (d_th).

[0133] The urgency of the dynamic object appearance possibility event (67) can be calculated by comprehensively considering the possibility of a dynamic object appearing around the entrance structure object, the distance between the entrance structure object and the indoor / outdoor driving robot (11), and the current driving speed of the indoor / outdoor driving robot (11).

[0134] Specifically, the distance (d) and the driving speed (vr) of the indoor / outdoor driving robot (11) can be calculated in the same way as described above. The probability of a dynamic object appearing (p_appear) can be calculated as a probability value of 0 or greater and 1 or less based on the type of the entrance / exit structure object (e.g., door, stairs, elevator, etc.), the history of the appearance of a dynamic object in the past around the entrance / exit structure object (e.g., the number of times a dynamic object appeared in a specific time period / the total number of observations), current time period information, and the current state of the entrance / exit structure object (e.g., whether the door or elevator door is open or closed).

[0135] For example, when the opening of an elevator door is detected, the probability of a dynamic object appearing can be immediately raised above a preset threshold. The urgency (U_appear) of the probability of a dynamic object appearing event can be calculated using a formula such as, for example, U_appear = w1 × (1 / d) + w2 × vr + w3 × p_appear, and if the probability of a dynamic object appearing (p_appear) is greater than or equal to a preset threshold probability (p_th) and the distance (d) is less than a preset threshold distance (d_th), it can be determined that the event has occurred.

[0136] Meanwhile, if the urgency is calculated using a weighted sum method in which weights are assigned to each parameter, each weight (w1 to w5) may be adjusted according to the operating environment and requirements. However, the scope of the present disclosure is not limited thereto, and the urgency may be calculated using a machine learning model (e.g., regression model, neural network model, etc.) or using fuzzy logic-based rules.

[0137] In some other embodiments, when calculating the urgency, the Time to Collision (TTC) may be additionally calculated based on the driving trajectory of the indoor / outdoor driving robot (11) and the movement trajectory of the dynamic object, and the calculated Time to Collision may be reflected in the calculation of the urgency. For example, the shorter the Time to Collision, the higher the urgency may be calculated, and if the Time to Collision is long, the urgency may be calculated relatively low. This will be described later with reference to FIGS. 8 and 9.

[0138] In some other embodiments, the urgency may be calculated by dividing it into multiple levels rather than a single number. Specifically, the urgency level may be divided into four stages, for example, low, medium, high, and very high, and a threshold value corresponding to each level may be set in advance.

[0139] For example, the calculated urgency value (U) may be determined as a low level if it is less than the first threshold (U_th1), a normal level if it is greater than or equal to the first threshold (U_th1) but less than the second threshold (U_th2), a high level if it is greater than or equal to the second threshold (U_th2) but less than the third threshold (U_th3), and a very high level if it is greater than or equal to the third threshold (U_th3) (e.g., U_th1 < U_th2 < U_th3).

[0140] Each threshold (U_th1, U_th2, U_th3) can be adjusted according to the operating environment and requirements. For example, in environments with frequent movement of objects, such as logistics centers, the threshold can be raised to reduce the occurrence of unnecessary high-urgency events, while in environments where pedestrian safety is critical, such as hospitals or schools, the threshold can be lowered to generate alarm sounds more sensitively.

[0141] In addition, the urgency level is not determined solely by the urgency value at a single point in time, but may also be determined based on the moving average or exponentially weighted moving average of urgency values ​​calculated over multiple consecutive frames. Accordingly, the impact of abrupt fluctuations in urgency values ​​caused by transient sensor noise or measurement errors on the determination of the urgency level can be reduced.

[0142] The determined urgency level can be utilized to adjust the acoustic parameters of the alarm sound during the subsequent alarm sound generation stage. For example, at a low level, the volume of the event alarm sound can be minimized or the repetition period set to a longer duration; at a normal level, predefined basic acoustic parameters can be applied; at a high level, the volume can be increased, the repetition period shortened, or the pitch raised; and at a very high level, the event alarm sound can be generated by applying maximum volume, the shortest repetition period, and the highest pitch, or by enhancing a timbre that is most clearly distinguishable from ambient sounds. By classifying urgency into multiple levels in this way, there is an advantage in that the acoustic parameters of the alarm sound can be adjusted more systematically and intuitively according to the urgency level during the subsequent alarm sound generation stage.

[0143] In step S44, a robot alarm sound is generated considering the calculated urgency. For example, the alarm sound generation system (10) analyzes the environmental sound for the driving path of an indoor / outdoor driving robot (11), selects an environmental sound source from a plurality of pre-stored candidate environmental sound sources, generates a robot driving alarm sound by adjusting the acoustic parameters of the selected environmental sound source according to the driving speed of the indoor / outdoor driving robot (11), generates an event alarm sound by adjusting the acoustic parameters of a pre-defined event-specific notification sound source corresponding to the determined object event according to the calculated urgency, generates a robot alarm sound draft by synthesizing the generated robot driving alarm sound and the event alarm sound, and generates a final robot alarm sound by adjusting the acoustic parameters of the generated robot alarm sound draft according to the detected noise level.

[0144] In some embodiments, for example, as shown in FIG. 7, the alarm sound generation system (10) can generate a robot alarm sound based on an environmental sound source (71) and an event-specific notification sound source (74).

[0145] First, the selection process of the environmental sound source (71) can be performed as follows. The alarm sound generation system (10) can collect environmental sounds for each section of the driving path of the indoor / outdoor driving robot (11) through sensors such as microphones, and can extract environmental sound characteristics including frequency band, volume, and timbre from the collected environmental sounds.

[0146] Based on the extracted environmental sound characteristics, the alarm sound generation system (10) calculates a suitability score and a distinguishability score for each of the multiple candidate environmental sound sources, and can select an assimilated environmental sound source and a contrasting environmental sound source, respectively, according to the calculated suitability score and the calculated distinguishability score. An assimilated environmental sound source is a candidate environmental sound source that has a suitability score with the extracted environmental sound characteristics equal to or greater than the first score, and can be used to announce the presence of an indoor / outdoor driving robot (11) in a manner that naturally blends with the atmosphere of the space.

[0147] A comparative environmental sound source is a candidate environmental sound source that has a distinction score of at least the second score from the extracted environmental sound characteristics, and can be used to strongly induce auditory attention so that surrounding people and moving objects clearly perceive the approach of the indoor / outdoor driving robot (11). Here, the first score and the second score can be set according to the operating environment and requirements.

[0148] For example, in a quiet space such as a hallway or office, an immersive environmental sound source can be selected as a candidate sound source having a soft tone in the low-frequency (100–500 Hz) band, and a contrasting environmental sound source can be selected as a candidate sound source having a volume that is at least a certain decibel (dB) higher than the ambient sound and a short, regular beat.

[0149] In noisy spaces such as factories or outdoors, assimilation-type environmental sound sources can be selected as candidate sound sources having a timbre similar to the mid-to-high frequency (1kHz~4kHz) band, and contrast-type environmental sound sources can be selected as candidate sound sources in the high frequency (4kHz or higher) series that are clearly distinguishable from the band.

[0150] At this time, a sound source selection unit (34) may be utilized for generating an environment sound source. The process of generating an environment sound source using the sound source selection unit (34) can be performed as follows.

[0151] First, the alarm sound generation system (10) collects ambient sound data for each section through a microphone while the indoor / outdoor driving robot (11) travels along the driving path, and extracts ambient sound characteristics including the main frequency band, volume (dB), and timbre of the corresponding section from the collected ambient sound data. The extracted ambient sound characteristics can be used as selection criteria information to calculate the suitability score for each candidate ambient sound source.

[0152] The sound source selection unit (34) calculates a suitability score and a distinction score for each of the plurality of candidate environment sound sources based on at least one of frequency band similarity, volume difference, timbre similarity, possibility of avoiding masking by ambient noise, suitability for alarm delivery, and estimated user discomfort, and selects at least one of an animated environment sound source and a contrasting environment sound source based on the above.

[0153] Here, frequency band similarity refers to the degree of similarity between the main frequency band of the candidate environmental sound source and the main frequency band of the environmental sound in the current driving section, and volume difference refers to the difference between the volume (dB) of the candidate environmental sound source and the volume (dB) of the environmental sound in the current driving section.

[0154] In addition, timbre similarity refers to the degree of similarity between the timbre characteristics of a candidate environmental sound source and the timbre characteristics of the current driving section's environmental sound, and the possibility of avoiding masking by ambient noise refers to the possibility that a candidate environmental sound source can be transmitted without being audibly obscured by ambient noise.

[0155] Meanwhile, the suitability of the warning transmission refers to the degree to which the candidate environment sound source can effectively convey the presence or dangerous situation of the indoor / outdoor driving robot (11) to surrounding people and moving objects. The user discomfort estimate refers to the degree of auditory discomfort that the candidate environment sound source can cause to surrounding people.

[0156] For example, the sound source selection unit (34) may calculate a suitability score and a distinguishability score through the following process, but is not limited thereto.

[0157] Frequency band similarity is calculated as the overlap ratio between the main frequency bands of the candidate environmental sound source and the environmental sound. For example, if the main frequency band of the environmental sound is 100 to 500 Hz and the main frequency band of the candidate environmental sound source is 200 to 600 Hz, frequency band similarity can be calculated as the ratio of the overlap interval (200 to 500 Hz).

[0158] Timbre similarity can be calculated as the cosine similarity between the MFCC vector of the candidate environment sound source and the MFCC vector of the environment sound.

[0159] The volume difference can be calculated as the absolute difference value between the volume (dB) of the candidate environment sound source and the environment sound.

[0160] The possibility of masking avoidance can be calculated as the ratio of frequency bands exceeding the frequency spectrum of candidate environmental sound sources, based on the masking threshold calculated using an auditory masking model.

[0161] The suitability of alarm delivery can be calculated by comprehensively considering whether the volume of the candidate environment sound source is above a preset SNR threshold relative to ambient noise, whether the repetition period is within a preset range, and whether the frequency band is included in a band with high human hearing sensitivity (e.g., 1kHz to 4kHz).

[0162] The estimated user discomfort can be calculated by comprehensively considering whether the volume of the candidate environment sound source exceeds the maximum allowable volume (e.g., 85dB), whether it includes high frequency bands (e.g., 8kHz or higher) that cause discomfort, and whether the repetition cycle is excessively short and likely to cause auditory fatigue.

[0163] Based on each evaluation metric calculated as described above, the suitability score is calculated as a weighted sum with preset weights applied to frequency band similarity, timbre similarity, and volume difference, and the discriminability score can be calculated as a weighted sum with preset weights applied to masking avoidance possibility and alarm delivery suitability. In this case, candidate environment sound sources in which the estimated user discomfort exceeds a preset discomfort threshold may be excluded from the calculation of the suitability score and discriminability score.

[0164] Next, the animated environmental sound source can be selected from among candidate sound sources with a suitability score of at least a preset first score, as the sound source with the highest suitability score, and the contrastive environmental sound source can be selected from among candidate sound sources with a distinguishability score of at least a preset second score, as the sound source with the highest distinguishability score. Additionally, the volume, repeat period, pitch, and timbre of the candidate sound sources can be selected within a range that satisfies preset safety criteria (e.g., maximum allowable volume, shortest allowable repeat period, etc.) and discomfort criteria (e.g., maximum allowable high-frequency component ratio, etc.).

[0165] That is, the dynamic environmental sound source can be selected from among candidate sound sources with high frequency band similarity and timbre similarity, and the contrasting environmental sound source can be selected from among candidate sound sources that are unlikely to be masked by ambient noise and have high suitability for alarm delivery. In addition, the volume, repetition period, pitch, and timbre of the candidate sound sources can be selected within a range that satisfies pre-set safety and discomfort standards.

[0166] It can be verified whether the selected environmental sound source satisfies preset quality criteria (e.g., similarity to the target frequency band, error from the target volume, identifiability of ambient noise, user discomfort criteria, etc.), and if it does not satisfy the quality criteria, the next-ranked sound source among the candidate environmental sound sources may be selected.

[0167] Environmental sound sources verified to satisfy quality standards can be stored in advance by section to reduce the burden of on-device real-time computation. Specifically, the generated animated and contrasting environmental sound sources can be stored in a storage device of an indoor / outdoor driving robot (11) along with section identification information (e.g., section ID, location coordinates, etc.).

[0168] During actual driving, the environment sound source of the section corresponding to the current position of the indoor / outdoor driving robot (11) is retrieved from the storage device, and the robot can be operated in a way that only acoustic parameters such as volume, pitch, and tempo are adjusted in real time according to the driving speed of the robot, the urgency of the event, or the level of ambient noise detected in real time.

[0169] Additionally, when environmental sound data of a new section is collected or the environmental sound characteristics of an existing section change, the sound source selection unit (34) can recalculate the suitability score and distinguishability score of each candidate environmental sound source and re-select and update the environmental sound source suitable for the section.

[0170] Next, the alarm sound generation system (10) can generate a robot driving alarm sound (73) by performing acoustic parameter adjustment (72-1) based on an environmental sound source (71). At this time, the acoustic parameter adjustment can be performed according to the current driving speed (72-2) of the indoor / outdoor driving robot (11).

[0171] Specifically, the alarm sound generation system (10) can generate a robot driving alarm sound (73) by increasing the volume of an environmental sound source as the driving speed (vr) of the indoor / outdoor driving robot (11) increases. For example, the volume can be adjusted by a function proportional to vr (e.g., A = A_base + k_A × vr, where A_base is the base volume and k_A is the volume adjustment coefficient).

[0172] Additionally, as the driving speed (vr) of the indoor / outdoor driving robot (11) increases, the repetition cycle of the environment sound source can be shortened or the tempo increased to generate a robot driving warning sound (73). For example, the repetition cycle can be adjusted by a function inversely proportional to vr (e.g., T = T_base / (1 + k_T × vr), where T_base is the base repetition cycle and k_T is the repetition cycle adjustment coefficient).

[0173] In addition, a robot driving warning sound (73) can be generated by increasing the pitch of an environmental sound source as the driving speed (vr) of the indoor / outdoor driving robot (11) increases, similar to the Doppler effect. For example, the pitch can be adjusted by a function proportional to vr (e.g., P = P_base + k_P × vr, where P_base is the base pitch and k_P is the pitch adjustment coefficient). The adjustment of the volume, tempo, and pitch can be applied independently or in combination, and each adjustment coefficient (k_A, k_T, k_P) can be set according to the operating environment and requirements.

[0174] Next, the alarm sound generation system (10) can generate an event alarm sound (76) by performing acoustic parameter adjustment (75-1) based on an event-specific notification sound source (74). At this time, the acoustic parameter adjustment can be performed according to the calculated urgency level (75-2).

[0175] Specifically, the alarm sound generation system (10) can generate an event alarm sound (76) by selecting a predefined event-specific notification sound source (74) corresponding to a determined object event (e.g., person detection event, moving object detection event, semi-dynamic object movement possibility event, dynamic object appearance possibility event) and adjusting at least one acoustic parameter among the frequency, volume, repetition period, tempo, pitch, and timbre of the selected notification sound source according to the calculated urgency.

[0176] For example, if the calculated urgency level is very high, the event alarm sound (76) can be generated by increasing the volume of the event alarm sound (76) to the maximum, shortening the repetition cycle to the shortest possible length, raising the pitch to the highest level, or enhancing the tone that is most clearly distinguishable from the surrounding environment sound. Conversely, if the calculated urgency level is low, the event alarm sound (76) can be generated by minimizing the volume of the event alarm sound (76) and setting the repetition cycle to a long length so as not to cause an excessive sense of warning.

[0177] Next, the alarm sound generation system (10) can generate a robot alarm sound draft (77) by synthesizing the generated robot driving alarm sound (73) and the event alarm sound (76). The robot driving alarm sound (73) functions as a basic alarm sound that continuously indicates that the indoor / outdoor driving robot (11) is moving, and the event alarm sound (76) functions as an alarm sound that conveys the presence and urgency of an event when an object event occurs.

[0178] Specifically, the synthesis process can be performed as follows. First, the alarm sound generation system (10) can analyze the frequency spectrum of each sound source by performing a Fast Fourier Transform (FFT) on each of the robot driving alarm sound (73) and the event alarm sound (76). Based on the analyzed frequency spectrum, the alarm sound generation system (10) can pre-adjust the frequency band of each sound source through equalizer processing to prevent the two sound sources from overlapping in the same frequency band, thereby degrading the sound quality or causing a specific frequency band to become oversaturated.

[0179] For example, if the robot driving warning sound (73) and the event warning sound (76) overlap in the same frequency band, the volume of the robot driving warning sound (73) in that frequency band can be reduced, and the volume of the event warning sound (76) in that frequency band can be maintained or increased. Accordingly, the two sound sources can be clearly distinguished and transmitted without interfering with each other.

[0180] After frequency band adjustment is completed, the alarm sound generation system (10) can convert each sound source into the time domain through the Inverse Fast Fourier Transform (IFFT) and perform synthesis by superimposing the waveforms of the two sound sources in the time domain. At this time, during synthesis, the volume ratio of each sound source can be dynamically adjusted according to the calculated urgency.

[0181] Specifically, the volume ratio (r_driving) of the robot driving warning sound (73) and the volume ratio (r_event) of the event warning sound (76) can be set to satisfy r_driving + r_event = 1, and can be adjusted in such a way that r_event increases and r_driving decreases as the calculated urgency (U) increases.

[0182] For example, r_driving : r_event = 0.8 : 0.2 for a low level of urgency, r_driving : r_event = 0.6 : 0.4 for a medium level, r_driving : r_event = 0.4 : 0.6 for a high level, and r_driving : r_event = 0.2 : 0.8 for a very high level. Accordingly, the higher the level of urgency, the more clearly the event alarm sound (76) can be transmitted.

[0183] If, during the synthesis process, the maximum sound pressure of the synthesized waveform exceeds a preset allowable range due to the overlap of the waveforms of the two sound sources (i.e., if oversaturation occurs), the alarm sound generation system (10) can further perform Dynamic Range Compression (DRC) processing to adjust the sound pressure of the synthesized waveform to within the allowable range. Accordingly, the two sound sources can be delivered in a balanced manner without degrading the sound quality of the synthesized robot alarm draft (77).

[0184] Additionally, fade-in and fade-out processing may be further performed on the synthesized robot alarm draft (77). For example, when a new object event occurs, fade-in processing may be performed to gradually increase the volume of the event alarm (76) over a certain period of time instead of rapidly increasing it, and when the object event is resolved, fade-out processing may be performed to gradually decrease the volume of the event alarm (76) over a certain period of time. Accordingly, it is possible to prevent causing unnecessary surprise or discomfort to people around due to sudden changes in the alarm sound.

[0185] Finally, the alarm sound generation system (10) can generate a final robot alarm sound (79) by performing acoustic parameter adjustment (78-1) on the generated robot alarm sound draft (77) according to the ambient noise level (78-2) detected in real time.

[0186] Specifically, the ambient noise level (N) can be calculated as a sound pressure level (SPL, unit: dB) from noise data acquired in real time through a microphone. If the calculated ambient noise level (N) exceeds a preset threshold noise level (N_th), the alarm sound generation system (10) can adjust acoustic parameters by increasing the volume of the final robot alarm sound (79) or by emphasizing a frequency band that is distinct from the ambient noise.

[0187] Conversely, when the ambient noise level (N) is below the threshold noise level (N_th), the alarm sound generation system (10) can adjust acoustic parameters by lowering the volume of the final robot alarm sound (79) or increasing the proportion of the animated environmental sound source. Accordingly, the final robot alarm sound (79) can be effectively transmitted even in noisy environments such as factories, outdoor roads, and logistics centers, and unnecessary noise or discomfort can be minimized in quiet spaces such as offices, hospitals, and corridors.

[0188] In some other embodiments, the acoustic parameters of the final robot alarm sound (79) may be further adjusted according to the driving direction of the indoor / outdoor driving robot (11). Specifically, the indoor / outdoor driving robot (11) may be equipped with a plurality of speakers facing in different directions, and the alarm sound generation system (10) may obtain the current driving direction and rotation direction of the indoor / outdoor driving robot (11) from a sensor such as an IMU or encoder, and independently adjust the volume of each speaker based on the obtained driving direction information.

[0189] For example, when the indoor / outdoor driving robot (11) turns to the left, the acoustic parameters can be adjusted by increasing the volume of the left speaker and decreasing the volume of the right speaker. Accordingly, people and moving objects in the vicinity can more intuitively perceive the direction of movement of the indoor / outdoor driving robot (11). Additionally, when the indoor / outdoor driving robot (11) moves straight, the volume can be adjusted by increasing the volume of the front speaker and decreasing the volume of the rear speaker, and at this time, the volume ratio of each speaker can be further adjusted in proportion to the driving speed of the indoor / outdoor driving robot (11).

[0190] In some other embodiments, the acoustic parameters of the final robot alarm sound (79) may be adjusted by increasing the volume of the speaker in the direction in which the dynamic object or entry / exit structure object is detected. Specifically, the alarm sound generation system (10) may calculate the direction in which the object is located relative to the indoor / outdoor driving robot (11) from the location information of each object calculated in step S42, and increase the volume of the speaker corresponding to the calculated direction.

[0191] For example, if a person detection event occurs and the person is located to the right front of the indoor / outdoor driving robot (11), the volume of the right front speaker can be increased so that the person can more clearly perceive the approach of the indoor / outdoor driving robot (11). Additionally, if an event indicating the possibility of a dynamic object appearing near an entrance / exit structure object occurs, the volume of the speaker in the direction of the entrance / exit structure object can be increased to notify the dynamic object that may appear from a non-visible area of ​​the presence of the indoor / outdoor driving robot (11) in advance. At this time, the volume adjustment amount of each speaker can be determined in proportion to the distance between the object and the indoor / outdoor driving robot (11) and the calculated urgency.

[0192] FIG. 8 is an exemplary flowchart illustrating a method for generating a robot alarm sound with a collision prediction time taken into account according to some embodiments of the present disclosure.

[0193] Referring to FIG. 8, the alarm sound generation system (10) may generate an event alarm sound by taking into account the time to collision. Here, the time to collision (TTC) refers to the time expected until a collision occurs between the indoor / outdoor driving robot (94) and a dynamic object based on the current time. The time to collision can be calculated by considering not only the possibility of a collision with the currently detected dynamic object, but also all possible collision scenarios that may occur within a predetermined time (e.g., 5 seconds). Accordingly, the alarm sound generation system (10) can generate a more accurate and situation-appropriate event alarm sound by correcting the urgency calculated in step S43 based on the time to collision.

[0194] In step S81, the predicted collision time is calculated based on the driving trajectory of the indoor / outdoor driving robot (94) and the predicted movement trajectory of the dynamic object. Specifically, the alarm sound generation system (10) first calculates a driving trajectory within a predetermined time based on the current driving speed, driving direction, and pre-set driving path information of the indoor / outdoor driving robot (94).

[0195] Next, the alarm sound generation system (10) calculates all predicted movement trajectories that a dynamic object can move within a predetermined time. The predicted movement trajectories of the dynamic object can be calculated as time-position functions and time-velocity functions according to multiple movement scenarios based on a preset motion model (e.g., Constant Velocity Model (CV), Constant Acceleration Model (CA), Constant Turn Rate and Velocity Model (CTRV), etc.), using the currently detected position, movement speed, and movement direction of the dynamic object as initial state values. For example, the movement scenarios can be configured as follows.

[0196] The first scenario is when a dynamic object approaches an indoor / outdoor driving robot (94) while maintaining its current speed and direction of movement. This scenario can be calculated based on a constant velocity motion model (CV) and can be represented as a time-velocity graph (e.g., a horizontal line) that maintains the current speed without a change in speed for a set period of time.

[0197] The second scenario is when a dynamic object suddenly comes to a stop. This scenario can be calculated based on an equal acceleration motion model (CA) with maximum deceleration applied, and can be represented by a time-velocity graph (e.g., a horizontal line after deceleration) in which the velocity decreases from an initial velocity to the maximum deceleration until it reaches zero and then remains at zero.

[0198] The third scenario is a case where the dynamic object moves in the same direction as the driving direction of the indoor / outdoor driving robot (94), but the speed gradually decreases. In this scenario, the deceleration can be set to multiple values ​​within a preset range to calculate a time-speed function according to each deceleration scenario, and can be represented by multiple time-speed graphs (e.g., multiple straight lines with different slopes for each deceleration value).

[0199] The fourth scenario is when a dynamic object suddenly appears from an invisible area within a predetermined time around the entrance structure object (93). This scenario can be calculated based on the location of the entrance structure object (93), the direction of the opening, and the possibility of the dynamic object appearing.

[0200] Specifically, the alarm sound generation system (10) can set the initial position and the time of appearance of a dynamic object that may appear from the opening of the entrance structure object (93) in a plurality of combinations, and for each combination, calculate the movement trajectory after appearance based on a constant velocity motion model (CV) or a constant acceleration motion model (CA). Accordingly, a plurality of time-velocity graphs (e.g., a step shape starting with a velocity of 0 before the time of appearance and a preset initial velocity after the time of appearance) that differ depending on the combination of the time of appearance and the direction of movement after appearance can be calculated.

[0201] In addition, time-velocity and time-position functions can be further calculated according to various movement scenarios, such as when a dynamic object changes direction to the left or right (e.g., based on a constant velocity turning motion model (CTRV)) or when the speed of a dynamic object suddenly increases (e.g., based on a constant acceleration motion model (CA) with positive acceleration applied).

[0202] As illustrated in FIG. 9, the alarm sound generation system (10) can detect a driving environment including a person (91), a static object (92), and an entry / exit structure object (93) through a sensor (95) equipped in an indoor / outdoor driving robot (94), and can calculate the driving trajectory of the indoor / outdoor driving robot (94) and the predicted movement trajectory of the dynamic object based on the location and state information of each detected object. If there is a point where the calculated driving trajectory of the indoor / outdoor driving robot (94) and the predicted movement trajectory of the dynamic object for each movement scenario intersect, the alarm sound generation system (10) can calculate the difference between the time required for the indoor / outdoor driving robot (94) to reach the intersection point and the time required for the dynamic object to reach the intersection point as the predicted collision time for each movement scenario.

[0203] In step S82, a correction value for the calculated urgency is calculated by taking into account the calculated estimated collision time. Specifically, the alarm sound generation system (10) can calculate the correction value based on a plurality of estimated collision times calculated for each movement scenario and the probability of occurrence of each movement scenario.

[0204] First, the probability of occurrence for each movement scenario is calculated. The probability of occurrence can be calculated based on the dynamic object's current movement speed, direction of movement, robot recognition rate, and movement patterns learned from past driving data. For example, if the dynamic object's robot recognition rate is low and it is moving in a direction approaching the indoor / outdoor driving robot (94), the probability of occurrence for the first scenario (e.g., constant speed approach) can be calculated as high.

[0205] Conversely, if a large rate of change in the movement speed of a dynamic object is detected, the probability of occurrence of the second scenario (e.g., sudden stop) can be calculated as high. Additionally, the probability of occurrence of the fourth scenario (e.g., appearance in an invisible area) can be calculated based on the history of past appearances of the dynamic object around the entrance / exit structure object (93) and current time zone information. The probability of occurrence of each calculated movement scenario can be normalized so that it becomes 1 when summed for all scenarios.

[0206] Next, a weighted average collision time is calculated by applying normalized occurrence probabilities as weights. The weighted average collision time can be calculated by summing the values ​​obtained by multiplying the collision time for each movement scenario by the occurrence probability of that scenario. In this case, for a scenario where no collision trajectory exists (i.e., a scenario where no collision occurs within a predetermined time), the collision time for that scenario can be set to a predetermined time.

[0207] Next, a correction value is calculated based on the calculated weighted average collision time. The correction value can be calculated inversely proportional to the weighted average collision time; the shorter the weighted average collision time, the larger the correction value, and the longer the time, the smaller the correction value. Additionally, if the minimum value among multiple collision times is less than a preset threshold collision time, it is determined that a collision is imminent, and the correction value can be adjusted upward by applying additional weight. For example, if the minimum value among the collision times is less than 50% of the preset threshold collision time, a method of doubling the correction value can be applied. Accordingly, an event alarm sound can be generated more strongly in situations where the risk of collision is imminent.

[0208] In step S83, a final event alarm is generated by applying a calculated correction value to an event alarm with adjusted acoustic parameters. Specifically, the alarm generation system (10) can generate a final event alarm by applying a calculated correction value to an event alarm with adjusted acoustic parameters according to the urgency in step S44, thereby further adjusting the volume, repetition period, and pitch.

[0209] Specifically, the volume can be adjusted by adding an additional amplification amount proportional to the correction value to the volume of the event alarm sound adjusted according to the urgency. The repetition cycle can be adjusted by shortening the repetition cycle of the event alarm sound adjusted according to the urgency in proportion to the correction value.

[0210] In the case of pitch, it can be adjusted by adding an additional increase proportional to the correction value to the pitch of the event alarm sound adjusted according to the urgency. At this time, the adjustment amount of each acoustic parameter can be clipped so as not to exceed a preset allowable range (e.g., maximum volume, shortest repetition period, maximum pitch).

[0211] For example, if a correction value is calculated to be large and the adjusted volume exceeds the maximum allowable volume, the volume may be limited to the maximum allowable volume. In this way, by applying the calculated correction value to the acoustic parameters of the event alarm sound, the final event alarm sound can be generated such that the volume increases, the repetition cycle is shortened, and the pitch is raised as the expected collision time becomes shorter (i.e., the larger the correction value).

[0212] Up to now, with reference to FIGS. 4 to 9, a method for generating an indoor and outdoor driving robot alarm sound according to some embodiments of the present disclosure has been described. Below, with reference to FIG. 10, an exemplary computing device (100) capable of implementing the alarm sound generation system (10) described above will be described.

[0213] FIG. 10 is an exemplary hardware configuration diagram showing a computing device (100).

[0214] As illustrated in FIG. 10, a computing device (100) may include one or more processors (101), a bus (103), a communication interface (104), a memory (102) for loading a computer program (106) executed by the processor (101), and a storage (105) for storing the computer program (106). However, FIG. 10 illustrates only the components related to the embodiments of the present disclosure. Therefore, a person skilled in the art to which the present disclosure belongs will understand that other general-purpose components may be included in addition to the components (101 to 106) illustrated in FIG. 10. That is, the computing device (100) may include various additional components in addition to the components (101 to 106) illustrated in FIG. 10. Furthermore, depending on the case, the computing device (100) may be configured in a form in which some of the components (101 to 106) illustrated in FIG. 10 are omitted. Below, each component of the computing device (100) is described.

[0215] The processor (101) can control the overall operation of each component of the computing device (100). The processor (101) may be configured to include at least one of a CPU (Central Processing Unit), MPU (Micro Processor Unit), MCU (Micro Controller Unit), GPU (Graphic Processing Unit), or any form of processor well known in the art of the present disclosure. Additionally, the processor (101) may perform operations on at least one application or program to execute specific steps / operations / methods. The computing device (100) may have one or more processors.

[0216] Next, the memory (102) may store various data, commands and / or information. The memory (102) may load a computer program (106) from storage (105) to execute specific steps / operations / methods. The memory (102) may be implemented as volatile memory such as RAM, but the technical scope of the present disclosure is not limited thereto.

[0217] Next, the bus (103) can provide communication functions between components of the computing device (100). The bus (103) can be implemented as various types of buses, such as an address bus, a data bus, and a control bus.

[0218] Next, the communication interface (104) can support wired and wireless internet communication of the computing device (100). Additionally, the communication interface (104) may support various communication methods other than internet communication. To this end, the communication interface (104) may be configured to include a communication module well known in the art of the present disclosure.

[0219] Next, the storage (105) may store one or more computer programs (106) non-temporarily. The storage (105) may be configured to include non-volatile memory such as ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable ROM), flash memory, a hard disk, a removable disk, or any form of computer-readable recording medium well known in the art to which this disclosure belongs.

[0220] Next, the computer program (106) may include instructions that cause the processor (101) to perform specific steps / actions / methods when loaded into memory (102). That is, the processor (101) can perform specific steps / actions / methods by executing the instructions loaded into memory (102).

[0221] For example, a computer program (106) may include instructions for the operation of acquiring driving environment data regarding the driving environment of an indoor / outdoor driving robot, the operation of calculating object information including object type information and object state information for each of a plurality of objects identified from the acquired driving environment data, the operation of determining an object event based on the calculated object information and calculating the urgency of the determined object event, and the operation of generating a robot alarm sound considering the calculated urgency.

[0222] As another example, a computer program (106) may include instructions to perform at least some of the steps / actions / methods described with reference to FIGS. 1 through 9.

[0223] As illustrated, an alarm sound generating system (10) according to some embodiments of the present disclosure can be implemented through a computing device (100).

[0224] Meanwhile, in some embodiments, the computing device (100) illustrated in FIG. 10 may refer to a virtual machine implemented based on cloud technology. For example, the computing device (100) may be a virtual machine running on one or more physical servers included in a server farm. In this case, at least some of the processor (101), memory (102), and storage (105) illustrated in FIG. 10 may be virtual hardware, and the communication interface (104) may also be implemented as a virtualized networking element such as a virtual switch.

[0225] Up to now, with reference to FIG. 10, an exemplary computing device (100) capable of implementing an alarm sound generating system (10) according to some embodiments of the present disclosure has been described.

[0227] Various embodiments of the present disclosure and effects according to the embodiments have been described so far with reference to FIGS. 1 to 10.

[0228] The effects according to the technical concept of the present disclosure are not limited to those mentioned above, and other unmentioned effects will be clearly understood by a person skilled in the art from the description below.

[0229] Furthermore, just because the above embodiments describe a plurality of components being combined into one or operating in combination, the technical concept of the present disclosure is not necessarily limited to these embodiments. That is, within the scope of the purpose of the technical concept of the present disclosure, all such components may be selectively combined into one or more combinations to operate.

[0230] The technical concept of the present disclosure described above may be implemented as computer-readable code on a computer-readable recording medium. A computer program stored on a computer-readable recording medium may be transmitted to another computing device via a network such as the Internet and installed on said computing device, thereby being used on said computing device.

[0231] Although operations are depicted in a specific order in the drawings, it should not be understood that the operations must necessarily be executed in the specific order depicted or in a sequential order, or that all depicted operations must be executed to obtain the desired result. In certain situations, multitasking and parallel processing may be advantageous. Although various embodiments of the present disclosure have been described above with reference to the attached drawings, those skilled in the art will understand that the technical concept of the present disclosure may be implemented in other specific forms without altering the technical concept or essential features thereof. Therefore, the embodiments described above should be understood as illustrative in all respects and not restrictive. The scope of protection of the present disclosure shall be interpreted by the claims below, and all technical concepts within the equivalent scope shall be interpreted as being included within the scope of rights of the technical concept defined by the present disclosure.

Claims

Claim 1 A method for generating an alarm sound for an indoor / outdoor driving robot, comprising: a step of acquiring driving environment data regarding a driving environment of an indoor / outdoor driving robot, performed by at least one processor; a step of calculating object information including object type information and object state information for each of a plurality of objects identified from the acquired driving environment data; a step of determining an object event based on the calculated object information and calculating the urgency of the determined object event; and a step of generating a robot alarm sound considering the calculated urgency, wherein the step of calculating object information comprises: a step of identifying each of a plurality of objects from the acquired driving environment data; and a step of calculating object type information among a dynamic object, a semi-dynamic object, an access structure object, and a static object for each of the identified plurality of objects. The method includes a step of calculating object state information for each of the plurality of objects according to the object type information calculated above, wherein the step of calculating any one of the object type information comprises: determining whether the identified object is a person, and if the identified object is determined to be a person, calculating the object type information of the identified object as a person among dynamic objects; if the identified object is determined not to be a person, calculating the speed of the identified object, and if the calculated speed is greater than 0, calculating the identified object type information as a moving body among dynamic objects; if the calculated speed is 0, determining whether the identified object is a movable object, and if the identified object is determined to be a movable object, calculating the object type information of the identified object as a quasi-dynamic object; if the identified object is determined not to be a movable object, determining whether the identified object is a structure through which the dynamic object enters and exits, and if the identified object is determined to be a structure through which the dynamic object enters and exits, calculating the object type information of the identified object as an entry / exit structure object.A method for generating an indoor / outdoor driving robot alarm sound, comprising the step of calculating object type information of the identified object as a static object when the identified object is determined not to be a structure through which the dynamic object enters or exits. Claim 2 A method for generating an alarm sound for an indoor / outdoor driving robot according to claim 1, wherein the step of acquiring driving environment data comprises the step of acquiring driving environment data including at least one of image data and distance data using at least one sensor among a camera, a depth camera, and a lidar equipped in the indoor / outdoor driving robot. Claim 3 delete Claim 4 delete Claim 5 A method for generating an alarm sound for an indoor / outdoor driving robot, comprising: a step of acquiring driving environment data regarding a driving environment of an indoor / outdoor driving robot, performed by at least one processor; a step of calculating object information including object type information and object state information for each of a plurality of objects identified from the acquired driving environment data; a step of determining an object event based on the calculated object information and calculating the urgency of the determined object event; and a step of generating a robot alarm sound considering the calculated urgency, wherein the step of calculating object information comprises: a step of identifying each of a plurality of objects from the acquired driving environment data; and a step of calculating object type information among a dynamic object, a semi-dynamic object, an access structure object, and a static object for each of the identified plurality of objects. The method comprises a step of calculating object state information for each of the plurality of objects according to the object type information calculated above, wherein the step of calculating object state information includes: a step of calculating object state information including the distance between the person and the indoor / outdoor driving robot, the movement speed of the person, the movement direction of the person, and the robot recognition rate of the person when the calculated object type information is a person among dynamic objects; a step of calculating object state information including the distance between the moving object and the indoor / outdoor driving robot, the movement speed of the moving object, the movement direction of the moving object, and the robot recognition rate of the moving object when the calculated object type information is a moving object among dynamic objects; a step of calculating object state information including the distance between the semi-dynamic object and the indoor / outdoor driving robot and the possibility of movement of the semi-dynamic object when the calculated object type information is a semi-dynamic object; and a step of calculating object state information including the distance between the access structure object and the indoor / outdoor driving robot and the possibility of appearance of dynamic objects around the access structure object when the calculated object type information is an access structure object.A method for generating an indoor / outdoor driving robot alarm sound, comprising the step of calculating the distance between the static object and the indoor / outdoor driving robot as object state information when the object type information calculated above is a static object. Claim 6 A method for generating an alarm sound for an indoor / outdoor driving robot, comprising: a step of acquiring driving environment data regarding a driving environment of an indoor / outdoor driving robot, performed by at least one processor; a step of calculating object information including object type information and object state information for each of a plurality of objects identified from the acquired driving environment data; a step of determining an object event based on the calculated object information and calculating the urgency of the determined object event; and a step of generating a robot alarm sound considering the calculated urgency. In this method, the step of calculating object information includes: a step of identifying a plurality of objects from the acquired driving environment data; a step of calculating object type information among a dynamic object, a semi-dynamic object, an access structure object, and a static object for each of the identified plurality of objects; and a step of calculating object state information for each of the plurality of objects according to the calculated object type information. The step of determining an object event and calculating the urgency of the determined object event includes, when the calculated object type information is a dynamic object, determining the object event as a dynamic object detection event and calculating the urgency of the dynamic object detection event based on the object state information for the dynamic object. A method for generating an alarm sound for an indoor / outdoor driving robot, comprising the step of determining the object event as a dynamic object prediction event when the calculated object type information is a semi-dynamic object or an access structure object, and calculating the urgency of the dynamic object prediction event based on object state information for the semi-dynamic object or the access structure object. Claim 7 In claim 6, the step of calculating the urgency of the dynamic object detection event comprises: determining the dynamic object detection event as a person detection event when the dynamic object is a person, and calculating the urgency of the person detection event based on the distance between the person and the indoor / outdoor driving robot, the driving speed of the indoor / outdoor driving robot, the movement speed of the person, the direction of movement of the person, and the robot recognition rate of the person; and determining the dynamic object detection event as a moving object detection event when the dynamic object is a moving object, and calculating the urgency of the moving object detection event based on the distance between the moving object and the indoor / outdoor driving robot, the driving speed of the indoor / outdoor driving robot, the movement speed of the moving object, the direction of movement of the moving object, and the robot recognition rate of the moving object, a method for generating an indoor / outdoor driving robot alarm sound. Claim 8 In claim 6, the step of calculating the urgency of the dynamic object prediction event comprises: determining the dynamic object prediction event as a quasi-dynamic object movement possibility event when the object type information is a quasi-dynamic object, and calculating the urgency of the quasi-dynamic object movement possibility event based on the distance between the quasi-dynamic object and the indoor / outdoor driving robot, the driving speed of the indoor / outdoor driving robot, and the movement possibility of the quasi-dynamic object; and determining the dynamic object prediction event as a dynamic object appearance possibility event when the object type information is an access structure object, and calculating the urgency of the dynamic object appearance possibility event based on the distance between the access structure object and the indoor / outdoor driving robot, the driving speed of the indoor / outdoor driving robot, and the possibility of a dynamic object appearing around the access structure object. Claim 9 A method for generating an alarm sound for an indoor / outdoor driving robot, comprising: a step of acquiring driving environment data regarding the driving environment of an indoor / outdoor driving robot, performed by at least one processor; a step of calculating object information including object type information and object state information for each of a plurality of objects identified from the acquired driving environment data; a step of determining an object event based on the calculated object information and calculating the urgency of the determined object event; and a step of generating a robot alarm sound considering the calculated urgency. In this method, the step of generating the robot alarm sound comprises: a step of analyzing the environmental sound for the driving path of the indoor / outdoor driving robot and selecting an environmental sound source from a plurality of pre-stored candidate environmental sound sources; a step of generating a robot driving alarm sound by adjusting the acoustic parameters of the selected environmental sound source according to the driving speed of the indoor / outdoor driving robot; and a step of generating an event alarm sound by adjusting the acoustic parameters of a pre-defined event-specific notification sound source corresponding to the determined object event according to the calculated urgency. A method for generating an indoor / outdoor driving robot alarm sound, comprising the step of synthesizing the generated robot driving alarm sound and the generated event alarm sound to generate a draft robot alarm sound, and adjusting the acoustic parameters of the generated draft robot alarm sound according to the detected noise level to generate a final robot alarm sound. Claim 10 In claim 9, the step of selecting the environmental sound source comprises: a step of extracting environmental sound characteristics including at least one of a frequency band, volume, and timbre from the environmental sound for the driving path; a step of calculating a suitability score and a distinction score for each of a plurality of previously stored candidate environmental sound sources based on at least one of frequency band similarity, volume difference, timbre similarity, possibility of avoiding masking by ambient noise, suitability for warning delivery, and estimated user discomfort with respect to the extracted environmental sound characteristics; and a step of selecting at least one of an assimilative environmental sound source and a contrastive environmental sound source based on the calculated suitability score, wherein the assimilative environmental sound source is a candidate environmental sound source having a suitability score with respect to the extracted environmental sound characteristics of at least a first score, and the contrastive environmental sound source is a candidate environmental sound source having a distinction score with respect to the extracted environmental sound characteristics of at least a second score, a method for generating an alarm sound for an indoor / outdoor driving robot. Claim 11 In claim 9, the step of generating the robot driving warning sound comprises at least one of the following steps: generating the robot driving warning sound by increasing the volume of the selected environmental sound source as the driving speed of the indoor / outdoor driving robot increases; generating the robot driving warning sound by shortening the repetition cycle of the selected environmental sound source or increasing the tempo of the selected environmental sound source as the driving speed of the indoor / outdoor driving robot increases; and generating the robot driving warning sound by increasing the pitch of the selected environmental sound source as the driving speed of the indoor / outdoor driving robot increases. Claim 12 In claim 9, the step of generating the event alarm sound comprises: selecting a predefined event-specific notification sound source corresponding to the determined object event; and generating an event alarm sound in which at least one acoustic parameter among the frequency, volume, repetition period, tempo, pitch, and timbre of the selected event-specific notification sound source is adjusted according to the calculated urgency, wherein the step of generating the event alarm sound with adjusted acoustic parameters comprises increasing the volume of the selected event-specific notification sound source, shortening the repetition period, increasing the pitch, or enhancing the timbre distinct from the ambient sound as the calculated urgency increases, thereby generating the event alarm sound with adjusted acoustic parameters. Claim 13 In claim 12, the step of generating the event alarm sound further comprises: a step of calculating an estimated collision time based on the driving trajectory of the indoor / outdoor driving robot and the predicted movement trajectory of a dynamic object; a step of calculating a correction value for the calculated urgency considering the calculated estimated collision time; and a step of generating a final event alarm sound by applying the calculated correction value to the event alarm sound with the acoustic parameters adjusted. Claim 14 A system for generating an alarm sound for an indoor / outdoor driving robot, comprising: one or more processors; and a memory for storing a computer program executed by the one or more processors, wherein the computer program comprises: instructions for acquiring driving environment data regarding the driving environment of an indoor / outdoor driving robot; an operation for calculating object information including object type information and object state information for each of a plurality of objects identified from the acquired driving environment data; an operation for determining an object event based on the calculated object information and calculating the urgency of the determined object event; and instructions for generating a robot alarm sound considering the calculated urgency. In this system, the operation for calculating object information comprises: an operation for identifying a plurality of objects from the acquired driving environment data; and an operation for calculating object type information among a dynamic object, a semi-dynamic object, an access structure object, and a static object for each of the identified plurality of objects. The method includes an operation of calculating object state information for each of the plurality of objects according to the object type information calculated above, wherein the operation of calculating any one of the object type information comprises: determining whether the identified object is a person, and if the identified object is determined to be a person, calculating the object type information of the identified object as a person among dynamic objects; for the case where the identified object is determined not to be a person, calculating the speed of the identified object, and if the calculated speed is greater than 0, calculating the identified object type information as a moving body among dynamic objects; for the case where the calculated speed is 0, determining whether the identified object is a movable object, and if the identified object is determined to be a movable object, calculating the object type information of the identified object as a quasi-dynamic object.An indoor / outdoor driving robot alarm sound generation system comprising: determining whether the identified object is a structure through which the dynamic object enters and exits when the identified object is determined not to be a movable object, and, if the identified object is determined to be a structure through which the dynamic object enters and exits, calculating the object type information of the identified object as an entry / exit structure object; and, if the identified object is determined not to be a structure through which the dynamic object enters and exits, calculating the object type information of the identified object as a static object. Claim 15 A computer program stored on a computer-readable recording medium for executing the steps of: acquiring driving environment data regarding the driving environment of an indoor / outdoor driving robot combined with a computer processor; calculating object information including object type information and object state information for each of a plurality of objects identified from the acquired driving environment data; determining an object event based on the calculated object information and calculating the urgency of the determined object event; and generating a robot alarm sound considering the calculated urgency, wherein the step of calculating object information comprises: identifying a plurality of objects from the acquired driving environment data; and calculating object type information among a dynamic object, a semi-dynamic object, an access structure object, and a static object for each of the identified plurality of objects. The method includes a step of calculating object state information for each of the plurality of objects according to the object type information calculated above, wherein the step of calculating any one of the object type information comprises: determining whether the identified object is a person, and if the identified object is determined to be a person, calculating the object type information of the identified object as a person among dynamic objects; if the identified object is determined not to be a person, calculating the speed of the identified object, and if the calculated speed is greater than 0, calculating the identified object type information as a moving body among dynamic objects; if the calculated speed is 0, determining whether the identified object is a movable object, and if the identified object is determined to be a movable object, calculating the object type information of the identified object as a quasi-dynamic object; if the identified object is determined not to be a movable object, determining whether the identified object is a structure through which the dynamic object enters and exits, and if the identified object is determined to be a structure through which the dynamic object enters and exits, calculating the object type information of the identified object as an entry / exit structure object.A computer program comprising the step of calculating object type information of the identified object as a static object when it is determined that the identified object is not a structure through which the dynamic object enters or exits.

Images