Automated and Adaptive 3D Robotic Site Inspection

An autonomous unmanned robot system optimally captures local geometry for 3D reconstruction, addressing the inefficiencies of manual data collection by dynamically planning and adapting its mission, thus reducing time and computational costs while ensuring high-fidelity 3D model generation.

JP7580875B2Active Publication Date: 2024-11-12GENERAL ELECTRIC CO
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
JP2023101091
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Priority Date
2018-01-25
Filing Date
2023-06-20
Publication Date
2024-11-12
Estimated Expiration
2038-12-27

AI Technical Summary

Technical Problem

Existing 3D reconstruction systems for industrial assets rely heavily on manual data collection by human operators, which is time-consuming, error-prone, and computationally expensive, and often result in inadequate data quality, especially in complex and dynamic environments.

Method used

An autonomous unmanned robot system that plans its mission dynamically, adapts in real-time, and optimally captures local geometry for 3D reconstruction using multisensory data, minimizing the number of images required while ensuring accuracy.

Benefits of technology

Reduces the time and computational resources needed for 3D model generation, enhances data quality, and allows for efficient, adaptive data collection in complex environments, enabling rapid creation of high-fidelity 3D models.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007580875000001
    Figure 0007580875000001
  • Figure 0007580875000002
    Figure 0007580875000002
  • Figure 0007580875000003
    Figure 0007580875000003
Patent Text Reader

Abstract

To provide a method for generating a three-dimensional model for an asset.SOLUTION: A method for generating a three-dimensional model of an asset comprises steps of: receiving input parameters corresponding to constraints of a mission plan for operating an unmanned vehicle around an asset; generating the mission plan based on the input parameters including information of a representative asset type, in which the mission plan includes waypoints identifying locations and orientations of image sensors of the unmanned vehicle; generating a flight path for the unmanned vehicle connecting the waypoints that satisfy predefined criteria; monitoring a vehicle state of the unmanned vehicle during execution of the flight path from one waypoint to the next waypoint; determining, at each waypoint, a local geometry of the asset sensed by the image sensors; changing the mission plan on-the-fly based on the local geometry; and capturing images of the asset along the waypoints of the changed mission plan.SELECTED DRAWING: Figure 3
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The subject matter disclosed herein relates to site inspection, and more particularly to the autonomous collection of multi-sensory data around an industrial asset to build a three-dimensional model of the asset and its environment. It would be desirable for a robot, particularly one operating in an industrial environment, to have the ability to autonomously sense and model the environment. This model of the environment can be used to direct other tasks such as inspection, obstacle avoidance, robot localization, targeting, etc. [Background technology]

[0002] Many of the existing 3D reconstruction systems suffer from several limitations that affect their performance. Typically, the data required for three-dimensional (3D) reconstruction of an asset is collected by drones manually piloted by a human operator or by human-carried sensing systems. It is noted that having humans perform aspects of the data collection process can be a difficult and error-prone task. This is especially true when planned data collection takes a significant amount of time, when collection may take various routes, when there are many points of interest that need to be surveyed, when the asset and / or surrounding environment are complex and dynamically changing, when other people and / or robots are operating in the area simultaneously, etc.

[0003] Manual data collection does not guarantee that adequate data is collected to generate a 3D reconstruction that meets certain key performance indicators (KPIs), such as ensuring that images are in focus, cover the right areas of the asset, and have sufficient overlap between images to build an accurate model. Furthermore, manual data collection typically results in the collection of large amounts of data. As mentioned above, currently available systems typically involve a human operator manually piloting a drone to capture a large number of images (e.g., on the order of thousands), which are then used to build a 3D model of the asset. Sending this entire dataset to a reconstruction engine can be computationally prohibitive. Additionally, the presence of images that are out of focus or do not have enough overlap can result in poor quality reconstructions.

[0004] It would therefore be desirable to provide a system and method that facilitates three-dimensional robotic site inspection in an automatic and adaptive manner. Summary of the Invention [Problem to be solved by the invention]

[0005] The embodiments described herein may be implemented in a manner that allows an unmanned robot / vehicle (e.g., given an asset type) to The present invention improves upon the prior art by providing a system and method that autonomously plans (based on a single satellite) its mission and updates / adapts its mission in real time ("on the fly") based on new information.

[0006] In various embodiments, an unmanned aerial system can autonomously collect multi-sensory data about an asset to build a 3D model with minimal human intervention. The system generates a mission plan that considers various constraints of the task and asset (e.g., geofences, camera parameters, asset dimensions, asset type, etc.). The various constraints are combined to build a mission plan that collects the right amount of data to maximize the accuracy of the 3D reconstruction. Additionally, the system can automatically re-plan on the fly based on the local geometry of the asset as sensed by two-dimensional (2D) and three-dimensional (3D) cameras.

[0007] In some traditional approaches, the data required for 3D reconstruction of an asset is typically collected by a drone manually piloted by a human operator or by a human-ported sensing system. For example, the drone first captures data / initial images manually (e.g., by a human operator) or by flying a defined path, which is then used to build a 3D model of the asset offline. Offline analytics, such as machine learning, can be used to clean up the data before feeding it to the 3D reconstruction engine. For example, a large number of images collected by a human operator may be pruned down to an optimal number (e.g., a minimal, relevant set) of data samples required for accurate 3D reconstruction using sensory and domain models. However, this requires constant feedback from the operator to ensure both that the images are of the desired quality and that the desired images have been captured.

[0008] One limitation of traditional approaches is that building a 3D model of an asset is a computationally expensive process that depends heavily on the size and type of the asset and the number of images collected. Depending on the size of the asset and the power limitations of the drone, the model building operation can take a long time to complete, anywhere from hours to days (e.g., for a typical large industrial asset). Furthermore, these 3D reconstruction engines are typically better run on cloud servers where they can be parallelized. However, many industrial assets are located in areas where it is difficult to upload large amounts of information over a network. As a result, traditional systems typically take days to build an accurate model of an asset that can be used to regenerate a mission plan. [Means for solving the problem]

[0009] To overcome these issues, examples herein utilize online sensing to update the mission plan on the fly in a way that optimally captures the local shape for 3D reconstruction. Advantageously, access to a previously constructed 3D model is not required. This differs from traditional solutions where planning is done offline and the drone flies the mission in an open-loop fashion.

[0010] According to various embodiments, the system automatically calculates the appropriate position and camera / viewing angle to acquire images that achieve efficient image acquisition (e.g., with respect to image overlap) and identify the correct area of ​​interest for the asset. Furthermore, data from the image sensor can be fed back to the system to assess image quality and take appropriate action if necessary. Thus, the drone can fly a mission autonomously while its position and the camera positions are tracked. Images are captured only at planned locations and appropriate metadata such as drone position and orientation are stored. In this way, a minimal number of images are collected while still meeting the required reconstruction accuracy. Furthermore, the system allows the robot to adapt its plan on the fly and reposition parts of the asset based on local shape sensing. Optimal capture.

[0011] Technical advantages offered by 3D reconstruction systems include the ability to automatically determine the best locations to image an asset, leading to an accurate reconstruction of the asset with a minimum number of images, and the ability for the drone to track its mission and take corrective actions based on sensed information such as local geometry.

[0012] Commercial benefits brought by 3D reconstruction systems include the ability for operators (e.g., engineers) to quickly collect the right amount and fidelity of data (e.g., the minimum number of images required) to create a high-fidelity 3D model of an asset. Traditional methods of 3D reconstruction are computationally expensive processes that are highly dependent on the number of images taken. This method of capturing images leads to a reduction in the time required to build the model, thereby reducing the time before actionable information is available. Additionally, building an accurate 3D model not only helps to measure parts of the asset, but also allows for the creation of targeted inspection plans where drones can be directed to go to specific locations on the asset to collect images.

[0013] Other features and aspects will become apparent from the following detailed description taken in conjunction with the drawings and claims. [Brief description of the drawings]

[0014] The features and advantages of the illustrative embodiments, and the manner in which they are accomplished, will become more readily apparent from the following detailed description taken in conjunction with the accompanying drawings, which are briefly described below.

[0015] [Figure 1] FIG. 1 illustrates a robotic system according to some embodiments. [Diagram 2] FIG. 1 illustrates an overview of a site inspection system according to some embodiments. [Diagram 3] FIG. 1 is a block flow diagram illustrating an example process according to some embodiments. [Figure 4] FIG. 1 illustrates an overview of a user interface according to some embodiments. [Diagram 5] FIG. 1 illustrates an overview of a user interface according to some embodiments. [Figure 6] FIG. 13 illustrates an example of how a local plan changes based on the shape of the asset, according to some embodiments. [Figure 7] FIG. 1 is a block diagram of a computing system in accordance with some embodiments.

[0016] Throughout the drawings and detailed description, unless otherwise stated, the same drawing reference numbers will be understood to refer to the same elements, features, and structures. The relative size and depiction of these elements may be exaggerated or adjusted for clarity, explanation, and / or convenience. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0017] In the following description, specific details are set forth to provide a thorough understanding of various exemplary embodiments. Various modifications to the embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other embodiments and applications without departing from the spirit and scope of the present disclosure. Moreover, in the following description, numerous details are set forth for purposes of explanation. However, those skilled in the art should understand that the embodiments may be practiced without the use of these specific details. In other instances, well-known structures and processes are not shown or described so as not to obscure the description with unnecessary detail. Thus, the present disclosure is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features disclosed herein.

[0018] Some embodiments described herein relate to providing unmanned robots with intelligent and automated environmental sensing and modeling. Such embodiments may be used to or industrial assets associated with various entities, including corporate entities, governments, individuals, non-profit organizations, etc. As described herein, such assets are typically discrete, limited in scope (e.g., airplanes, helicopters, ships, submarines, space rockets, satellites, locomotives, etc.), or geographically distributed (e.g., roads or railroad tracks, ports or airports, pipelines or electrical infrastructure, power generation facilities or manufacturing plants, etc.). Some embodiments described herein may be used to inspect these types of assets (and others not listed) in a manual, autonomous, or semi-autonomous manner using robotic intermediation.

[0019] FIG. 1 illustrates a robotic system 100 according to some embodiments. With reference to FIG. 1, the robotic system 100 includes an unmanned robot 130 that collects multi-sensory data around an asset 140. For example, the unmanned robot 130 can include an unmanned aerial vehicle (UAV), such as a drone, a crawling robot, a rolling robot, a walking robot, an autonomous underwater vehicle (AUV), and the like. In the example of FIG. 1, the unmanned robot 130 includes a docking station 132 for recharging and storage. The unmanned robot 130 can also communicate with a controller 120 and / or a base station 110. For example, the controller 120 and the base station 110 can receive data from and transmit data to the unmanned robot 130 via a data communication channel. Also, in some examples, the base station 110 can receive input from a user (e.g., an engineer) that sends commands to the unmanned robot 110 via the data communication channel. The communication channel can be wireless, such as, for example, satellite, cellular, local, and / or wired in some cases. For example, a user can input specific information about the asset, such as asset dimensions, asset type, required data fidelity, etc., which is used to build a mission plan that adheres to various mission constraints. The mission is then wirelessly transmitted to the unmanned aerial vehicle 130, which performs the mission and collects the required imagery (e.g., both 2D and 3D). These images, along with metadata, are transmitted to a 3D reconstruction system to build an accurate 3D model of the asset.

[0020] The asset 140 may be a structure located at a geographic location accessible by the unmanned robot 130. For example, the asset 140 may be an oil platform located at sea, a tank or well located below the earth's surface, a pipeline located along the earth's surface, a bridge, a dam, a gas flare, a turbine, a power grid, an aircraft, a locomotive, etc. Based on a mission plan, the unmanned robot 130 may autonomously move around the asset 140 while inspecting (e.g., collecting data on) the asset 140. In the example of FIG. 1, the unmanned robot 130 is a drone that may fly around the asset 140. The drone may land at a location on the surface of the asset 140, pause in mid-air, etc. As described herein, the drone 130 may fly a path of travel around the asset 140 based on a 3D model of a virtual path of travel included in the flight plan. In some cases, the drone 130 may also receive instructions providing a path of travel from the docking station 132 to a start point of the flight plan.

[0021] It should be noted that the unmanned robot 130 may include at least one sensor to collect sensor data. According to some embodiments, the sensor may be associated with, for example, a camera (e.g., a red-green-blue ("RGB") camera), a video camera, an infrared ("IR") camera, a microphone, a chemical detector, a light detection and ranging ("LIDAR") sensor, a radiation detector, or the like. It should be noted that while some examples are described herein in connection with a single inspection device or robot, an inspection (e.g., site survey) plan may instead be associated with multiple inspection devices or robots that simultaneously collect information.

[0022] According to various embodiments, the unmanned robot 130 may perform inspection of multiple areas of interest of the asset 140 based on a 3D virtual mission plan received from the base station 110, the controller 120, or entered locally via the unmanned robot 130. The areas of interest may be located at various distributed locations throughout the asset 140, clustered within a predefined area of ​​the asset 140, and / or the like. Each area of ​​interest includes the points of interest to be inspected on the asset, the location of the area of ​​interest, etc., the angle at which the camera (or other sensor) is positioned relative to the surface of the asset in the area of ​​interest, the turning angle about the asset that the drone should move while capturing sensor data, etc.

[0023] 2 is a diagram illustrating an overview of a site inspection system according to some embodiments. The system uses asset-specific constraints 210 and site inspection constraints 220 provided by an operator (e.g., user, technician, engineer, etc.) to generate a site inspection plan (e.g., mission) at 230. The plan is then executed by a UAV (e.g., drone) and updated on-the-fly at 240 to optimize the local geometry. Once the necessary two-dimensional (2D) and three-dimensional (3D) images have been collected, they are passed to a 3D reconstruction engine at 250 to generate textures, metrics, and a 3D model of the asset.

[0024] Reference is now made to Figures 3 to 6, which are described together. Figure 3 is a block flow diagram illustrating an example process 300 according to some embodiments. For example, the process 300 may be performed by the software and / or systems described herein. The process 300 provides a structure for building an accurate 3D model of an asset.

[0025] Figure 4 is a diagram of a user interface 400 according to some embodiments.Figure 5 is a diagram of a user interface 500 according to some embodiments.

[0026] Figure 6 provides an overview of how the local / initial plan for a portion of an asset changes based on the geometry of the asset, which may be determined by 2D and 3D sensors on the drone.

[0027] With reference to FIG. 3 , the site reconnaissance planning and execution system includes a site reconnaissance user interface (UI) 310 , a mission planner 320 , a flight planner 330 , a mission executor 340 , and a perception engine 350 .

[0028] To generate a site review plan, a user interface (such as an application or other software) may be provided and may be interacted with via a user device, such as a base station, tablet, or other computer, whereby a user, such as an inspector or operator, can input asset-specific and site review constraints for the inspection site. Here, the site review UI 310 provides a means for a user of the system to input constraints (e.g., site and mission specific constraints). For example, as shown in FIG. 4, a user can select parameter values ​​(e.g., site and mission specific constraints) from a drop-down menu 402 or similar user interface feature for transmission to the mission planner 320.

[0029] In some embodiments, constraints may include asset constraints (e.g., asset type, asset dimensions, etc.), 3D reconstruction constraints (e.g., ground sampling distance (GSD), image overlap, camera parameters such as focal length and sensor size), desired resolution (pixels per inch) and / or minimum resolvable features such as cracks of a certain size), and mission constraints (e.g., minimum safe / backoff distance from the asset for the robot, geofence constraints, etc.).

[0030] As a result of the input operation 402, the user interface 400 displays the geofence constraints 40 4 and 406 may be presented, outlining virtual boundaries that designate one or more areas of interest and / or areas of no interest. For example, a flight plan should not include any areas within range / cube area 404, and should not exceed boundary 406.

[0031] The mission planner 320 uses the system constraints to calculate a mission plan that provides optimal coverage of the asset to ensure the required image quality and coverage. Since the mission planner 320 generates the mission plan without access to a three-dimensional model of the asset, representative asset type information is used to generate the plan. The representative asset 502 is represented by a primitive shape and parameters (e.g., cylindrical asset, rectangular asset, horizontal asset, etc.). The extent of the representative asset (e.g., area for inspection) may be determined by information entered via the user interface 400 or by user observation and judgment. This comprehensive shape information is used to generate a mission plan that best covers the asset given the constraints.

[0032] A non-limiting example of a travel path environment is shown in FIG. 5 via a screenshot of a user interface 500. FIG. 5 shows an example of a mission plan and flight plan generated for a representative rectangular asset. The geofence boundaries 404, 406 indicate exclusion zones that a drone cannot violate. The generated mission plan includes a list of camera locations / positions and orientations with respect to the asset 502. In this example, the arrow 505 indicates a waypoint returned from the mission planner 320. The position of the arrow 505 indicates the location of the camera, and the orientation of the arrow indicates the camera orientation (e.g., the viewpoint or viewing angle of one or more sensors on the drone). The line 515 indicates the generated flight plan / path (e.g., a drone flight inspection plan).

[0033] Each travel path (e.g., flight path) may correspond to a series of travel paths and waypoints in real space. For example, a series of travel paths (e.g., areas of interest on a flight path) may be connected by a solid line 515 between a virtually selected start point 510 and a virtually selected end point 520 of the travel path 515 in virtual space. The line of the travel path 515 may move continuously while generally changing in each of the X, Y, and Z coordinate axes in virtual space, thereby simulating the continuous three-dimensional movement of a highly accurate robot. Furthermore, the travel path 515 may circle the asset 502 one or more times while simultaneously changing in each of the X, Y, and Z coordinate axes along a continuous line generated by the system described herein.

[0034] The travel path 515 in the virtual space may be aligned with the physical travel path in the physical space (e.g., the real world) by, for example, a flight plan application. Waypoints in the real space correspond to areas of interest in the virtual space and are points from which the robot captures data and photos. For example, in addition to video and still images, the captured data may include thermal images, ultraviolet images, chemical concentration sniffing, etc. of various asset parts, asset sub-areas, etc. The path planning may be performed in multiple passes, for example using RGB and IR cameras separately. Additionally, the site inspection plan may be adjusted at the inspection site to focus on specific points of interest. For example, an IR camera pass may detect hot spots that are set for further site inspection using the RGB camera.

[0035] The interfaces 400, 500 may be presented on any type of display device (e.g., desktop monitor, smartphone display, tablet display) provided by any type of client (e.g., desktop system, smartphone, tablet computer). The application executed to provide the interfaces 400, 500 may be a web browser, a stand-alone application, , or any other application. Embodiments are not limited to interfaces 400, 500 and may include user interfaces having any particular type, format, or functionality.

[0036] The mission plan information is then provided to the flight planner 330. The flight planner 330 calculates feasible flight plans that the drone can execute given one or more predefined criteria including environment / site, drone, and task constraints. The environment / site constraints include exclusion zones where the drone is not allowed to fly for safety reasons. The drone constraints include information about the drone's location and gimbal limitations. The task constraints can range from finding the shortest path (minimum distance / time, etc.) from one waypoint to the next to the path with the least amount of energy consumed. In some cases, the least amount of energy consumed may include the minimum amount of climb required by the unmanned robot, since going against gravity can consume the most power. The flight planner 330 finds the optimal path that meets all the above constraints and generates a flight trajectory / plan that connects the various waypoints.

[0037] The flight plan information is then provided to the mission executor 340. The mission execution module 340 receives the flight plan from the flight planner 330 and executes the sub-plan (e.g., local plan) to each waypoint. The mission executor also monitors the state of the robot (drone state, payload state, etc.) and takes action accordingly. For example, if the drone is flying in windy conditions and the drone is off its trajectory, the mission executor provides control commands to the drone to bring it back to the required trajectory or abort the execution if necessary. Once the drone reaches the target waypoint, the mission executor adjusts the gimbal orientation and captures the required imagery.

[0038] Referring to FIG. 6, as described above, the generated mission plan 615 is based on the assumption that the asset resembles a particular representative shape 610. However, as shown in FIG. 6, in reality, the actual shape 620 of the asset may have many local variations. The perception engine 350 provides a method to infer the local shape 625 of the asset and adapt the mission plan accordingly. The perception engine uses information from the 2D camera and LIDAR system to calculate the local shape 625 (e.g., local area constraints) of the asset 620 at the waypoint. This information is used to modify the mission plan. Some ways to modify the mission include changing the orientation of the camera that captures the image, adding a waypoint near the waypoint that best captures the local shape, and modifying an already planned waypoint that is locally nearby based on the additional information.

[0039] FIG. 7 illustrates a computing system 700 for generating a site survey (e.g., data collection) plan for building a 3D model of an asset, according to some embodiments. For example, the computing system 700 may be a device or group of devices that communicates with and controls a robot via a wired or wireless connection. With reference to FIG. 7, the computing system 700 includes a network interface 710, a processor 720, a display 730, an output 740, and a storage 750. Although not shown in FIG. 7, the robotic system 700 may include other components such as an input unit, one or more storage devices, etc. The network interface 710 may be a data communication interface that transmits and receives data over a network, such as the Internet, a private network, a public network, via a satellite network, a cellular network, a local network, etc.

[0040] The processor 720 may include one or more processing devices, each including one or more processing cores. In some examples, the processor 720 is a multi-core processor or multiple multi-core processors. The processor 720 may also be deterministic or reconfigurable. According to various embodiments, the processor 720 may execute a site inspection planning application that includes a user interface configured to display a site inspection plan for execution by the unmanned robot. The user interface may receive, and the processor 720 may detect, a selection of a number of asset-specific and site inspection constraints for collecting images.

[0041] The storage 750 may store a site inspection plan including the robot's travel path and the area of ​​interest of the asset to be inspected. The storage 750 may also store data captured while the computing system 700 executes the site inspection process. In an exemplary embodiment, images are captured at the planned location and metadata such as the drone's position and orientation are stored in the storage 750 associated with the system 700. The storage device 750 is not limited to a particular storage device and may include any known memory device such as RAM, ROM, hard disk, etc. The output 740 may output the site inspection plan including the unmanned robot's travel path generated for the asset to a display or to another device or system such as the unmanned robot itself and / or a control device.

[0042] The display 730 can display a user interface and other data, such as via an embedded display of the system 700 or another device externally connected to the system 700. The output 740 can output the user interface and other data to the display 730.

[0043] While examples herein are provided mostly with respect to land-based assets, it should be understood that embodiments can be incorporated into underwater drones for inspecting underwater assets. In this situation, a virtually created 3D travel path is created in the same manner, but instead of through air, the travel path is through water or other liquid-based environments such as oil.

[0044] As will be appreciated based on the foregoing specification, the above examples of the present disclosure may be implemented using computer programming or engineering techniques, including computer software, firmware, hardware, or any combination or subset thereof. Such resulting programs having computer readable code may be embodied or provided in one or more non-transitory computer readable media, thereby making a computer program product, i.e., an article of manufacture, in accordance with the described examples of the present disclosure. For example, the non-transitory computer readable medium may be, but is not limited to, a fixed drive, a diskette, an optical disk, a magnetic tape, a semiconductor memory such as a flash memory, a read-only memory (ROM), and / or a transmitting / receiving medium such as the Internet, cloud storage, the Internet of Things, or other communication networks or links. An article of manufacture including the computer code may be manufactured and / or used by executing the code directly from one medium, by copying the code from one medium to another, or by transmitting the code over a network.

[0045] A computer program (also referred to as a program, software, software application, "app", or code) may include machine instructions for a programmable processor and may be implemented in a high-level procedural and / or object-oriented programming language and / or assembly / machine code. As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, device, cloud storage, Internet of Things, and / or any other software or hardware that may be used to implement the computer program product. Refers to devices used to provide machine instructions and / or data to a programmable processor including machine-readable media that receive the machine instructions as machine-readable signals (e.g., magnetic disks, optical disks, memory, programmable logic devices (PLDs)). However, "machine-readable media" and "computer-readable media" do not include transitory signals. The term "machine-readable signal" refers to any signal that can be used to provide machine instructions and / or any other type of data to a programmable processor.

[0046] The above description and illustration of the processes herein should not be construed as implying a fixed order for performing the process steps. Rather, the process steps may be performed in any order possible, including simultaneous performance of at least some steps. Although the present disclosure has been described in conjunction with specific examples, it should be understood that various changes, substitutions, and alterations apparent to those skilled in the art may be made to the disclosed embodiments without departing from the spirit and scope of the present disclosure as set forth in the appended claims.

[0047] Further aspects of the invention are provided by the subject matter of the following sections.

[0048] [Section 1] 1. A system for generating a three-dimensional model of an asset, comprising: A memory for storing instructions; and a processor configured to execute the instructions, wherein the executed instructions cause the processor to: an operation of receiving input parameters corresponding to constraints of a mission plan for operating the unmanned vehicle around the asset; generating the mission plan based on the input parameters including asset type information representative of the asset, the mission plan including waypoints identifying positions and orientations of one or more image sensors of the unmanned vehicle; generating a flight path for the unmanned vehicle that connects the waypoints that meet one or more predefined criteria; determining, at each waypoint, a local shape of the asset as sensed by the one or more image sensors; modifying the mission plan on the fly based on the local geometry; and and causing the system to perform operations to capture images of the asset along waypoints of the modified mission plan.

[0049] [Section 2] 2. The system of any preceding claim, wherein the processor is further configured to monitor a vehicle state of the unmanned vehicle during execution of the flight path from one waypoint to a next waypoint.

[0050] [Section 3] 13. The system of any preceding claim, wherein the input parameters include at least one of an asset constraint, a 3D reconstruction constraint, and a mission constraint.

[0051] [Section 4] 2. The system of any preceding claim, wherein the input parameters include at least one of a geofence, a camera parameter, and an asset dimension.

[0052] [Section 5] The system of any preceding paragraph, wherein the one or more image sensors are associated with at least one of: (i) a camera, (ii) a video camera, (iii) an infrared camera, (iv) a microphone, (v) a chemical detector, (vi) a light detection and ranging ("LIDAR") sensor, and (vii) a radiation detector.

[0053] [Section 6] 13. The system of any preceding claim, wherein the asset types are expressed in terms of primitive shapes and parameters.

[0054] [Section 7] 2. The system of any preceding claim, wherein modifying the mission plan includes at least one of changing the orientation of the one or more image sensors, adding a waypoint, and modifying the waypoint.

[0055] [Section 8] 13. The system of any preceding claim, wherein the one or more predefined criteria include environmental constraints, vehicle constraints, and task constraints.

[0056] [Section 9] 1. A method for generating a three-dimensional model of an asset, comprising: receiving input parameters corresponding to constraints of a mission plan for operating the unmanned vehicle around the asset; generating the mission plan based on the input parameters including asset type information representative of the asset, the mission plan being configured to include one or more of the unmanned vehicles; comprising waypoints identifying positions and orientations of a plurality of image sensors; generating a flight path for the unmanned vehicle that connects the waypoints that meet one or more predefined criteria; monitoring a vehicle state of the unmanned vehicle during execution of the flight path from one waypoint to a next waypoint; determining, at each waypoint, a local shape of the asset as sensed by the one or more image sensors; modifying the mission plan on the fly based on the local geometry; and The method includes capturing images of the asset along waypoints of the modified mission plan.

[0057] [Section 10] 20. The method of any preceding claim, further comprising monitoring a vehicle state of the unmanned vehicle during execution of the flight path from one waypoint to a next waypoint.

[0058] [Section 11] 13. The method of any preceding claim, wherein the input parameters include at least one of an asset constraint, a 3D reconstruction constraint, and a mission constraint.

[0059] [Section 12] 13. The method of any preceding claim, wherein the input parameters include at least one of a geofence, a camera parameter, and an asset dimension.

[0060] [Section 13] The method of any preceding claim, wherein the one or more image sensors are associated with at least one of: (i) a camera, (ii) a video camera, (iii) an infrared camera, (iv) a microphone, (v) a chemical detector, (vi) a light detection and ranging ("LIDAR") sensor, and (vii) a radiation detector.

[0061] [Section 14] 13. The method of any preceding claim, wherein the asset types are expressed in terms of primitive shapes and parameters.

[0062] [Section 15] 13. The method of any preceding claim, wherein modifying the mission plan includes at least one of changing the orientation of the one or more image sensors, adding a waypoint, and modifying the waypoint.

[0063] [Section 16] 13. The method of any preceding claim, wherein the one or more predefined criteria include environmental constraints, vehicle constraints, and task constraints.

[0064] [Section 17] 1. A non-transitory computer readable medium having stored thereon instructions that, when executed, cause a computer to perform a method for generating a three-dimensional model of an asset, the method comprising: receiving input parameters corresponding to constraints of a mission plan for operating the unmanned vehicle around the asset; generating the mission plan based on the input parameters including asset type information representative of the asset, the mission plan including waypoints identifying positions and orientations of one or more image sensors of the unmanned vehicle; generating a flight path for the unmanned vehicle that connects the waypoints that meet one or more predefined criteria; monitoring a vehicle state of the unmanned vehicle during execution of the flight path from one waypoint to a next waypoint; determining, at each waypoint, a local shape of the asset as sensed by the one or more image sensors; modifying the mission plan on the fly based on the local geometry; and capturing images of the asset along waypoints of the modified mission plan.

[0065] [Section 18] 2. The non-transitory computer-readable medium of any preceding claim, wherein the method further includes monitoring a vehicle state of the unmanned vehicle during execution of the flight path from one waypoint to a next waypoint.

[0066] [Section 19] 7. The non-transitory computer-readable medium of any preceding claim, wherein the input parameters include at least one of an asset constraint, a three-dimensional reconstruction constraint, and a mission constraint.

[0067] [Section 20] 7. The non-transitory computer-readable medium of any preceding claim, wherein the one or more predefined criteria include environmental constraints, vehicle constraints, and task constraints.

Claims

1. 1. A system for generating a three-dimensional model of an asset, comprising: a memory for storing instructions; and a processor configured to execute the instructions, wherein the executed instructions cause the processor to: an operation of receiving input parameters corresponding to constraints of a mission plan for operating the unmanned robot around the asset; generating the mission plan based on the input parameters including asset type information representative of the asset, the mission plan including a plurality of waypoints, the mission plan identifying a position and orientation of one or more image sensors at each waypoint of the unmanned robot; generating a flight path for the unmanned robot that connects the plurality of waypoints that meet one or more predefined criteria; detecting, along the flight path, by the one or more image sensors, a localized portion of the asset as being deformed; calculating a shape of the detected local portion and modifying the mission plan on the fly based on the shape of the local portion; and and causing the system to perform operations to capture images of the localized portion along a plurality of waypoints of the modified mission plan.

2. The system of claim 1 , wherein the processor is further configured to monitor a robotic state of the unmanned robot during execution of the flight path from one waypoint to a next waypoint.

3. The system of claim 1 , wherein the input parameters include at least one of an asset constraint, a three-dimensional reconstruction constraint, and a mission constraint.

4. The system of claim 3 , wherein the input parameters include at least one of a geofence, a camera parameter, and an asset dimension.

5. 2. The system of claim 1, wherein the one or more image sensors are associated with at least one of: (i) a camera; (ii) a video camera; (iii) an infrared camera; (iv) a microphone; (v) a chemical detector; (vi) a light detection and ranging ("LIDAR") sensor; and (vii) a radiation detector.

6. The system of claim 1 , wherein the asset types are represented in terms of primitive shapes and parameters.

7. 2. The system of claim 1, wherein modifying the mission plan includes at least one of changing the orientation of the one or more image sensors, adding a waypoint, and modifying the waypoint.

8. The system of claim 1 , wherein the one or more predefined criteria include environmental constraints, robot constraints, and task constraints.

9. 1. A method for generating a three-dimensional model of an asset, comprising: receiving input parameters corresponding to constraints of a mission plan for operating the unmanned robot around an asset; generating the mission plan based on the input parameters including asset type information representative of the asset, the mission plan including a plurality of waypoints, the mission plan identifying a position and orientation of one or more image sensors at each waypoint of the unmanned robot; generating a flight path for the unmanned robot that connects the plurality of waypoints that meet one or more predefined criteria; detecting, along the flight path, by the one or more image sensors, a localized portion of the asset as being deformed; calculating a shape of the detected local area and modifying the mission plan on the fly based on the shape of the local area; and capturing images of the localized portion along a plurality of waypoints of the modified mission plan.

10. The method of claim 9 , further comprising monitoring a robotic state of the unmanned robot during execution of the flight path from one waypoint to a next waypoint.

11. The method of claim 9 , wherein the input parameters include at least one of an asset constraint, a three-dimensional reconstruction constraint, and a mission constraint.

12. The method of claim 11 , wherein the input parameters include at least one of a geofence, a camera parameter, and an asset dimension.

13. 10. The method of claim 9, wherein the one or more image sensors are associated with at least one of: (i) a camera; (ii) a video camera; (iii) an infrared camera; (iv) a microphone; (v) a chemical detector; (vi) a light detection and ranging ("LIDAR") sensor; and (vii) a radiation detector.

14. The method of claim 9 , wherein the asset types are represented in terms of primitive shapes and parameters.

15. 10. The method of claim 9, wherein modifying the mission plan includes at least one of changing the orientation of the one or more image sensors, adding a waypoint, and modifying the waypoint.

16. The method of claim 9 , wherein the one or more predefined criteria include environmental constraints, robot constraints, and task constraints.

17. 1. A non-transitory computer readable medium having stored thereon instructions that, when executed, cause a computer to perform a method for generating a three-dimensional model of an asset, the method comprising: receiving input parameters corresponding to constraints of a mission plan for operating the unmanned robot around an asset; generating the mission plan based on the input parameters including asset type information representative of the asset, the mission plan including a plurality of waypoints, the mission plan identifying a position and orientation of one or more image sensors at each waypoint of the unmanned robot; generating a flight path for the unmanned robot that connects the plurality of waypoints that meet one or more predefined criteria; detecting, along the flight path, by the one or more image sensors, a localized portion of the asset as being deformed; calculating a shape of the detected local area and modifying the mission plan on the fly based on the shape of the local area; and capturing images of the localized portion along a plurality of waypoints of the modified mission plan.

18. 20. The non-transitory computer-readable medium of claim 17, wherein the method further comprises monitoring a robotic state of the unmanned robot during execution of the flight path from one waypoint to a next waypoint.

19. 20. The non-transitory computer-readable medium of claim 17, wherein the input parameters include at least one of an asset constraint, a three-dimensional reconstruction constraint, and a mission constraint.

20. 20. The non-transitory computer-readable medium of claim 17, wherein the one or more predefined criteria include environmental constraints, robot constraints, and task constraints.

Citation Information

Patent Citations

  • Inspection system and inspection method of structure

    JP2017138162A

  • Determination of object data by template-based UAV control

    US20160253808A1

  • monitoring

    WO2017005976A1

  • Image capture system for shape measurement of structure, method of capturing image of stucture used for shape measurement of structure, onboard control device, remote control device, program, and recording medium

    WO2017073310A1