Robotic ai agent for construction guidance, monitoring, and documentation

The laser-based spatial augmented reality system addresses the challenge of translating design documents into accurate construction instructions by projecting precise guidance onto construction sites, enhancing alignment and documentation efficiency.

WO2026241494A1PCT designated stage Publication Date: 2026-11-19UNIV OF FLORIDA RESEARCH FOUNDATION INC
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
PCT/US2026/028146
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2026-02-17
Filing Date
2026-05-14
Publication Date
2026-11-19

AI Technical Summary

Technical Problem

The construction industry faces significant challenges in translating design documents into accurate and timely construction instructions, leading to errors, rework, and cost overruns due to the lack of contextual, digital, and as-design information on unstructured and dynamic construction sites.

Method used

A laser-based spatial augmented reality system that uses a projector and sensors on a movable platform to project precise guidance onto an as-built surface, aligning design data with the actual site conditions, measure deviations, and generate corrective outputs, while creating an exportable task record for documentation.

Benefits of technology

This system provides real-time, accurate construction guidance, reducing errors and rework by bridging the gap between digital design intent and physical execution, ensuring precise alignment and documentation across teams.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure US2026028146_19112026_PF_FP_ABST
    Figure US2026028146_19112026_PF_FP_ABST
Patent Text Reader

Abstract

A method of the present disclosure comprises providing a field execution device; receiving design data representing an installation target; generating projector-ready guidance data comprising a pattern and one or more projection parameters for projecting guidance associated with the installation target; estimating an alignment between the design data and the as-built surface using one or more alignment techniques; directing a movable platform of the field execution device to move into position to achieve the alignment between a projector and the as-built surface; causing the projector to project guidance of the installation target onto an as-built location based on the projector-ready guidance data and device-specific projector calibration parameters; and / or measuring a deviation between the projected guidance and the as-built surface and generating a corrective output based on the deviation.
Need to check novelty before this filing date? Find Prior Art

Description

Docket T19763W0001 (222112-2650)ROBOTIC Al AGENT FOR CONSTRUCTION GUIDANCE, MONITORING, AND DOCUMENTATIONCROSS-REFERENCE TO RELATED APPLICATION

[0001] This application claims priority to and the benefit of co-pending U.S. provisional application number 63 / 984,595, filed February 17, 2026, and co-pending U.S. provisional application number 63 / 806,421 , filed May 15, 2025, each of which is entirely incorporated herein by reference.STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT

[0002] This invention was made with government support under Grant No.2524239, awarded by the National Science Foundation. The government has certain rights in the invention.BACKGROUND

[0003] The construction industry in the United States is experiencing significant growth, with projections suggesting that its overall revenue will reach $15.5 trillion by 2030. Despite this promising outlook, an estimated $17 billion to $36 billion is lost during translating design documents to real construction. As such, construction workers often struggle with insufficient access to precise and actionable information on unstructured and dynamic construction sites. Many still rely on paper drawings and verbal instructions from site supervisors, a practice that frequently results in errors, rework, delays, and cost overruns in the construction industry. The primary issue is the absence of contextual, digital, and as-design information that can effectively guide on-site activities. Hence, there is an urgent need to devise systems and methods thatDocket T19763W0001 (222112-2650)ensure design documents are translated into clear and accurate construction instructions vividly and timely, thereby minimizing potential errors.SUMMARY

[0004] Embodiments of the present disclosure provide systems and methods for construction guidance, monitoring, and documentation. One such method comprises providing a field execution device comprising a projector and one or more sensors, the field execution device being coupled to a movable platform, and providing computing hardware in communication with the field execution device; receiving design data representing an installation target, the design data comprising geometric information including object positions, vertex coordinates, and dimensions; generating projectorready guidance data comprising a pattern and one or more projection parameters for projecting guidance associated with the installation target; storing, in a projection library, the projector-ready guidance data associated with (i) an identifier of the installation target and (ii) a revision of the design data; receiving, via a user interface, a selection of the installation target for field execution, wherein the selection is associated with the identifier and the revision; retrieving, by the computing hardware, the projector-ready guidance data from the projection library; retrieving, by the computing hardware from the projection library, device-specific projector calibration parameters for projecting onto an as-built surface corresponding to the installation target; estimating, using the one or more sensors, an alignment between the design data and the as-built surface using one or more alignment techniques; directing, by the computing hardware, the movable platform to move into position to achieve the alignment between the projector and the as-built surface; causing the projector to project guidance of the installation target onto an as-built location, based on the projector-ready guidance data and the device-specific projector calibrationDocket T19763W0001 (222112-2650)parameters; measuring a deviation between the projected guidance and the as-built surface and generating a corrective output based on the deviation; and / or generating an exportable task record including the identifier, the revision, one or more alignment parameters, the deviation, and a timestamp.

[0005] The present disclosure can also be viewed as providing a laser-based spatial augmented reality system comprising: one or more processors; a projection library storing projector-ready guidance data for installation targets, each installation target being associated with an identifier and a revision of design data; a communication interface configured to receive a selection of an installation target for field execution; a field execution device comprising a projector and one or more sensors coupled to a movable platform; and instructions that, when executed by the one or more processors, cause the system to: retrieve projector-ready guidance data from the projection library based on the identifier and the revision; retrieve, from the projection library, device-specific projector calibration parameters for projecting onto an as-built surface; estimate an alignment between the design data and the as-built surface using one or more alignment techniques; control the projector to project guidance corresponding to the installation target onto the as-built surface based on the retrieved projector-ready guidance data and the device-specific projector calibration parameters; measure a deviation and generate a corrective output; and / or generate an exportable task record associated with the identifier and the revision.

[0006] The present disclosure can also be viewed as providing a non-transitory computer-readable medium having instructions stored therein, wherein the instructions, when executed by one or more processors, cause the one or more processors to: receive design data representing an installation target; generate projector-ready guidance data for the installation target and store the projector-readyDocket T19763W0001 (222112-2650)guidance data in a projection library, wherein the projector-ready guidance data is associated with an identifier of the installation target and a revision of the design data; receive a selection of the installation target for field execution; retrieve the projectorready guidance data from the projection library based on the identifier and the revision; retrieve, from the projection library, device-specific projector calibration parameters for projecting onto an as-built surface; estimate an alignment between the design data and the as-built surface using (i) one or more sensors and (ii) one or more alignment techniques; position a field execution device comprising a laser projector and the one or more sensors to align the laser projection with the as-built surface based on the alignment; cause the laser projector to project the projector-ready guidance data onto the as-built surface based on the retrieved guidance data and the device-specific projector calibration parameters; perform a closed-loop verification by measuring a deviation between the projected guidance data and the as-built surface and generating a corrective output based on the deviation; and / or generate an exportable task record including the identifier, the revision, at least one alignment parameter, the deviation, and a timestamp.

[0007] In one or more aspects for such methods, systems, or mediums, the projector projects visual guidance directly onto the as-built surface of the as-built location; the visual guidance comprises a step-by-step sequence of instructions for installing a physical representation of the installation target at the as-built location; the selection of the installation target comprises a user’s spoken request, a text input, a graphical user interface selection, a gesture input, a button input, ora scanned code; measuring the deviation comprises computing at least one of a translational offset, a rotational offset, or a fit error relative to a tolerance threshold, and wherein generating the corrective output comprises updating at least one projection parameter; theDocket T19763W0001 (222112-2650)projector is mounted on at least one of: an end-effector, a wrist, a mast, a gimbal, a base of the movable platform, a tripod, or a stationary mount; the projector projects visual guidance directly onto the as-built surface of an as-built location; and / or the visual guidance comprises a step-by-step sequence of instructions for installing a physical representation of the installation target at the as-built location.

[0008] In one or more aspects, such methods, systems, or mediums further perform computing a scale of the installation target required to render an outline of the installation target at a correct physical location on the as-built location.

[0009] Other systems, methods, features, and advantages of the present disclosure will be or become apparent to one with skill in the art upon examination of the following drawings and detailed description. It is intended that all such additional systems, methods, features, and advantages be included within this description and be within the scope of the present disclosure.BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Many aspects of the present disclosure can be better understood with reference to the following drawings. The components in the drawings are not necessarily to scale, emphasis instead being placed upon clearly illustrating the principles of the present disclosure. Moreover, in the drawings, like reference numerals designate corresponding parts throughout the several views.

[0011] FIG. 1 is a block diagram of a laser-based spatial augmented reality system in accordance with various embodiments of the present disclosure.

[0012] FIG. 2A illustrates an example table that maps object identifiers of a building information model (BIM) into appropriate projector-ready laser patterns and parameters, in accordance with various embodiments of the present disclosure.Docket T19763W0001 (222112-2650)

[0013] FIG. 2B provides an illustration of an exemplary graphical worker interface in accordance with various embodiments of the present disclosure.

[0014] FIGS. 2C and 2D provide illustrations of exemplary graphical laser control projection interfaces in accordance with various embodiments of the present disclosure.

[0015] FIG. 3 illustrates an exemplary model information extraction process applied by an analysis and decision software module of the laser-based SAR system of FIG. 1.

[0016] FIG. 4 illustrates an exemplary process for converting a shape of a BIM object into a laser projection file, in accordance with various embodiments of the present disclosure.

[0017] FIG. 5 illustrates an exemplary workflow of the laser-based SAR system of FIG. 1.

[0018] FIG. 6 is a block diagram showing power module and laser control module components of an exemplary laser projector in the laser-based SAR system of FIG. 1.

[0019] FIG. 7 illustrates an example projection distortion that can be corrected using embodiments of the laser-based SAR system of FIG. 1.

[0020] FIG. 8 illustrates an operation workflow of an exemplary laser-based SAR system, in accordance with various embodiments of the present disclosure.

[0021] FIG. 9 shows conversion laser projection outputs of various objects of a BIM model, in accordance with various embodiments of the present disclosure.

[0022] FIG. 10 shows laser patterns used to evaluate the projection accuracy of a prototype of the laser-based SAR system of FIG. 1.

[0023] FIG. 11 shows geometric patterns used to evaluate a shape stability of a prototype of the laser-based SAR system of FIG. 1.Docket T19763W0001 (222112-2650)

[0024] FIG. 12 shows a reference BIM model and its embedded objects that were used to study the effectiveness of a prototype of the laser-based SAR system of FIG.1.

[0025] FIG. 13 depicts a schematic block diagram of a computing device that can be used to implement various embodiments of the present disclosure.DETAILED DESCRIPTION

[0026] Systems and methods of the present disclosure involve laser-based spatial augmented reality (SAR) technology that is designed to interact with building information modeling (BIM) data and / or computer-aided design (CAD) data. In various embodiments, such systems and methods translate natural language task intent into BIM-referenced laser projections on as-built surfaces with closed-loop visual verification and, in some embodiments, robotic navigation and end-effector pose control to realize those projections in the field. An exemplary laser-based SAR system projects vivid, real-time construction information directly onto workspaces to assist users, such as workers. Accordingly, an exemplary laser-based SAR system may be considered as an artificial intelligence (Al) agent that performs tasks, reasoning, planning, and acts autonomously on behalf of users, such as workers in the construction industry. One contribution of this research is the creation of a workflow that computationally translates BIM data from a data library / store into laser projection files, enabling real-time on-site visualization of tasks such as layout, alignment, assembly, installation, and inspection. This approach provides workers with in-situ guidance, bridging the gap between digital design intent and physical construction execution.Docket T19763W0001 (222112-2650)

[0027] The terminology used herein is for the purpose of describing particular examples and is not intended to be limiting. For example, in various embodiments, the disclosed system may be implemented as a construction coordination and spatial guidance platform that connects design stakeholders and field stakeholders through a traceable field-execution loop. As used herein, a “field execution device” refers to any device configured to project spatial guidance in a worksite environment and to sense an as-built condition, including, by way of non-limiting examples, a projector coupled with one or more sensors (e.g., RGB camera, depth camera, RGB-D camera, LiDAR, I MU) and associated computing hardware. The field execution device may be coupled to a “movable platform,” such as a mobile robot base, a robotic arm, a gimbal, a handheld rig, a cart, or a tripod with adjustable pose, and may alternatively be stationary in some embodiments. An “installation target” refers to any physical object, assembly, region, or work item to be located, aligned, installed, assembled, inspected, or verified at an as-built surface, including subcomponents and step-specific intermediate states. “Design data” may include BIM, CAD, drawings, manuals, specifications, schedules, submittals, RFIs (Request for Information), change orders, and other digital artifacts that encode geometry and / or installation intent. An “identifier” may include a GUID (globally unique identifier), object ID, tag, barcode / QR code value, textual label, database key, or other unique reference that enables unambiguous association of an installation target with stored guidance. A “revision” refers to a version designation of the design data (e.g., revision number, timestamp, issue date, submittal revision, change order number, RFI response state, or hash) used to preserve traceability between what was designed and what was executed in the field. “Projector-ready guidance data” refers to a pattern and one or more projection parameters (e.g., object coordinate parameters, scale parameters, dimensionDocket T19763W0001 (222112-2650)parameters, etc. from a design file) enabling deterministic projection of guidance corresponding to an installation target, optionally including tolerances, alignment marks, step sequencing metadata, and verification criteria. An “exportable task record” refers to a machine-readable and / or human-readable record that can be transmitted, stored, and audited across stakeholders and software tools, including at least the identifier, the revision, one or more alignment parameters, one or more deviation values, and a timestamp, and optionally images, depth / point-cloud evidence, and pass / fail determinations or quality scores.

[0028] In various embodiments, the disclosed system is organized as a layered field-execution stack that transforms design intent into on-surface guidance with a traceable closed-loop outcome. A first layer (intent and retrieval layer) interprets user input (e.g., voice / text / GUI selection) and resolves an installation target and step context to a normalized intent record (e.g., {identifier, plane, step, tolerance, revision}). A second layer (guidance authoring and library layer) generates and / or retrieves projector-ready guidance data and device-specific projector calibration parameters (e.g., galvo angle parameters, projector pose parameters, hardware control parameters, etc.) from a projection library keyed by the installation target identifier and design-data revision, thereby enabling deterministic and version-traceable projection. A third layer (alignment and planning layer) estimates alignment between the design data and an as-built surface using one or more alignment techniques (e.g., markerbased localization, feature-based registration, depth / point-cloud registration, and / or inertial fusion) and determines a positioning plan for a field execution device relative to the surface. A fourth layer (execution, verification, and record layer) performs closed-loop projection and measurement (project-sense-correct), computes deviations against tolerance criteria, generates corrective outputs (e.g., updatedDocket T19763W0001 (222112-2650)projection parameters and / or repositioning cues), and generates an exportable task record associated with the identifier and revision for cross-team traceability.

[0029] While the disclosure will be described in connection with the following drawings, there is no intent to limit the disclosure to the embodiment or embodiments disclosed herein. With reference to FIG. 1, there is shown an embodiment of the system 10 of the present disclosure in a manufacturing application, which includes a laser projector 1 that is integrated with a mobile robot vehicle 2. In various embodiments, an end effector 3 of the robot 2 can position the laser projector 1 in a proper pose for accurately displaying spatial information to be relayed to a worker in the field. Accordingly, the laser projector 1 can project visual guidance directly onto physical surfaces without wearable devices, offering hands-free operation and increased worker comfort. This approach allows multiple users to work simultaneously without personal device calibration. Computing hardware 4 is provided in or with the mobile robot vehicle 2 and laser projector 1 to enable robotic navigation and endeffector pose control to realize those projections in the field. The computing hardware 4 may generally include a processor circuit which processes executable instructions of the various software modules of the laser-based SAR system 10. For example, in various embodiments, the computing hardware 4 is configured with a data library 7 that stores one or BIM files and a perception software module 5 that is programmed to resolve a user’s spoken or typed request to a specific BIM object and parent surface and compute a transform that transforms objects in the target BIM to the parent surface. Further, an analysis & decision software module 6 is programmed to retrieve a projector ready pattern from a projection library 17 and compute the pose and scale required to render that pattern at the correct physical location on the parent surface. The computing hardware 4 is further configured with an action execution softwareDocket T19763W0001 (222112-2650)module 8 that can materialize the projection from a mobile base of the robot 2 with an end effector 3 (arm / gimbal) and execute a project-sense-correct loop that measures millimeter-level offsets, issues corrective cues if needed, and logs pass / fail results against the BIM object and installation step. In doing so, the system 10 closes the gap between language, digital design intent, and field-accurate layout. The system 10 may also continuously gather and synthesize distributed data from the job site — via sensors, cameras, and user inputs — and integrate this information into the BIM environment. This two-way feedback project-sense-correct loop can maintain a living digital twin of the construction project, enabling better coordination, traceability, and decision-making. In alternative embodiments, the projector 1 and sensors are coupled to other movable platforms, including an arm-mounted end-effector, a mast, a gimbal, a handheld rig, a cart, or a tripod, and the system is not limited to any particular robot make, model, or mobility method. In some embodiments, multiple field execution devices are used concurrently across a jobsite, and each device is configured to retrieve consistent guidance from a shared projection library to maintain a single source of truth across teams and devices.

[0030] In various embodiments, the projection library 17 is a project specific database that maps each GUID (globally unique identifier) of a BIM object (and selected projection plane) into a projector-ready pattern and its parameters. In some embodiments, each library entry is additionally associated with a revision of the design data so that changes to the design intent are traceable and older guidance can be archived, retrieved, or compared against newer guidance. For every target object, the entry stores the following: the vector pattern (e.g., polyline vertices / ILDA), the designated projection plane, alignment marks, nominal dimensions and scale / margin rules, tolerance bands, and — when applicable — the installation order. At runtime,Docket T19763W0001 (222112-2650)when a user (e.g., operator) selects an object and plane in a laser projection control interface 16, the computing hardware 4 uses the object’s GUID to retrieve the corresponding pattern and parameters for immediate rendering. Because entries are indexed for rapid access, the projector 1 acquires the correct outline quickly and consistently, enabling accurate placement and orientation in the as-built scene. This organization is especially effective for standardized, industrialized building workflows: the stored installation order lets the laser-based SAR system 10 guide crews step-by-step through a logical sequence.

[0031] Next, FIG. 2A illustrates an example library, in the form of a table, for a house project, where each row lists the object type and object ID together with the laser files for different planes (e.g., left, top, front) and the prescribed installation step. This structured approach streamlines retrieval and setup, providing clear, repeatable guidance on site.

[0032] In various embodiments, to enhance communication between a user and the laser-based SAR system 10, two distinct interface modules are implemented in the perception software module 5. A worker interface module 15 allows users (e.g., workers) to directly interact with the BIM model 18, enabling them to select, manipulate, and examine objects within the BIM model 18. Built on xBIM, in various non-limiting embodiments, this interface 15 facilitates BIM model visualization, allowing workers to upload I FC (Industry Foundation Classes (I FC)) files, rotate, pan, zoom, and select target objects. With simple clicks, workers can easily extract object properties, such as GUIDs, which may be used for further processing. In accordance with embodiments of the present disclosure, this module streamlines the interaction with the BIM model 18, making object selection and property extraction straightforward. FIG. 2B provides an illustration of an exemplary graphical workerDocket T19763W0001 (222112-2650)interface, where a user can display a BIM model, select a target object (e.g., window) within the BIM model, and be provided the GIIID for the selected object.

[0033] A second module, the laser projector control interface 16, enables direct interaction with the laser projector 1 itself. In certain embodiments, designed as a webbased control module, it is built on the ESP32 Async Web Service, in various nonlimiting embodiments, providing users with flexible access options. In such implementations, the projector 1 may operate in two modes: station mode and Access Point (AP) mode. In station mode, the projector 1 connects to a local Wi-Fi router, allowing users to control it through a local IP address. In AP mode, a processor or controller (e.g., ESP32 board, processor 1302 (FIG. 13)) of the computing hardware 4 creates a Wi-Fi hotspot, enabling users to access the projector when no Wi-Fi is available. Through this interface 16, users can upload and manage projection files, including ILDA (International Laser Display Association) files, directly on onboard memory (e.g., SD card, memory 1304 (FIG. 13)). For example, users can create playlists for sequential laser projections, delete files, and adjust the projection’s scale and position based on its location, as shown in FIGS. 2C and 2D, which also demonstrate the interface’s functionalities, showing options for file management, playlist creation, and projection calibration. Together, these interfaces enhance the system’s interactivity by providing users (e.g., workers) with tools to interact with both the BIM model 18 and the laser projector 1, streamlining the process from BIM model selection to precise projection setup.

[0034] In various embodiments, the computing hardware generates “projectorready guidance data” for an installation target, where the projector-ready guidance data includes (i) a pattern to be projected (e.g., vector outlines, polylines, alignment marks, fiducial shapes, or error-localizing overlays) and (ii) one or more “projectionDocket T19763W0001 (222112-2650)parameters” that specify how the pattern is rendered on an as-built surface. Nonlimiting examples of projection parameters include scale, translation, rotation, brightness, color channel assignment, line thickness, blanking paths, alignment marks, tolerance bands, and a selected projection plane or surface coordinate frame. In some embodiments, projector-ready guidance data is stored as ILDA files; in other embodiments, the data is stored in other projector-consumable formats and / or as intermediate representations (e.g., JSON or vector graphics) that are compiled into a projector-specific format at runtime.

[0035] In various embodiments, the perception software module 5 is configured to deploy artificial intelligence (Al) to convert a worker’s spoken or typed request, together with the live scene, into a structured task intent and the next installation step and then retrieve the correct projector-ready pattern from the projection library 17. In various embodiments, spoken audio may be transcribed to text and paired with one or more on-site images; and a lightweight captioner (e.g., BLIP / LLaVA-class) produces a short scene caption and object tags that summarize what is visible at the workface. The text, caption, tags, and scheduling context form the retrieval query against three indexed sources: (i) BIM metadata (object IDs, names / aliases, parent surfaces, local placements), (ii) the projection library 17 (pattern vertices, tolerances, installation order), and (iii) tutorials / standards plus the worker’s daily plan (which each may also be stored in or accessed by data library 7). A step retriever of the perception software module 5 ranks candidate tutorial steps and (GUID, plane) pairs using a hybrid similarity score that fuses language, image, and schedule priors; the top candidates and their evidence are passed to an instruction-tuned large language model (LLM) that resolves ambiguities (“second metal frame” vs. “left wall”), enforces dependency rules, and emits a normalized intent record with {guid, plane, offsets, tolerance,Docket T19763W0001 (222112-2650)sequence_step} together with the selected step ID. When the spoken audio is underspecified or conflicts with the schedule, the LLM returns a concise clarification (“Do you mean the front face of Wall W-103?”) rather than proceeding. One or more LLM models are trained on image-caption-question-answer-step ID (tutorial ID) tuples and, for grounding, annotated mappings from utterances to (guid, plane); the captioner and step retriever are fine-tuned on project photos, while the LLM(s) are instruction-tuned to call the retrievers, check dependencies, and format outputs. In various embodiments, in practice, a retrieval augmented model (RAG) produces deterministic projection plan inputs for the analysis layer, while keeping the language interface natural for workers (e.g., one user type) and auditable for managers (e.g., second user type).

[0036] To place a pattern at the correct physical location, the analysis & decision software module 6 aligns BIM geometry to the jobsite and expresses target objects relative to their parent surface. As shown in FIG. 3, a user selects the target object in BIM (e.g., a toilet roughing on a wall), in which the computing hardware 4 of the laserbased SAR system 10 is configured to read the object’s local placement Ttarget(e.g., using ifcOpenShell and the Industry Foundation Classes (I FC) file format) and the parent surface’s placement and Tparent, together with the object’s dimensions. The object’s pose relative to the parent surface is then'T*r— 1rrilrel1parent1target

[0037] This transform expresses the target in the parent-surface coordinate frame (origin, in-plane axes, and normal), which is the frame used for both projection and verification. In parallel, BIM can export a CAD view of the same object. The computing hardware 4 of the laser-based SAR system 10 can extract the object’s vertex coordinates on the chosen plane (e.g., using via AutoLISP in the AutoCADDocket T19763W0001 (222112-2650)environment), yielding a compact, quantitative shape description (vertices, type, size). Combining this surface-relative pose Tretwith the plane-metric pattern from the projection library 17 provides all data required to compute the projector pose and scale, ensuring the outline is placed precisely on the as-built surface and ready for closed-loop quality assurance (QA) in subsequent steps.

[0038] In determining the precise location of the laser projector 1 relative to the target object, the computing hardware 4 of the laser-based SAR system 10 integrates both digital information from the BIM model 18 and physical on-site calibration. For example, after a worker selects the target object and retrieves its associated laser file and projection plane ID from the projection library 17, the computing hardware 4 of the laser-based SAR system 10 may use the relative positioning data obtained (e.g., through ifcopenshell). This data includes the placement of the target object and its parent surface, allowing the computing hardware 4 of the laser-based SAR system 10 to establish the expected projection location in the digital model’s coordinate frame.

[0039] Extracting the design information from the 2D CAD model and 3D BIM model is detailed in FIG. 3. Here, in various embodiments, a user first selects the target object within the BIM model 18 and adjusts the view angle to optimize the projection alignment. For example, when projecting the layout of a toilet on a wall to identify built-in pipeline positions, the wall serves as the parent plane of the toilet. In various embodiments, by using ifcopenshell, key data are extracted according to the GUIDs of both the target object and its parent plane, thereby obtaining their local placements, and the dimensions of the targets. T o determine the toilet’s position relative to the wall, a transformation matrix is computed as Equation 1:T=T ^parentT target (1)Docket T19763W0001 (222112-2650)where Ttarget and Tparent represent the local placement matrices of the target object and the parent surface, respectively.

[0040] This matrix transformation enables the representation of the target object's position relative to the parent surface, with the surface origin serving as the origin of the coordinate system. In this way, the laser projector 1 can know where to project on the projection plane. Simultaneously, the 3D BIM model 18 can export a 2D CAD model for any view and object. For example, through AutoCAD LISP (AutoLISP), the object’s vertex coordinates may be extracted from the CAD model, defining the component’s shape in terms of vertex points, type, and size. This quantized shape representation, combined with the relative position from the transformation matrix, provides all necessary data for precise laser projection. In this way, the foundation for guiding the laser projector 1 in the subsequent steps is established.

[0041] Building on the foregoing output, the analysis & decision software module 6 is configured to convert the quantized object shape into a laser file. Accordingly, FIG.4 illustrates this process using a cross pattern as an example. First, in various embodiments, the data generated, including component vertices and types, are stored in a JSON file and reordered to follow the desired path of point movement. This reordering ensures that the laser projector’s path appears as a continuous shape to the human eye. The vertices are rearranged in sequence, following the intended path of the laser. Since laser projection is represented by a series of point movements, point interpolation is used between the component vertices to create smooth transitions. After interpolation, in various embodiments, these points are scaled up to fit the International Laser Display Association (ILDA) coordinate system, which spans from -32768 to 32767. To maintain consistency in the laser’s movement, in various embodiments, a blanking path is designed where the laser turns off to move betweenDocket T19763W0001 (222112-2650)different segments of the pattern without projecting light. This blanking path is represented by the lines (shown for the reordered JSON File) in FIG. 4, showing the laser’s off-path movement. When projected, only the main cross pattern is visible, as the laser remains off during the blanking movements along the lines (shown for the reordered JSON File). Finally, in various embodiments, the ILDA file is generated by converting the organized and scaled JSON data into a laser-compatible format. This structured approach ensures precise pattern projection, with only the intended shape displayed to the viewer.

[0042] In various embodiments, the action execution software module 8 governs both projection alignment and installation acceptance on the selected parent surface. After a pattern is projected, a vision-language-action model of the action execution software module 8 receives the current projection plan (pattern vertices, tolerance bands, projector pose Tproj, synchronized images of the projection, and the surface homography Hplanethat converts pixel displacements to millimeters in the surface’s metric frame "77" = {0, u , v, n . The action execution software module 8 detects projected polylines and alignment marks (narrow-band thresholding when color-coded, or a lightweight contour / line head trained for laser strokes), optionally detects physical references (edges, hole centers), and registers the expected pattern to the observed strokes to compute in-plane offsets. From these, the action execution software module 8 reports maximum / mean error (mm), coverage / contrast quality, and occlusion flags. If within tolerance, the projection is accepted; otherwise, the action execution software module 8 derives a small update A = (Au,Av,A0) in the (u, v) frame, maps A to a refined projector pose, re-projects, and re-checks within bounded iterations (“project-sense-correct”).Docket T19763W0001 (222112-2650)

[0043] Once the worker completes the step, the action execution software module 8 performs post-install inspection to decide whether the part is installed correctly. Using images (optionally depth) of the finished state, the action execution software module 8 segments and recognizes the installed object, estimates its as-built pose Tasbunt relative to "77" , and registers it against the expected geometry from the Projection Library entry (GUID, plane) and BIM placements. In various embodiments, the action execution software module 8 computes: (1) in-plane translation / rotation errors (u,v,9) (2) edge gaps / flushness along specified sections; (3) level / plumb deviations by combining image cues with n; and (4) compliance of fasteners / cutouts (presence, count, spacing, and offsets to marks). All measurements are expressed in mm via Hpianeand compared with features specific bands (e.g., outline ±3 mm, hole centers ±2 mm, fastener spacing ±5 mm). The module then issues a pass / fail decision with an itemized punch-list: failures trigger corrective guidance — either concise textual cues (e.g., “raise right edge 4 mm; add fastener at mark #3”) ora re-projection overlay that highlights out-of-tolerance regions; passes unlock step completion and advance the installation sequence. Only steps that receive a post-install pass are marked finished.

[0044] In various embodiments, artificial intelligence training data comprise image-caption-question-answer-step ID tuples for language grounding and projection / installation images with known offsets for metric supervision, augmented for brightness, glare, dust, and partial occlusion. Runtime outputs — alignment parameters, deviations, pass / fail determinations, corrective updates, and evidence imagery — may be logged and associated with the installation target identifier, plane, and design-data revision.Docket T19763W0001 (222112-2650)

[0045] In various embodiments, the system generates an exportable task record for each installation target and / or step. The exportable task record may include, by way of non-limiting examples, the installation target identifier, the associated design revision, selected guidance pattern identifiers, alignment parameters, measured deviation metrics (e.g., translational offset, rotational offset, fit error, etc.), tolerance thresholds, corrective outputs applied (e.g., updated projection parameters), timestamps, operator / user identifiers, and one or more evidence artifacts (e.g., images, depth snapshots, or summaries). In some embodiments, the exportable task record is transmitted to one or more external platforms for cross-team coordination and documentation, including email, messaging, project management systems, and / or document control systems, thereby providing a cross-platform traceable workflow for owners, designers, general contractors, and subcontractors. In some embodiments, the record is stored in a tamper-evident or append-only log to support auditing and dispute resolution.

[0046] This disclosed laser-based SAR system brings design information vividly into the construction site to provide workers with visual guidance. As shown in FIG. 5, an exemplary workflow comprises seven main steps. In various embodiments, the process begins with Step 1, where design data representing an installation target is received from one or more sources (e.g., BIM, CAD, drawings, manuals, or specifications), the design data including geometric information such as object positions, vertex coordinates, and dimensions. In Step 2, the extracted or derived geometric information is converted into projector-ready guidance data, including a pattern and one or more projection parameters for projecting guidance associated with the installation target. In Step 3, the projector-ready guidance data is stored in a projection library, where each entry is associated with (i) an identifier of the installationDocket T19763W0001 (222112-2650)target (e.g., GUID or other unique reference) and (ii) a revision of the design data, thereby enabling traceable retrieval of guidance corresponding to a particular design version. In Step 4, a user selects the installation target for field execution through a user interface (e.g., voice, text, GUI selection, gesture, button, or scanned code), wherein the selection is associated with the identifier and the revision, and the corresponding projector-ready guidance data and device-specific projector calibration parameters are retrieved from the projection library for an as-built surface corresponding to the installation target. In Step 5, the computing hardware estimates an alignment between the design data and the as-built surface using one or more alignment techniques (e.g., marker-based localization, feature-based registration, and / or depth / point-cloud based alignment), and determines a pose or positioning plan for the field execution device relative to the as-built surface. In Step 6, the movable platform is directed to move into position to achieve the alignment between the projector and the as-built surface, and the projector projects the guidance pattern onto the as-built surface to guide field execution. In subsequent steps (not shown), the system may perform a closed-loop verification by measuring one or more deviations (e.g., a translational offset, a rotational offset, or a fit error relative to a tolerance threshold, etc.), including (i) a deviation between the projected guidance and the as-built surface during alignment and / or (ii) a deviation between an as-built condition of the installation target and an expected geometry or tolerance criterion, generating a corrective output based on the deviation (e.g., by updating at least one projection parameter and re-projecting), and generating an exportable task record including the identifier, the revision, one or more alignment parameters, the deviation, and a timestamp, optionally with associated evidence for documentation and cross-team traceability.Docket T19763W0001 (222112-2650)

[0047] In various embodiments, the projection library 17 may be custom-built for each BIM model 18 and play an important role in guiding workers efficiently, especially in standardized constructions like industrialized buildings. When a worker selects a target object and its projection plane in the control interface, the computing hardware 4 of the laser-based SAR system 10 uses the object’s GUID to retrieve the corresponding projection file and parameters from the projection library database 17. With this data, the laser projector 1 can precisely project the object outline onto the as-built location, ensuring accurate placement and orientation.

[0048] In various embodiments, the organization of the projection library 17 also simplifies the work sequence by storing the installation order, allowing the projector 1 to follow a logical guidance sequence. Referring back to FIG. 2A, this figure demonstrates an example of this database for a house project, where each row lists the target object type, ID, and corresponding laser files for different projection planes (e.g., left, top, front), along with the installation order. This structured approach streamlines projection retrieval and setup, providing clear and reliable guidance for workers on site.

[0049] In various embodiments, alignment between the design data and the as-built surface is estimated using one or more alignment techniques, including markerbased localization, feature-based registration using edges / corners / lines, plane detection and surface fitting, depth-based or point-cloud registration (e.g., ICR or variants), and / or fusion with inertial and odometry measurements. In some embodiments, alignment is performed without pre-installed markers by registering projected strokes and / or sensed geometric features of the as-built surface to expected design geometry. The selection of alignment technique may depend on the worksite conditions, sensor availability, and required tolerance, and multiple techniques mayDocket T19763W0001 (222112-2650)be combined to improve robustness. As discussed, to bridge the coordinate systems of the projection plane (derived from the digital BIM model) and the real-world environment, in various embodiments, a marker system may be pre-installed on-site as a reference point for alignment. In various embodiments, an exemplary laser projector system 10 includes a camera 20 (FIG. 1), which captures images of the marker to determine the projector’s pose relative to it. This marker’s position is preregistered in the local world coordinate system, allowing the camera 20 to estimate its precise location. Additionally, to verify if the camera 20 is parallel to the projection plane (reducing distortion from skewed angles), a checkerboard pattern may be affixed to the projection plane. In various embodiments, using OpenCV’s built-in camera calibration functions, the computing hardware 4 detects the checkerboard pattern and computes the camera’s extrinsic parameters. The checkerboard pattern provides well-defined reference points, enabling the computing hardware 4 to determine both the position and orientation of the camera relative to the projection plane. If the camera 20 is misaligned, adjustments can be made to ensure precise alignment. This calibration step ensures that the laser projector’s projection remains undistorted.

[0050] For projector pose estimation, the computing hardware 4 may estimate the camera's pose by detecting the marker's 2D image coordinates (w,n) and their corresponding 3D world coordinates (Xt,Yi,Zi). Solving the Perspective-n-Point problem provides the camera's rotation matrix Rcand translation vector tc, which defines its pose in the world coordinate system (see Equation 2):Tcam" 1](2)where Rcis the camera’s orientation in world coordinate system, and tcis the camera’s position in the world coordinate system.Docket T19763W0001 (222112-2650)

[0051] The camera’s position is then translated into the position of the laser projector 1. In various embodiments, both the camera 20 and projector 1 are housed together, e.g., in a 3D-printed enclosure, meaning their relative position is fixed and can be expressed as a known offset transformation. This mechanical offset is given as Equation 3:Toffset = [ / 30x3 toqset] (3)where / 3x3 is the identity matrix (since no additional rotation exists between the camera 20 and projector 1 ) and tOffset=[ x, y, z]Tis the known translation vector representing the projector’s position relative to the camera 20 in the enclosure’s local coordinate system.

[0052] The projector’s position in the world coordinate system is then computed as Equation 4:tproj ~ tc Reto ff set (4)where tproj - [xP,yp,zP]Tis the projector’s position in world coordinate system. While tcis the camera’s position in world coordinate system, Rcis the camera’s rotation matrix in world coordinate system, and toff set is the known mechanical displacement of the projector relative to the camera 20.

[0053] Once the laser file and projection parameters are determined, the laser projector 1 is configured to project the pattern at the designated location. In various embodiments, an exemplary laser projector 1 includes the following components, as shown in FIG. 6 and can be divided into two main subsystems: a power module 30 and a laser control module 40, as follows. In a non-limiting embodiment, the power module 30 is a laser galvanometer power supply that provides power to both a driver control board and a Transistor-Transistor Logic (TTL) control module of the laser control module 40. Since specific components require different voltage levels, inDocket T19763W0001 (222112-2650)various embodiments, a buck module of the power module 30 reduces the voltage from 12V to 5V for safe operation and ensures a stable power supply for the rest of the laser projector 1. In various embodiments, the laser control module 40 contains multiple components, including the driver control board, TTL control module, and galvanometer drive board. The driver control board manages both the TTL control module and the galvanometer drive board. In various embodiments, the TTL control module regulates laser emission from the RGB laser dot module and controls a vibration mirror for each color, enabling precise color generation. The galvanometer drive board controls the movement of the laser galvanometer, guiding the laser beam along the predefined path stored in the laser file. The reflecting prism handles the combination and separation of the three laser channels (red, green, and blue), ensuring colorful, high-quality projections.

[0054] In various embodiments, the laser galvanometer of the laser control module 40 is a high-speed scanning device that deflects the laser beam, creating the projected patterns. The laser galvanometer plays a key role in controlling parameters such as scale and position, ensuring accurate beam positioning and movement; the RGB laser dot module generates the primary colors (Red, Green, and Blue) used in the projection. By combining and modulating these colors, the laser control module 40 can produce a wide range of colors and intensities for the projected pattern; the reflecting prism combines and directs the red, green, and blue laser beams to produce a unified, full-color output for accurate and vibrant projections; the galvanometer drive board powers and controls the galvanometer’s movement, precisely positioning the laser beam along the defined path; the TTL control module utilizes Transistor-Transistor Logic (TTL) to control the on / off states of the laser, allowing modulation of laser intensity and enabling various projection effects; the laser galvanometer power supplyDocket T19763W0001 (222112-2650)supplies power to the galvanometer and laser modules to ensure consistent and accurate operation; the buck module steps down voltage from 12V to 5V, providing the appropriate voltage levels needed for certain components; and / or the driver control board serves as the central control unit, decoding ILDA files and controlling the projection of patterns.

[0055] In various embodiments, the laser galvanometer of the laser control module 40 operates at 20 kHz aiming to ensure high-speed and precise laser beam deflection, which minimizes flicker and distortion during projection. Additionally, the galvanometer driver board utilizes an initial hardware adjustment for x-scale and y-scale calibration. This adjustment follows a predefined principle: when projecting a test pattern, the projection length should be 60 inches when the projection distance is 82 inches. This manual calibration step ensures proper scaling before software-based corrections are applied.

[0056] In various embodiments, the laser projector 1 is integrated with a selfcalibration mechanism within the laser projector 1 to ensure precise projection. The projector 1 may also utilize an optical angle method for displaying objects and figures, which may lead to distortion due to uneven scaling at the outer edges. In accordance with various embodiments, the ILDA file format stores laser objects as points in 3D space, but these points require adjustment when projected onto a flat 2D surface.

[0057] As illustrated in FIG. 7, the distortion due to uneven scaling occurs because the laser beam is evenly distributed but gets truncated at the edges of the projection area. This effect results from a combination of beam speed and the angle of the galvanometer head, causing the display to deviate from the intended output, especially near the projector’s maximum display angle.Docket T19763W0001 (222112-2650)

[0058] To counter this distortion and ensure that the projected output matches the expected result, an experimentally determined angle scale function is applied, in various embodiments. This function adjusts the laser points based on Equation 5: arc tan fp™t‘aiXLjx—Pfinai = 2048 + -x 1024 (5)where Pfinaimeans calibrated laser point coordinates, Ptnuiai is the current laser point coordinates, L is the projection range when the projection angle achieves maximum, and D is the distance between laser center and projection plane.

[0059] Assuming the laser is perpendicular to a flat wall, this equation utilizes the Pythagorean Theorem to calculate the calibrated coordinates according to angular proportion. The final coordinate Pfinai is sent to the drive board, with each angular adjustment corresponding to approximately 0.02 degrees at a projection distance of 5 m, which translates to a positional adjustment of around 1.7 mm. This calibration process minimizes distortion, ensuring that the projected pattern maintains its intended shape and proportions.

[0060] For system evaluation purposes, several aspects are evaluated to validate the performance of the disclosed system. To begin, FIG. 8 illustrates the operational workflow of an exemplary laser-based SAR system 10. The evaluation process begins with preparing the worksite by establishing reference points to ensure accurate alignment between the physical environment and design models. This involves placing markers based on factors such as visibility, accessibility, and structural stability. Typically, 4 reference points are required to calculate the transformation matrix for a calibrated camera. In real practice, workers can attach markers — such as Arllco markers or checkerboard with at least 4 reference points — using methods suited to the surface material. For smooth surfaces like glass, metal, or painted walls, adhesive backing provides a secure and non-invasive attachment. On rough or porous surfaces,Docket T19763W0001 (222112-2650)such as concrete or unfinished wood, mechanical fasteners like screws ensure stability despite surface irregularities. The size and placement of markers should be adjusted to the specific workspace to optimize localization accuracy. To streamline the setup process, pre-gridded templates or laser alignment tools can help workers quickly identify marker positions. Pre-prepared markers with adhesive backing or pre-drilled holes further reduce installation time. Based on controlled experiments, two workers can install five 4x4 ArUco markers (spaced at 10” intervals from 29” to 69”) over an approximately 3 m2area within 2.25-3.3 minutes. In more complex environments, establishing reference points may require additional time. However, this setup is a one-time process for most projects and is justified by the improvements in alignment accuracy and reduction in rework during projection.

[0061] The placement of the laser projector 1 is obtained by measuring relative pose to the established reference points. And then workers activate the laser-based SAR system 10 and initialize it for operation. They select the components to be installed and specify the installation plane, which determines the projection targets and the corresponding projection surface. Following this, the computing hardware 4 of the laser-based SAR system 10 undergoes calibration by projecting a standard test pattern to verify functionality and establish the correct scaling. Once calibration is complete, workers can seamlessly control the projection process through the laser projection control interface module 16, toggling it on or off as needed to provide precise visual guidance during installation tasks.

[0062] To evaluate the system setup time, three groups of two participants performed the workflow three times each, and the setup time was recorded for each trial. In one experiment, the markers were placed at (0”, 40”), and the projector 1 was positioned at a fixed projection distance of 82.4”. The designated test pattern lengthDocket T19763W0001 (222112-2650)was set to 30” for consistency and convenience. Each trial began with the workers setting up the markers and continued through to determining the scaling factor. The results indicate that the system setup time averages approximately 2 minutes, which is acceptable given the trade-off between initial calibration effort and the improved precision and efficiency provided by the system.

[0063] To rigorously validate the system's capability to convert design models into fully functional laser projection files, five different patterns were chosen from real architectural drawings. These patterns encompass various geometric elements — such as lines, arcs, circles, and semicircles — derived from top-view plans of a washing basin, a window, a beach chair, a single sofa, and a restroom layout. The conversion process accurately transformed each design model component into corresponding laser projection files without compromising detail. As illustrated in FIG.9, the system 10 preserved the original shapes from CAD drawings while vividly translating design models into laser-guided projections, thereby demonstrating its effectiveness for real-world installation guidance.

[0064] Two types of validations were conducted to assess the projection performance of the disclosed laser projector: shape stability and localization accuracy. These evaluations involved projecting basic patterns at varying distances from the projection plane to analyze how well the system maintained projection integrity and assess the shape stability and localization accuracy of the projection system. The experiments were conducted within a 3 m x 3.5 m workspace. In this setup, the projector’s position was pre-determined and fixed parallel to the wall due to the stationary base of the projector. A BOSCH Professional GLL 30 laser level was used to assist in constructing the coordinate system, ensuring precise alignment for accurate performance evaluation.Docket T19763W0001 (222112-2650)

[0065] To evaluate the system’s projection accuracy, five different patterns were projected onto a wall at distances ranging from 1 m to 3 m, in 0.5 m increments. At each distance, the position of both the projector 1 and the projected patterns was recorded using a Tacklife Hd-60 classic laser measurer and Keson NR30M nylon coated steel blade measuring tape. Each pattern's position was measured five times by three different participants (all with basic engineering backgrounds) to ensure data reliability. A different starting point on the tape measurer was used for each trial to reduce errors arising from tape stretching. The primary objective was to compare the observed pattern positions with their expected positions, aligned with the projector’s center, to assess the system’s stability and precision at various distances. FIG. 10 shows the five test laser patterns, with each pattern’s center point recorded during the measurement process.

[0066] The localization accuracy of the projection system is evaluated using three key metrics: localization error, consistency across trials, and overall stability across distances. These metrics assess how well the projected patterns align with the projector's center and maintain accuracy under varying conditions. The localization error measures the deviation between the observed position of the pattern's center (x observ d, y observe d) and the projector's center (xprojectory projector). The error is calculated using the Euclidean distance (Equation 6):Error = K ■^-observed ^-projector) T (^observed V projector) (®)

[0067] This metric provides a direct measure of how accurately the patterns are projected relative to the projector's alignment. To ensure reliability, data is recorded 15 times for each pattern and distance. The mean error is computed as the average error across the 15 trials (Equation 7):Docket T19763W0001 (222112-2650)Error = ■^'Zt=iErrori(7)and the standard deviation (std) of error is calculated to capture the variation in these measurements (Equation 8):°Error = J^i=1(Errori ~ Error)2. (8)

[0068] These metrics provide insights into whether the projection accuracy is consistent as the projector's distance changes.

[0069] TABLE I (below) presents the mean center-location errors and associated standard deviations for the five projected patterns at distances of 1.0 m, 1.5 m, 2.0 m, 2.5 m, and 3.0 m. In general, the average position error increases from roughly 0.05 cm at 1 .0 m to around 0.17 cm at 3.0 m, consistent with the geometric principle that small angular deviations in the mirrors produce larger linear displacements on the wall as distance grows. Despite this increase, the errors remain at or below a couple of millimeters, indicating that the system’s stable positional accuracy.Mean position error for each projected pattern at different distances....Pattern J Pattern 2 [ Pattern 3 [ Pattern 4 Overall Distance Mean Position Error (cm) Avg | Std 1.0 in 0.0133 0.0267 0.0745 0.1067 0.0481 0.0539 i 0.0336 1.5 m 0.0550 0.11.33 0.1166 0.1373 0.1043 0.1053 i 0.0274 2.0 m "01320 "671253 '672004 "61054 "61323 ”6T39T"T16322"' 2.5 m 0.0629 0.1406 0.1480 0.1350 0.0907 0.1154 i 0.0330 3.0 m 00835 0.2885 0.2496 0.0825 0.1649 0.1738 i 00842 Avs 0.0693 0.1389 0.1578 0.1134 0.1081"ltd ' ”6.6388 ”676'846 ”16616 ”676205 ”676393TABLE I

[0070] Among the individual patterns, Pattern 1 (a diagonal cross) shows the lowest overall mean error (about 0.07 cm) and moderate variability. Pattern 3, which features both a cross and a circle, records the highest mean error (approximately 0.16 cm). Importantly, all center measurements — across all patterns — are taken at the intersection of their respective crosses, so the circle in Pattern 3 does not itself define the center point. However, the additional circle segment alters the scanning trajectoryDocket T19763W0001 (222112-2650)that the galvanometer mirrors must follow. A circle requires continuously changing angles in both the horizontal and vertical axes, which can exacerbate minor controlloop or mechanical nonlinearities. Even though the operator measures only the cross intersection for the center location, any scanning irregularities introduced by tracing the circle can slightly shift or distort the cross lines themselves, causing the measured cross center to deviate more compared to patterns without circular elements.

[0071] Patterns 2 and 4 help illustrate the influence of a shape’s geometry and repeated corner points on projection accuracy. Pattern 2 contains only a simple cross, whereas Pattern 4 builds upon that cross to form a 2 x 2 grid. Interestingly, Pattern 4 shows lower variability (standard deviation of -0.02 cm) than Pattern 2 (-0.08 cm), which suggests that the additional rectangular outlines in Pattern 4 may help stabilize the scanning motion by repeatedly forcing the mirrors back to precise angular positions at the grid’s corners. In Pattern 2, the scanning path is constrained to the horizontal and vertical lines, leaving fewer distinct angular checkpoints; any small drift or asymmetry in mirror response can accumulate over multiple scans, contributing to greater variation in the measured center.

[0072] Another factor influencing the results is the inherent beam shape and divergence. Although the cross intersection is used to determine the location error, the peripheral portions of the lines or the circle may appear dimmer or slightly thicker at longer distances, making it subtly harder for participants to pinpoint the exact line intersection. Moreover, participants used a tape measure and a laser distance meter in multiple trials, and, despite efforts to mitigate human error (by changing the tape measure’s starting point each time, for example), there could still be slight inconsistencies in how the center point is identified. These effects, combined with theDocket T19763W0001 (222112-2650)galvanometer’s mechanical tolerances, create the variability observed at certain distances and in certain patterns.

[0073] Overall, the measurement data confirm that the projector’s position accuracy remains within a millimeter or two over distances up to 3 m, with only moderate fluctuations stemming from differences in scanning geometry, human measurement procedures, and distance-related magnification of small angular errors. The consistently higher error for Pattern 3 — despite the center point always being the cross intersection — indicates that additional or more complex geometry (a circle) can introduce small scanning irregularities. Meanwhile, patterns with more frequent corner re-alignments (e.g., the 2 x 2 grid in Pattern 4) may exhibit tighter repeatability even when they include the same center cross. These findings suggest that designers of laser-projected alignment systems should carefully consider both the underlying mirror dynamics and the chosen pattern geometry when seeking high accuracy, especially at greater throw distances.

[0074] To evaluate the shape stability of the projection system, four geometric patterns were tested: (1) a square with middle lines, (2) a square with diagonal lines, (3) an isosceles right-angle triangle, and (4) a circle. These patterns were projected onto a flat surface, and the lengths of specific edges (highlighted and numbered as dashed lines in FIG. 11 were measured using a Keson NR30M nylon coated steel blade measuring tape. To ensure reliability, each edge was measured five times by three different individuals, and the projector was tested at distances ranging from 1.0 m to 3.0 m with 0.5 m increments.

[0075] For the square with middle lines (Pattern 4), the ideal length of each edge is 65.1 cm. For the square with diagonal lines (Pattern 5), the edge length remains 65.1 cm, while the diagonal line should measure approximately 92 cm. For theDocket T19763W0001 (222112-2650)isosceles right-angle triangle (Pattern 6), the base is 65.1 cm, the height is half the base (32.5 cm), and the hypotenuse is 46 cm. Finally, for the circle (Pattern 7), the ideal diameter is 65.1 cm. This setup enables an assessment of whether the projector 1 maintains consistent and accurate geometric dimensions across varying distances, validating its shape stability.

[0076] To quantify the shape stability of the projected patterns, the following metrics were used (Equation 9-11). The relative deviation of the measured edge length from the ideal (expected) length:Relative Deviation = ^^ed-Lideai\x 100o / o (g)Lldealwhere Lmeasured is the average of the three recorded measurements for an edge, and Lideai is the theoretical length.

[0077] For patterns with multiple edges (e.g., square, triangle), the aspect ratio deviation (AR) was calculated to assess geometric distortion:Aspect Ratio =Lon9est Ed9e. (10)rShortest Edge

[0078] The average relative deviation (Ave Dev) across all measured edges of a pattern:Deviation = -x100% (11)N Lideaiwhere N is the number of edges measured for the pattern.

[0079] TABLE II (below) summarizes the AR and Ave Dev for the four projected patterns across distances of 1.0 m, 1.5 m, 2.0 m, 2.5 m, and 3.0 m. In each case, the physical dimensions of the patterns (e.g., a ~65 cm side) remain the same, forcing the galvanometer mirrors to operate at different scanning angles depending on the distance. These changing angles can introduce minor geometric distortions that become evident in both AR and Ave Dev metrics.Docket T19763W0001 (222112-2650)Shape stability evaluation for each projected pattern at different distances. Patern 4 Patern 5 Patern 6 Pattern 7 Ave iiror Ave Ave Ave Ave .Ave Distance AR AR AR AR AR Dev Dev Dev Dev Dev 1.0 m 1.009 0.003 1.427 0.004 1.958 0.013 1.000 0.000 0.016 0.005 1.5 m 1.005 0.003 1.420 0.002 1.975 0.008 1.000 0.000 0.009 0.003 2.0 m 1.003 0.001 1.420 0.003 1.992 0.002 1.001 0.001 0.005 0.001 2.5 m 1.005 0.007 1.421 0.002 2.010 0.002 1.001 0.003 0.006 0.003 3.0 m 1.005 0.0 ...1..5..... 1.428 .. 0...0 _1.7..... 1.980 0.002 1.002 .. 0....01.3..... 0010 0 0147’415“ “”2’0'"” "”" o ’“To“’“Ave Error 0.006 0.006 0.009 0.006 0.021 0.007 0.001 0.003TABLE II

[0080] An overall trend shows that the laser-based SAR system 10 performs most consistently near the nominal calibration range of 2.0-2.5 m: the AR values hover close to each shape’s ground-truth ratio, and Ave Dev tends to be at its minimum. By contrast, the data reveal some fluctuations at shorter (1.0-1.5 m) and longer (3.0 m) distances. In particular, Pattern 6 (the isosceles right-angle triangle) does not always produce the single largest AR error among the patterns, but its AR values do appear to fluctuate more than those of the squares or circle. Since triangles have fewer edges to constrain the geometry, a small misalignment at one corner can produce a relatively large change in the base-to-height ratio, causing AR to vary more between distance increments. Nevertheless, this variability is not strictly monotonic, as the AR error at 1.0 m can exceed that at 3.0 m, illustrating that near-field scanning angles (which are larger) do not necessarily correlate to a smooth progression of errors at greater distances.

[0081] A related observation concerns the differences in average performance across all patterns at 1.0 m versus 3.0 m. Although the average AR errorforall shapes combined may be higher at 1.0 m than at 3.0 m, the Ave Dev can show the opposite behavior, ending up larger at 3.0 m. This discrepancy occurs because AR and Ave Dev capture different aspects of shape fidelity. Some shapes might have fairly small edge-length errors yet yield a noticeable AR offset (for example, if two adjacent edgesDocket T19763W0001 (222112-2650)deviate slightly but in opposite directions), whereas others might maintain near-ideal ratios but accumulate corner or diagonal errors that inflate the total Ave Dev. At longer distances, a small galvanometer offset can translate into a bigger linear shift, thereby raising certain edges’ measured error and pushing up the Ave Dev metric. Conversely, at short range, overshoot in the mirror angles might distort the shape’s proportions, increasing AR error but not necessarily the net sum of edge-length deviations.

[0082] Despite these nuanced differences, the projector 1 maintains an acceptable level of geometric consistency for squares, triangles, and circles from 1 .0 m to 3.0 m, with the best fidelity typically near its calibrated distance of 2.0-2.5 m. Shapes with sharp corners and high aspect ratios, like the isosceles right-angle triangle, tend to be more sensitive to corner overshoot or slight mirror misalignments at the extremes of the scanning range. Although the overall average deviation at 3.0 m is larger than at 1.0 m for all shapes combined, the absolute magnitudes are still modest, generally within a few percentage points for AR and often below 1-2% for Ave Dev. These results illustrate both the strengths of the disclosed laser-based SAR system 10 in holding a target size across varying distances and the importance of considering a shape’s complexity and the system’s distance-dependent calibration nuances when assessing final accuracy.

[0083] Following the controlled performance tests, two case studies were conducted to illustrate how the disclosed laser-based SAR system 10 functions in practical applications. The first study involved guiding the installation of a small guard house using a BIM model, where three distinct objects — a window, an HVAC unit, and a countertop — were projected and assessed. The second study concerned the assembly of two metal frames, allowing further examination of the system’s accuracyDocket T19763W0001 (222112-2650)when installing linear components. Both case studies evaluated different subsets of the system’s functionality under varying object sizes, shapes, and distances.

[0084] In the first case study, a guard booth was chosen as the reference BIM model, and three embedded objects — one window, one HVAC unit, and one countertop — were selected for a detailed evaluation (see FIG. 12). The BIM model provided center coordinates of (114 cm, 149.9 cm) for the window, center coordinates of (61 cm, 48.3 cm) for the HVAC, and center coordinates of (61 cm, 101.9 cm) for the countertop, with dimensions of 92.4 x 103.8 cm (window), 58.4 x 40 cm (HVAC), and 121.3 x 4.4 cm (countertop). Although laser line brightness appeared somewhat diminished in camera images — taken at a 1 / 15 frame rate with a CANON EOS M6 Mark II — visual observation confirmed sufficient clarity for installation guidance.

[0085] To quantify the system’s accuracy in reproducing BIM-defined positions and shapes, three observers independently measured each projected object’s corner coordinates, performing five repeated measurements per observer. The projected center location was computed by averaging these corner coordinates, and its distance from the BIM-model center defined the location error. Each individual edge was also measured to derive an Ave Dev, indicating how closely the projected perimeter adhered to its intended shape. The window provided 17 measurable edges, while the HVAC and countertop had 8 and 4 edges, respectively. TABLE III (below) provides the results in terms of center offsets, total location error in centimeters, and average deviations as percentages. Because these percentages can seem small in absolute numeric terms, the “Location Error (%)” is also shown with respect to each object’s shorter side, offering a conservative measure of how the projection might shift relative to the most critical dimension.Docket T19763W0001 (222112-2650)Experiment results of the BIM object projection.Dimension. Projector Ay | Location Location Ave Dev Object (cm) Distance (m) (cm) (cm) Error (cm) Error (%) (%) Window 92.4 X 103.8 2.309 0.47 0.38 0.60 0.65% 1.3% HVAC 58.4 x 40 2.170 0.27 0.05 0.28 0.48% 0.4%Countertop 121.3 X 4.4 2.125 0.84 0.14 j 0.85 1.93% 0.8%TABLE III

[0086] In absolute terms, the largest center error (0.85 cm for the countertop) remains within about 0.33 inches, which is comparable to or below typical construction tolerances for window or door openings of ±1 / 4 inch or ±1 / 8 inch for many framing and placement tasks. Specifically, industry guidelines often permit a horizontal or layout deviation of upto±1 / 4 inch (6.35 mm) when installing windows, doors, or other building components, with slightly tighter constraints (±1 / 8 inch) cited for certain precision alignments. Here, the largest observed offset still falls near or below ±1 / 4 inch, while the smaller objects (window and HVAC) remain well under 1% in relative offset — reliably meeting typical light-frame or interior-finish positioning requirements.

[0087] Similarly, the average edge deviations for all three objects remain below 2%, indicating that the system reproduces the overall shapes with acceptable fidelity for most jobsite needs. The elongated countertop shows a higher relative center offset (1.93%) because its narrow width amplifies any corner misalignment, yet even there, the actual difference is limited to under 1 cm. Since many interior or light-frame construction tasks allow for slight discrepancies compensated by trim or flexible sealants, a minor shift of less than 10 mm typically does not present major installation issues.

[0088] Nonetheless, if higher precision is required — such as for specialized commercial glazing or custom millwork where tolerances might be as low as ±1 mm to ±2 mm — operators could reduce remaining deviations by refining projector calibration at the specific distance used and ensuring minimal tilt or rotationalDocket T19763W0001 (222112-2650)misalignment. More frequent reference checks or repeated corner measurements can also help mitigate potential user reading errors. Overall, even under standard calibration, the laser-based SAR system 10 demonstrates that sub-centimeter placement is readily achievable, aligning with established tolerances in the construction industry and supporting efficient on-site installation.

[0089] The second case study evaluated the system’s capacity to guide the assembly of two metal frames using the same laser projector. Each frame contained three bars that had to be installed at specified reference positions indicated by projected lines or points. Case 1 used bars of 49 cm in length, whereas Case 2 employed bars of 23.6 cm. The projector was positioned at a height of 0.874 m, allowing a moderately angled beam approach to the assembly surface.

[0090] For both cases, the intended bar positions were recorded, and after using laser guidance to place each bar, the actual final positions were measured with a steel tape measure. TABLE IV (below) details the deviations, both in absolute millimeters and as a percentage of each bar’s length. This approach verifies whether laser-projected references can reliably lead to the correct placement of linear components under typical workshop or onsite conditions.Metal flame assembly guidance resultsBar Length Reference Position Measured Position Deviation Deviation Case(cm) (mm) (mm) (mm) (%) Case 1 (Bar 1) 49.0 115.0 113.6 1.40 0.29% Case 1 (Bar 2) 49.0 60.83 56.20 4.63 0.95% Case 2 ( Bar 1) 23.6 46.71 45.80 0.91 0.39%Case 2 (Bar 2) 23.6 39.73 40.30 0.57 0.24%TABLE IV

[0091] In Case 1, the longer bars manifest higher absolute deviations (up to 4.63 mm), equivalent to about 0.95% of the 49 cm length. This is partly because a longer span makes minor angular or alignment errors more noticeable. Although the second bar exhibits a larger offset than the first, both remain below 1% of the bar length,Docket T19763W0001 (222112-2650)indicating that even with longer components, the system’s laser references keep installations within a relatively tight tolerance. Case 2, by contrast, shows absolute deviations under 1 mm and corresponding deviations under 0.5%. These smaller figures reflect the simpler task of aligning shorter bars, which naturally reduces the impact of any slight misplacement or measurement discrepancy.

[0092] Overall, these two case studies — installing BIM-defined objects in a guard booth and assembling metal frames — demonstrate that the disclosed laser-based SAR system 10 provides reliable positional and dimensional guidance (e.g., outline, annotations, step cues, tolerances, etc.) for different object types, sizes, and shapes. The BIM object projection results confirm sub-centimeter accuracy for both location and shape, even for elongated components like a narrow countertop. The assembly guidance findings reveal that linear elements can be placed within 1 % of their intended positions, whether they are nearly half a meter long (49 cm) or under a quarter meter (23.6 cm). These deviations are typically acceptable in routine construction or fabrication tasks, where a few millimeters can be managed through ordinary installation adjustments. In scenarios requiring greater precision, users could fine-tune the system’s calibration at the relevant distance, minimize tilt, or use additional alignment checks to ensure the projected lines or points match critical tolerances.

[0093] The present disclosure successfully demonstrate a laser-based SAR system empowered by robotics and artificial intelligence (Al) technologies to deliver in-situ, real-time guidance to construction workers through laser-ink-based projection, process monitoring, inspection, and automated documentation. Embodiments of this system integrate multiple advanced components to enhance productivity, accuracy, and safety on construction sites using programmable spatial augmented reality projection to project construction details and work plans directly onto physical surfacesDocket T19763W0001 (222112-2650)using laser lines. In various embodiment, these projections are dynamically generated from the Building Information Model (BIM) data, providing intuitive and precise guidance to workers in real time. In various embodiments, an integrated visionlanguage-action model enables the system to interpret, retrieve, and contextualize information from a wide range of construction documents — including drawings, schedules, and specifications — allowing both workers and superintendents to receive instant support and actionable insights in the field. By combining laser projection, BIM data, and Al processing, the laser-based SAR system 10 can conduct real-time inspections of ongoing construction activities. For example, an exemplary system 10 may detect deviations, verify task completion, and ensure alignment with design intent, reducing errors and rework. The system 10 may also continuously gather and synthesize distributed data from the job site — via sensors, cameras, and user inputs — and integrate this information into the BIM environment. This two-way feedback project-sense-correct loop can maintain a living digital twin of the construction project, enabling better coordination, traceability, and decision-making.

[0094] Systems and methods of the present disclosure were validated in controlled scenarios, confirming its ability to map digital designs onto physical construction sites. Further, the present disclosure shows successful development of a laser-based SAR system for on-site guidance (e.g., outline, annotations, step cues, tolerances, etc.). Experimental evaluations and case studies confirmed that a prototype of the system achieves sufficient brightness and accuracy for typical construction tasks, proving its viability as a reference design for broader industry adoption. As such, the present disclosure delivers a practical laser-based SAR framework that addresses persisting layout and alignment challenges. At the same time, it underscores that substantial opportunities remain for improving projector hardware, enhancing calibration, andDocket T19763W0001 (222112-2650)broadening deployment scenarios through mobile or multi-projector solutions. Overall, the present disclosure provides a foundation for leveraging open-source SAR in construction, guiding future investigations into even more efficient and scalable ways to bridge the gap between digital design intent and on-site reality.

[0095] Through controlled experiments, the disclosed system and related methods consistently demonstrated sub-centimeter projection accuracy and stable shape fidelity over variable distances. The evaluation revealed that even complex or elongated objects could be projected with minimal distortions, substantially reducing manual measurement and layout tasks. These results align with common industry tolerances for interior framing and assembly, thus illustrating the practical feasibility of the approach.

[0096] FIG. 13 depicts a schematic block diagram of a computing device 1300 that can be used to implement various embodiments of the present disclosure, such as, but not limited to, computing hardware 4. An exemplary computing device 1300 includes at least one processor circuit, for example, having a processor (CPU) 1302 and a memory 1304, both of which are coupled to a local interface 1306, and one or more input and output (I / O) devices 1308. In various embodiments, additional hardware devices may also be included, such as graphical processing unit (GPU) 1320, etc. The local interface 1306 may comprise, for example, a data bus with an accompanying address / control bus or other bus structure as can be appreciated.

[0097] Stored in the memory 1304 are both data and several components that are executable by the processor 1302. In particular, stored in the memory 1304 and executable by the processor 1302 of computing device 1300 and / or across multiple computing devices may include (or may not include) machine learning functionality, such as in the form of artificial intelligence appl ication(s) or machine learning model(s)Docket T19763W0001 (222112-2650)1310 and training code 1312, in accordance with embodiments of the present disclosure. Various software modules 1316 may also be stored that is executed by the computing device 1300, such as perception software module 5, analysis & decision software module 6, action execution software module 8, etc. Also stored in the memory 1304 may a data library or store 1314, and / or other data, such as BIM models, BIM metadata, tutorials, etc. In addition, an operating system may be stored in the memory 1304 and executable by the processor 1302. The I / O devices 1308 may include input devices, for example but not limited to, a keyboard, mouse, sensors, cameras or sensors for vision systems, etc. Furthermore, the I / O devices 1308 may also include output devices, for example but not limited to, a display, etc. Also, the I / O devices 1308 may include a communication interface component, such as a network adapter or interface (e.g., WiFi network adapter, Bluetooth adapter, 4G wireless adapter, ethernet adapter, etc.), that allows for wired or wireless communications with external devices and networks.

[0098] Certain embodiments of the present disclosure can be implemented in hardware, software, firmware, or a combination thereof. If implemented in software, logic or functionality can be implemented in software or firmware that is stored in a computer-readable medium (e.g., memory) and that is executed by a suitable instruction execution system. If implemented in hardware, logic or functionality can be implemented with any or a combination of the following technologies, which are all well known in the art: a discrete logic circuit(s) having logic gates for implementing logic functions upon data signals, an application specific integrated circuit (ASIC) having appropriate combinational logic gates, a programmable gate array(s) (PGA), a field programmable gate array (FPGA), etc.Docket T19763W0001 (222112-2650)

[0099] In the context of this document, a "computer-readable medium" can be any means that can contain, store, communicate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can be, for example but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device. More specific examples (a nonexhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic) having one or more wires, a portable computer diskette or drive (magnetic), a random access memory (RAM) (electronic), a read-only memory (ROM) (electronic), an erasable programmable read-only memory (EPROM or Flash memory) (electronic), an optical fiber (optical), and a portable compact disc read-only memory (CDROM) (optical).

[0100] In various embodiments, artificial intelligence and machine learning algorithms are deployed to build models that based on a large set of training data. These models may identify patterns and features within the training data that do not have meaning to human software engineers, but that can be used to accurately classify entities, organize data, optimize solutions, and make predictions or decisions.

[0101] Machine learning refers to the use of artificial intelligence (Al) computer algorithms to build predictive models that can learn and improve through experience. The prediction can be, for example, a classification of an image (e.g. identifying objects in images) or as another example, a recommendation (e.g. a decision). Supervised machine learning algorithms can use sets of labeled data to build models that make predictions for unlabeled input data (e.g., regression analysis, predicting output values from input values or predicting classifications for new input data). Unsupervised machine learning algorithms can use unlabeled data to build models that identifyDocket T19763W0001 (222112-2650)structure, patterns, and relationships among the unlabeled data (e.g., clustering or filtering of input data).

[0102] In various embodiments, a supervised machine learning algorithm can be used to build a model based on a training sample selected from among the data records stored in a data storage and their corresponding output values. The building (e.g., training) of the model can involve an iterative process of updating the model in order to minimize a loss function that quantifies the difference between the model's prediction and the target output values. As such, the machine learning algorithm “learns” how to make better predictions over successive iterations. Various different machine learning techniques, having different model structures and training methods, can be used to build the model. For example, the model can be built using linear regression, nearest neighbor, gradient boosting, or neural network algorithms. Once the model is built, it can be validated using the records stored in the data storage.

[0103] The model can be built according to various predetermined model training settings that control various parameters of the machine learning process. The model training settings can include settings for selecting and shuffling the training data (e.g., different sampling methods), parameters for modifying the data (e.g., normalizing or weighting certain aspects of the data), settings to indicate which type of machine learning algorithm will be used to train the model, a parameterto set a maximum model size (e.g. , in bytes), a parameter to limit the number of iterations or passes performed by the machine learning algorithm, and parameters to set initial weights or variables used by the particular machine learning algorithm. The predetermined model training settings may be determined through experimentation or research.

[0104] After building and validating the model, it can be used to make predictions on new data where the target output is unknown. For instance, computing hardware 4Docket T19763W0001 (222112-2650)storing the model can receive a projector ready pattern and it can make a prediction of the correct physical location and scale on a projection surface. The computing hardware 4 can then project based on the predicted physical location and scale. To do this, the computing hardware 4 can input a set of data based on the record into the model, which determines a predicted output value. The request can also be stored in the data storage such that it could potentially be used for training later builds of the model.

[0105] While the model can determine a predicted output value, the predicted output value may not be particularly useful in of itself. Accordingly, in addition to running the input data through the model, the computing hardware 4 can perform decision making by applying the predicted output value to a set of decision rules. In order to update the model over time, more training data may be accumulated to provide new accurate model builds.

[0106] As will be apparent to those of skill in the art upon reading this disclosure, each of the individual embodiments described and illustrated herein has discrete components and features which may be readily separated from or combined with the features of any of the other several embodiments without departing from the scope or spirit of the present disclosure. Any recited method can be carried out in the order of events recited or in any other order that is logically possible.

[0107] It is also to be understood that the terminology used herein is for purposes of describing particular embodiments only, and is not intended to be limiting. It is also possible in the present disclosure that steps can be executed in different sequence where this is logically possible.Docket T19763W0001 (222112-2650)

[0108] It must be noted that, as used in the specification and the appended claims, the singularforms “a,” “an,” and “the” include plural referents unless the context clearly dictates otherwise.

[0109] A number of embodiments of the invention have been described. Nevertheless, it will be understood that various modifications may be made without departing from the spirit and scope of the invention. Accordingly, other embodiments are within the scope of the following claims.

[0110] Unless defined otherwise, all technical and scientific terms used herein have the same meanings as commonly understood by one of skill in the art to which the disclosed invention belongs. Publications cited herein and the materials for which they are cited are specifically incorporated by reference.

[0111] Those skilled in the art will recognize, or be able to ascertain using no more than routine experimentation, many equivalents to the specific embodiments of the invention described herein. Such equivalents are intended to be encompassed by the following claims.

Claims

Docket T19763W0001 (222112-2650)WHAT IS CLAIMED IS:

1. A computer-implemented construction coordination and spatial guidance method comprising:providing a field execution device comprising a projector and one or more sensors, the field execution device being coupled to a movable platform, and providing computing hardware in communication with the field execution device;receiving design data representing an installation target, the design data comprising geometric information including object positions, vertex coordinates, and dimensions;generating projector-ready guidance data comprising a pattern and one or more projection parameters for projecting guidance associated with the installation target;storing, in a projection library, the projector-ready guidance data associated with (i) an identifier of the installation target and (ii) a revision of the design data; receiving, via a user interface, a selection of the installation target for field execution, wherein the selection is associated with the identifier and the revision; retrieving, by the computing hardware, the projector-ready guidance data from the projection library;retrieving, by the computing hardware from the projection library, devicespecific projector calibration parameters for projecting onto an as-built surface corresponding to the installation target;estimating, using the one or more sensors, an alignment between the design data and the as-built surface using one or more alignment techniques;directing, by the computing hardware, the movable platform to move into position to achieve the alignment between the projector and the as-built surface;Docket T19763W0001 (222112-2650)causing the projector to project guidance of the installation target onto an as-built location, based on the projector-ready guidance data and the device-specific projector calibration parameters;measuring a deviation between the projected guidance and the as-built surface and generating a corrective output based on the deviation; andgenerating an exportable task record including the identifier, the revision, one or more alignment parameters, the deviation, and a timestamp.

2. The method of claim 1, wherein the projector projects visual guidance directly onto the as-built surface of the as-built location.

3. The method of claim 2, wherein the visual guidance comprises a step-by-step sequence of instructions for installing a physical representation of the installation target at the as-built location.

4. The method of claim 1 , wherein the selection of the installation target comprises a user’s spoken request, a text input, a graphical user interface selection, a gesture input, a button input, or a scanned code.

5. The method of claim 1, further comprising computing a scale of the installation target required to render an outline of the installation target at a correct physical location on the as-built location.Docket T19763W0001 (222112-2650)6. The method of claim 1, wherein measuring the deviation comprises computing at least one of a translational offset, a rotational offset, or a fit error relative to a tolerance threshold, and wherein generating the corrective output comprises updating at least one projection parameter.

7. The method of claim 1 , wherein the projector is mounted on at least one of: an end-effector, a wrist, a mast, a gimbal, a base of the movable platform, a tripod, or a stationary mount.

8. A laser-based spatial augmented reality system comprising:one or more processors;a projection library storing projector-ready guidance data for installation targets, each installation target being associated with an identifier and a revision of design data;a communication interface configured to receive a selection of an installation target for field execution;a field execution device comprising a projector and one or more sensors coupled to a movable platform; andinstructions that, when executed by the one or more processors, cause the system to:retrieve projector-ready guidance data from the projection library based on the identifier and the revision;retrieve, from the projection library, device-specific projector calibration parameters for projecting onto an as-built surface;Docket T19763W0001 (222112-2650)estimate an alignment between the design data and the as-built surface using one or more alignment techniques;control the projector to project guidance corresponding to the installation target onto the as-built surface based on the_retrieved projector-ready guidance data and the device-specific projector calibration parameters;measure a deviation and generate a corrective output; and generate an exportable task record associated with the identifier and the revision.

9. The system of claim 8, wherein the projector projects visual guidance directly onto the as-built surface of an as-built location.

10. The system of claim 9, wherein the visual guidance comprises a step-by-step sequence of instructions for installing a physical representation of the installation target at the as-built location.

11. The system of claim 8, wherein the selection of the installation target comprises a user’s spoken request for the installation target, a text input, a graphical user interface selection, a gesture input, a button input, or a scanned code.

12. The system of claim 8, wherein the instructions, when executed by the one or more processors, cause the system to compute a scale of the installation target required to render an outline of the installation target at a correct physical location on an as-built location.Docket T19763W0001 (222112-2650)13. The system of claim 8, wherein measuring the deviation comprises computing at least one of a translational offset, a rotational offset, or a fit error relative to a tolerance threshold, and wherein generating the corrective output comprises updating at least one projection parameter.

14. The system of claim 8, wherein the projector is mounted on at least one of: an end-effector, a wrist, a mast, a gimbal, a base of the movable platform, a tripod, or a stationary mount.

15. A non-transitory computer-readable medium having instructions stored therein, wherein the instructions, when executed by one or more processors, cause the one or more processors to:receive design data representing an installation target;generate projector-ready guidance data for the installation target and store the projector-ready guidance data in a projection library, wherein the projector-ready guidance data is associated with an identifier of the installation target and a revision of the design datareceive a selection of the installation target for field execution;retrieve the projector-ready guidance data from the projection library based on the identifier and the revision;retrieve, from the projection library, device-specific projector calibration parameters for projecting onto an as-built surface;estimate an alignment between the design data and the as-built surface using (i) one or more sensors and (ii) one or more alignment techniques;Docket T19763W0001 (222112-2650)position a field execution device comprising a laser projector and the one or more sensors to align the laser projection with the as-built surface based on the alignment;cause the laser projector to project the projector-ready guidance data onto the as-built surface based on the retrieved guidance data and the device-specific projector calibration parameters;perform a closed-loop verification by measuring a deviation between the projected guidance data and the as-built surface and generating a corrective output based on the deviation; andgenerate an exportable task record including the identifier, the revision, at least one alignment parameter, the deviation, and a timestamp.