BOOKCase Study
A spatial AI system used to review warehouse delivery conditions before large aerospace equipment, spaceflight hardware, and high-value industrial tools arrived on site.
Industrial Sites · Digital Twins · Logistics Planning · Spatial AI
Large, expensive, and difficult-to-transport aerospace equipment, spaceflight hardware, and high-value industrial tools had to be delivered into a warehouse with limited room for error. The team needed to review the main entry point, road access, staging areas, crane operations, and final placement assumptions before equipment arrived.
The warehouse was captured and reconstructed as a Gaussian splat, creating a navigable 3D digital twin. Reviewers used the scene to inspect the delivery route, evaluate staging options, check placement assumptions, and ask AI-assisted questions about the environment.
The digital twin helped the team review delivery feasibility before execution, reduce the risk of disruption, and scope the contract around the physical conditions of the warehouse.
The project involved pre-delivery planning for large, expensive, and difficult-to-transport aerospace equipment, spaceflight hardware, and high-value industrial tools. The team had to account for the physical limits of the warehouse, including access points, crane operations, staging space, and final placement areas.
The warehouse had one primary entry point for large deliveries. Access to that entry point also depended on conditions outside the building, including whether other delivery trucks could occupy or block the road leading to the warehouse entrance.
This was necessary to ensure that the contract scope was defined correctly. If the equipment did not fit, if access was blocked, or if the planned workflow could not operate inside the warehouse, the team could face delays, rework, and budget pressure.
A digital twin of the warehouse was created so the team could review the site and operational workflows before equipment was delivered.
Warehouse planning is often done through floor plans, photos, walkthrough notes, equipment dimensions, and stakeholder discussions. Those materials are useful, although they can be difficult to evaluate together.
A floor plan can show layout and dimensions. Photos can show specific viewpoints. A spreadsheet can list equipment sizes. Walkthrough notes can capture observations. The planning team still has to combine those materials into a single understanding of the delivery path and warehouse workflow, which often becomes a hurdle during scoping.
For this project, the main question was whether the planned delivery and workflow could operate inside the captured warehouse conditions so contract scoping and pricing could be done with better confidence.
The review focused on:
The warehouse was captured and reconstructed as a Gaussian splat. This created a navigable 3D scene that reviewers could inspect from multiple viewpoints.
The digital twin became a planning reference. Reviewers could move through the environment, examine the proposed route, inspect staging areas, and compare the planned movement of equipment against the warehouse layout.
A typical review path looked like this:
Capture warehouse environmentCreate 3D digital twinInspect primary entry pointReview access from the road to the warehouse entranceEvaluate delivery route through the facilityInspect staging and crane accessCompare equipment placement against available spaceIdentify constraints before deliveryAdjust planning or contract scope
The system also supported AI-assisted questions about the captured scene. This made the digital twin useful as an interactive planning reference beyond a static 3D model.
For example, a reviewer could ask where a specific asset could be staged before final placement. The system could review visible warehouse areas, identify open floor regions near the intended route, flag nearby obstructions, and point the reviewer toward areas that appeared more suitable for temporary staging.
A reviewer could also ask what might interfere with delivery through the main entrance. The system could inspect the visible entry path, identify possible obstructions near the entrance, and note that access also depended on whether delivery trucks were occupying the road leading to the warehouse.
The goal was to help reviewers inspect the site, surface planning constraints, and decide where further measurement or human review was needed.
The digital twin helped connect site conditions to contract scope. Delivery plans for large equipment can require special sequencing, crane access, staging space, additional handling, or operational changes. With the reconstructed 3D scene, the team was able to review those assumptions before delivery, and it gave reviewers a clearer basis for identifying whether the proposed scope matched the warehouse conditions.
The review surfaced issues such as:
Finding these issues during planning helped the team adjust delivery assumptions, workflow design, and scope before equipment arrived.
Gaussian splatting was chosen for this particular task because it offered the following advantages:
The system was used for planning review and site understanding. Final clearance decisions still depended on approved measurements, equipment dimensions, drawings, and human review.
The largest impact was better planning confidence. The team could review the warehouse as a connected environment, examine the primary entry point, account for road access, inspect staging areas, and compare the proposed workflow against the physical conditions of the site. Identifying constraints during planning, rather than on delivery day, kept those issues from becoming rework after equipment was on site, which helped keep the delivery portion of the project within scope and budget.