Engineering Process & CAD Validation

Subway Sentinel — SIP408 Fusion 360 Development

This page documents the engineering process behind Subway Sentinel, a rail-mounted robotic inspection platform concept for the NYC subway tunnel. The current SIP408 work focuses on building a realistic Fusion 360 CAD layout that integrates the tunnel geometry, track placement, train reference, bogie interface, platform frame, robot placement, and the future GPR end-of-arm tool.

Earlier SIP work helped define the motion concept, but the current focus is on physical layout and mechanical feasibility. The goal is to show how each design decision was tested, corrected, and moved closer to a defensible engineering model.

Current Status — Updated May 26, 2026

The tunnel reference model, rail placement, train visual reference, bogie reference model, and robot selection direction have been established. The current design task is building the rail-mounted platform frame and rough GPR EOAT layout. By the end of Week 3, the goal is to show functional CAD progress and early system integration, not a fully validated final MVP.

Design Iteration Log

This project changed as the CAD work became more realistic. I started with a broad tunnel and robot motion concept, but the design had to be corrected once I began working through tunnel geometry, rail spacing, train clearance, bogie sizing, and platform mounting. This section shows the design trail instead of only showing the final model.

  • What I tried:

    I started with an early tunnel model and a larger ABB robot motion concept.

    What did not work:

    The early setup was useful for exploring the idea, but it was not tied closely enough to the actual tunnel, rail, bogie, and platform constraints I am using now.

    Current decision:

    I moved the project into Fusion 360 and rebuilt the layout around physical geometry first.

  • What I tried:

    I researched tunnel geometry along the 1 Line and found that different contractors and construction methods created different tunnel forms.

    What I learned:

    There is not one universal tunnel shape that represents the entire 1 Line.

    Current decision:

    I selected a deep two-track underground tunnel reference as the working CAD baseline for this phase of the project.

  • What I considered:
    I looked at a few possible directions for the rail platform, including designing a custom railcar, using a maintenance-style flat car, or building the system around an existing bogie reference.

    What I learned:
    Designing a full railcar from scratch would add too much engineering scope for this project. It would require detailed structural, suspension, and wheelset design, braking assumptions, load certification, manufacturing planning, and validation, all of which are outside the realistic timeline for SIP408. A maintenance flat car also does not align with the design direction, as Subway Sentinel needs a compact robotic inspection platform, not a full work car.

    Current decision:
    For this phase, I am using an existing bogie-based platform approach instead of designing my own railcar. The bogie already represents a proven rail vehicle support concept, so I can focus my engineering effort on the platform frame, robot mounting structure, GPR EOAT, clearance envelope, and validation plan.

    Why this matters:
    This keeps the project realistic. The goal is not to redesign rail vehicle technology from scratch. The goal is to design and validate a robotic inspection system that uses an existing rail-based support geometry.

  • What I tried:

    I added rail geometry into the tunnel and placed two tracks from the tunnel centerline.

    What worked:

    This gave me a controlled layout for the left and right track centerlines, rail spacing, train reference, bogie placement, and future platform design.

    Current decision:

    The rail layout is now being used as the foundation for the rest of the Fusion 360 assembly.

  • What I tried:

    I imported and scaled an R62A subway car model to show the scale of the train inside the tunnel.

    What did not work:

    The model was built as a mesh/OBJ file, not clean engineering CAD. It was not worth converting to STEP because it would create a heavy, faceted body.

    Current decision:

    The R62A model is being used only as a visual size reference, not as official engineering geometry. This is also the model the 1 line uses currently for its rolling stock.

  • What I tried:

    I spent time looking for a usable R62A or MTA-style bogie CAD model.

    What did not work:

    Official MTA bogie CAD was not available, and many public models were either the wrong size or did not include useful mounting features.

    Current decision:

    I selected a TR22 bogie reference because it has usable center pin, bolster, and side bearer geometry for early platform interface design.

  • What I am doing now:

    I am designing the platform that will sit on top of the two bogies and support the robot, power components, controls, and GPR payload.

    Main engineering concern:

    The frame has to resist bending and torsion from the robot, not just hold static weight.

    Current decision:

    The platform will use a rigid box-section/HSS frame concept with reinforced bogie adapter zones and a dedicated robot mounting area.

Previous SIP311 Motion Feasibility Study

Before the current Fusion 360 CAD layout, SIP311 included an early RobotStudio motion feasibility study. That work tested whether a large ABB IRB 8700 robot could follow a sectional tunnel inspection arc while maintaining the tool's orientation toward the tunnel surface.

This study helped define the inspection motion concept, but it does not represent the current SIP408 CAD assembly. The current work has moved into Fusion 360, so the platform, bogies, rail layout, tunnel geometry, and GPR tool can be developed before motion validation is repeated.

Tunnel Geometry

The tunnel geometry for Subway Sentinel is based on the IRT 1 Line in Upper Manhattan, specifically the Washington Heights deep-tunnel section around 168th Street, 181st Street, and 191st Street. I selected this section after researching the West Side Line and finding that the 1 Line includes different construction types, including cut-and-cover subway, viaduct, and rock tunnel sections. Because of that, I did not model the entire 1 Line as one generic tunnel.

For SIP408, I narrowed the CAD baseline to a typical two-track deep tunnel reference. This gives the project a controlled envelope for track placement, train clearance, bogie layout, platform sizing, robot reach, and GPR inspection planning. The model is a public-source conceptual reference, not an official MTA as-built drawing.

Track & Rail Layout

The rail layout uses an A-Division straight/tangent track gauge of 1,429 mm for the Fusion 360 CAD model. I measured the gauge between the inside faces of the rail heads, not from rail centerline to rail centerline. This gives the project a consistent rail baseline for placing the bogies, platform frame, train reference, and future robot/GPR assembly. Wider gauge values are only considered for tighter curve sections, so the current model uses the straight-track baseline.

R62A Train Visual Reference

I added a scaled R62A subway car model to illustrate how much space a train occupies in the tunnel. This is important because Subway Sentinel has to be evaluated against the train envelope, tunnel wall clearance, and track location.

The model was originally an OBJ mesh, so I am using it as a visual reference only. I did not convert it to STEP because that would create a heavy, faceted CAD body instead of clean mechanical geometry.

Bogie Interface Study

The bogie became one of the more challenging CAD problems because the official R62A bogie geometry was unavailable. I reviewed several public models, but most were either the wrong size, too simplified, or missing useful mounting features.

I selected a TR22 bogie reference because it includes a usable center pin, bolster area, and side bearer geometry. This gives me something practical to build the early platform interface around.

The TR22 bogie is not an official MTA bogie. After scaling it to match the rail gauge assumption, it is slightly wider than ideal. I am documenting this as a design constraint because it affects access to the lower wall and platform inspection path.

Platform Frame Development

Status: I am currently in the design phase for this as of May 26, 2026.

This platform has to connect to the bogies, support the robot, carry power and control components, and eventually support the GPR inspection tool.

The frame cannot just be a flat deck. It has to act as a rigid torque structure because the robot will impose bending and torsional loads during motion. For that reason, the current design direction uses rectangular box-section/HSS-style members instead of open I-beams.

Design Focus:

  • Two-bogie rail-mounted platform layout

  • Reinforced front and rear bogie adapter zones

  • Rigid rectangular tube / box-section frame concept

  • Dedicated robot mounting torque box

  • Space reserved for power, controls, and GPR support hardware

  • Future validation through FEA, deflection checks, and anti-tip review

Integrated Fusion 360 Assembly

The integrated assembly brings the individual design decisions together into one Fusion 360 model. This view is used to verify that the tunnel geometry, rail placement, train reference, bogies, platform frame, robot, and future GPR EOAT are working together as a single system.

This is the main CAD model that will support the next phase of validation, including robot reach, platform structure, clearance review, and motion planning.