Proof of Concept — Operational Demonstration
The proof of concept demonstrates how autonomous drone capture and automated analytics can support yard operations without disrupting current workflows. The proof of concept focuses on demonstrating operational workflows for railcar safety hardware inspection, railcar identification, and yard scanning and visibility.
POC Overview
The POC uses a 90-day operational demonstration to validate that autonomous drone capture and automated analytics can reliably collect inspection and yard intelligence data. The goal is to validate operational workflows, not replace existing inspection processes.
POC duration
90 days
Capture infrastructure
Autonomous drone dock deployment
Primary capabilities
Railcar safety hardware inspection, railcar identification, and yard scanning.
Technology components
Drone-based capture, computer vision inspection models, and operational dashboards.
Mission Types
Five mission types support the POC. Types 1 and 2 are Phase 1 core. Type 3 is optional in this POC. Types 4 and 5 are suggested for later phases as capabilities mature.
Mission Type 1 — Yard Survey Mission
POCBroad coverage of the yard to capture inventory and build the operational map. Sets the baseline for car counts and layout.
Mission Type 2 — Close Inspection Mission (Rail Car)
POCTargeted close-up inspection of rail cars for defect detection and safety appliance checks.
Mission Type 3 — Close Inspection Mission (Material Inventory)
Optional in POCFocused capture of material and equipment inventory for cycle counts and layout optimization.
Mission Type 4 — On-Demand Spot Inspection (Blue Flag / Brake Engaged)
Suggested Phase 2Quick verification of blue signal and hand-brake status when requested for work-zone safety.
Mission Type 5 — Close Inspection Mission (Switch Position Verification)
Suggested Phase 3Verification of switch positions to support switching logic and yard operations.
Operational Workflow
The system operates in four stages:
Drone platforms collect imagery across railcars and yard environments.
Computer vision models detect railcars and safety hardware.
Detected observations are evaluated using configurable rules.
Results are presented through dashboards for inspection review and asset location verification.
90-Day POC Timeline
Work runs in parallel over 12 weeks. Tasks grouped by phase: Setup → Build → Integrate. Click a task for details.
POC Outcomes
The POC demonstrates:
- Operational drone capture workflows
- Automated inspection analytics
- Railcar identification capability
- Yard scanning workflows
The outcome is validation that drone capture and analytics can reliably collect inspection and yard intelligence data. The system assists inspectors and yard operations rather than replacing them.
TTX Use Case Difficulty vs Value
Operational use cases mapped by implementation difficulty and suggested phasing. Phase 1 builds foundational capabilities; Phase 3 targets advanced operations.
| Use Case | Difficulty | Why | Phase |
|---|---|---|---|
| Material audits / cycle counts | Easy | Requires OCR + car location | Phase 1 |
| Yard mapping | Easy | Mostly photogrammetry / manual mapping | Phase 1 |
| Track inspections | Easy | Simple visual imagery review | Phase 1 |
| Lost railcar search | Easy | OCR + yard map | Phase 1 |
| Layout optimization | Easy | Analysis of mapped yard data | Phase 2 |
| Yard audits | Medium | Requires consistent car location accuracy | Phase 2 |
| Box car inspection | Medium | Object detection models | Phase 2 |
| Auto rack inspection | Medium | Tall structure but visually consistent | Phase 2 |
| Tall car inspection | Medium | Requires different capture geometry | Phase 2 |
| Blue signal verification | Medium | Object detection but clear visual signal | Phase 3 |
| Hand brake verification | Medium | Requires better resolution + angles | Phase 3 |
| Auto rack bellows inspection | Hard | Flexible structure + lighting variation | Phase 3 |
| Switch position tracking | Hard | Small objects + occlusion | Phase 3 |
| Switching optimization | Hard | Requires movement tracking + operations logic | Phase 3 |
| Yard movement planner | Hard | Needs historical movement + predictive logic | Phase 3 |