Automated Dwell Equity Score (ADES)

GMI-RAIL-029Rail TransportationInfrastructure-Dependent

Definition

Ratio of platform-stranded boarding attempts to total boarding attempts under ATO dwell management, segmented by passenger mobility category, measuring whether automated dwell decisions equitably serve all passenger types.

The Measurement Gap

The Measurement Gap

Transit agencies deploying ATO (Automated Train Operation) measure dwell time adherence: whether trains depart on schedule. What they don't measure is who gets left behind when the algorithm decides to close the doors. ATO systems optimize for headway maintenance and schedule adherence, but passengers don't all move at the same speed. A dwell time that works for able-bodied commuters may systematically strand passengers with mobility impairments, parents with strollers, or elderly riders.

This gap has legal implications. ADA compliance isn't just about ramp availability: it's about whether the automated system's dwell decisions create discriminatory service outcomes for protected populations. No transit agency currently measures this at scale because the data infrastructure (platform sensors, boarding-event detection, mobility categorization) largely doesn't exist.

Decision Supported:

Is ATO closing doors and departing while mobility-impaired, elderly, or encumbered passengers are still boarding? Does dwell management need station-specific or time-specific adjustments? ADES provides the equity data to answer these questions before they become regulatory complaints or litigation.

Formula

Plain Language: Count the times a passenger was visibly attempting to board when doors closed under ATO control. Divide by total boarding events. Segment by passenger mobility category if sensors or platform cameras can distinguish.

Notation:

ADES = 1 − (Stranded_boarding_attempts / Total_boarding_attempts)

Variables:

Variable Definition Unit Source Classification
Stranded_boarding_attempts Boarding attempts where a passenger was in the door zone when ATO initiated door closure and departed Count Platform camera/sensor system INFRASTRUCTURE-DEPENDENT
Total_boarding_attempts All passengers who approached a door during the dwell period Count Platform camera/sensor system INFRASTRUCTURE-DEPENDENT

Sign Convention: 1.0 = no passengers stranded (ideal). Lower values indicate more passengers left behind. Segment by station, time of day, and passenger type.

Edge Cases:

  • Distinguish between passengers who were in the boarding zone when ATO decided to close and passengers who arrived at the platform after doors were already closing. The first is an ATO failure; the second is a passenger-timing issue.
  • Peak vs. off-peak: ATO may perform differently at crush-load stations (aggressive departure to maintain headway) vs. lightly-loaded stations. Track separately.
  • Privacy: passenger mobility categorization from camera data raises privacy concerns. May need to measure aggregate stranding rate without individual classification, or use opt-in data from accessibility assistance requests.

Proposed formula, pending validation pilots.

At a Glance

Category
Autonomous Train Operations and Human Capability
← Time-to-Correct-Classification (TCC)Energy Optimizer Override Value Index (EOOVI) →

Assess Your Organization

The AI-GPS Assessment evaluates your AI governance maturity and recommends KPIs like this one for your implementation roadmap.

Take the Free AI-GPS Assessment