Lighthouse AdvisorySLED AI Adoption Intelligence

InfrastructureIssue 074 records

Transportation

Two U.S. state DOT items and two international pilots, no independent outcome data. Alabama began an AI-enabled photo speed pilot in an I-65 work zone with legislative reporting and human review. Ohio's Honda connected-car inspection pilot reported 89-99% detection rates (vendor-reported) alongside privacy concerns. Incheon's port authority is piloting an AI agent for pre-work risk assessments. Kapsch's Barcelona AI signal pilot keeps safety parameters under deterministic control. No tolling, parking, fleet or U.S. transit evidence was found.

Evidence records
4
Cross-source patterns
1
Evidence classes
3 independent reporting1 vendor claim
Outcomes
3 emerging1 mixed
Source freshness
4 new this fortnight
Research completed
2026-11-05

Choose a role to see its takeaway beside every record in the ledger.

Patterns across the evidence

Synthesis · Lighthouse Advisory interpretation

1 pattern, each supported by at least two sources

  1. Agencies are fencing AI with explicit human or deterministic limits

    Alabama routes every AI-flagged violation to ALEA human review with civil-only penalties, and Barcelona's pilot locks pedestrian and emergency parameters outside AI control; neither has outcome data yet.

    Operating questionWhich decisions in the agency's AI use must remain outside model control, and how is that enforced?

    Supporting evidenceAlabama Department of Transportation (ALDOT) and Alabama Law Enforcement Agency (ALEA)Kapsch TrafficCom with City of Barcelona and CENIT (CIMNE)

Evidence ledger

Full record · every source keeps its link and limitations

4 records
  1. Independent reportingEmergingNew this fortnight

    Alabama DOT and ALEA launch AI-enabled photo speed enforcement pilot in I-65 work zone

    Under the April 2026 Alabama Work Zone Safety Act, ALDOT and ALEA are piloting photo speed enforcement near I-65 milepost 209 at the Walnut Creek Bridge replacement. Vendor Elovate describes its cameras as machine learning/AI-enabled. Fines are $250 civil penalties for more than 10 mph over the limit, issued only while workers are present, no earlier than Nov. 16. ALEA reviews evidence and runs hearings. The act requires effectiveness reports to the legislature. Context: 2,543 work-zone crashes, about 11 deaths and 709 injuries last year.

    Alabama Department of Transportation (ALDOT) and Alabama Law Enforcement Agency (ALEA)Chilton County, Alabama, USA2026-11-04

    Why it matters, evidence and limitations
    Why it matters
    A live U.S. state DOT pilot with a statutory evaluation requirement, a split DOT/law-enforcement role model and worker-presence limits that other states weighing work-zone cameras can compare.
    Evidence and measured results
    Press report of agency statements; AI role is the vendor's website claim, not an ALDOT description. No accuracy, speed or crash results yet.
    Limitations and uncertainty
    Pilot not yet issuing citations; AI accuracy and false-positive rates are undisclosed; the AI description comes from vendor marketing.
    Independent reporting of agency announcement Role takeaways & implementation
  2. Independent reportingEmergingNew this fortnight

    Incheon Port Authority pilots an AI agent to automate pre-work risk assessments

    IPA launched a pilot in which an AI agent takes work procedures, equipment and on-site photographs, draws on accumulated accident history, and within seconds identifies hazards by zone, assigns risk ratings and recommends safety measures. It targets hazards easy to miss in paperwork, such as fall zones and equipment-worker path intersections. Previously safety officers manually cross-referenced incident records and regulations. IPA already uses AI CCTV, a pothole-repair robot and AI fire surveillance.

    Incheon Port Authority (IPA)Incheon, South Korea2026-11-04

    Why it matters, evidence and limitations
    Why it matters
    A concrete agentic-AI use in port safety knowledge work, directly comparable to U.S. port authority and DOT job-hazard analysis processes.
    Evidence and measured results
    Agency announcement as reported; no accuracy, time-saving or incident outcomes reported; pilot just launched.
    Limitations and uncertainty
    Korean regulatory context; no published results; vendor and model not named.
    Independent reporting of agency announcement Role takeaways & implementation
  3. Vendor claimEmergingNew this fortnight

    Kapsch TrafficCom MATCAB wins Barcelona Innova Lab challenge; AI adaptive signal pilot to follow

    Kapsch won the fifth Barcelona Innova Lab Mobility Challenge (organized by Barcelona City Council and Fira de Barcelona) with MATCAB, an AI agent system integrated with EcoTrafiX that replaces static signal plans using multimodal sensor data and simulation. Pedestrian crossing times, emergency priority and regulatory constraints stay under deterministic control; every AI change is validated in a digital twin and reversible. Pilot targets: at least 5% lower pedestrian waits and at least 10% fewer severe congestion events. Site not yet chosen.

    Kapsch TrafficCom with City of Barcelona and CENIT (CIMNE)Barcelona, Spain2026-11-04

    Why it matters, evidence and limitations
    Why it matters
    Shows a guardrail pattern (deterministic safety constraints, digital-twin validation, reversibility) that U.S. signal agencies can require in AI signal procurements.
    Evidence and measured results
    Vendor press release; targets are aims, not results; pilot location not selected.
    Limitations and uncertainty
    Vendor claim; European city; no measured outcomes; innovation-challenge selection is not an evaluation.
    Vendor press release Role takeaways & implementation
  4. Independent reportingMixedNew this fortnight

    Ohio DOT DriveOhio and Honda connected-vehicle road-defect detection pilot results

    Honda and DriveOhio finished a two-year pilot in January 2026 covering about 3,000 miles. Vehicle cameras and LiDAR with edge AI flagged potholes, damaged guardrails, blocked signs, rough pavement and shoulder drop-offs; Honda cloud analysis and Parsons asset software turned findings into prioritized ODOT work orders. Honda reported 99% detection for damaged or obstructed signs, 93% for guardrails and 89% average for potholes. The article pairs this with Northeastern/Consumer Reports findings of extensive connected-car data sharing.

    Ohio Department of Transportation (DriveOhio) with Honda and ParsonsCentral and southeastern Ohio, USA2026-11-02

    Why it matters, evidence and limitations
    Why it matters
    Why now: the 11-02 coverage packages the results with a privacy critique, relevant as DOTs consider crowd-sourced inspection data. Measured detection rates inform maintenance AI business cases.
    Evidence and measured results
    Detection rates are Honda-reported; methods, false-positive rates and ground truth are not described.
    Limitations and uncertainty
    Pilot ended January 2026; vendor-reported accuracy; secondary reporting; no cost or crew-time outcomes.
    Independent reporting of pilot results Role takeaways & implementation

How to read this edition

Source findings, measured results and limitations come from the cited publications. Patterns, operating questions, role takeaways and implementation considerations are Lighthouse Advisory interpretation, stated as questions to validate locally rather than guaranteed outcomes. Vendor and operator claims are labeled as claims. Full research method.

Independent reporting
Independent reporting with attributable sources but without a formal evaluation design.
Vendor claim
A supplier-provided assertion that has not been upgraded to independent evidence.