LAMbDA — home Faculty of Transportation Sciences, CTU in Prague
1 Research pillar 1

Cooperative Traffic Management and CCAM

Uncoordinated automation makes intersections worse, not better.

  • MAVEN (H2020)
  • GLOSA and V2X message sets
  • Multi-agent signal control
  • Variable speed limits
  • Mobility on Demand
  • Impact assessment
Why this pillar exists

What must a control centre become when it no longer commands the network?

Left to optimise locally, automated vehicles produce phantom jams, unstable queue formation and lower intersection throughput. The question is not whether automation arrives — it is what a traffic management centre has to become in order to stay in control of a network it no longer fully commands.

We treat this as one field rather than two. Connectivity and cooperation are not a separate topic from traffic control: they are the basis on which distributed, multi-agent control becomes possible at all. The arc runs from a control system operating today on a European capital's ring road, through three generations of soft-computing controllers, to negotiation-based cooperative control validated in European field trials — and the deployed system at the start of that arc is what makes the rest credible.

Scientific contribution

  • Hierarchical self-organising control. Multi-level top-down guidance of self-organised dynamic platoons outperforms both fully centralised and fully decentralised control — the architectural conclusion of the MAVEN programme.
  • Negotiation as a control primitive. Mutually beneficial information exchange between a signal controller and an approaching platoon, formalised, simulated and then reduced to a signal-plan stabilisation cost function.
  • Impact attribution under partial penetration. Queue-state estimation and effect attribution when only a fraction of the fleet is connected — the regime every city will actually operate in for the next two decades.

Operational evidence

+34 %Network capacity
−74 %Average queue length
−52 %Delay
−12 %CO₂ emissions

From H2020 MAVEN, where CTU led the impact assessment work package; figures quoted from the project's CORDIS report. Queue-length estimation improved by 50 % at low connected-vehicle penetration.

Also: a motorway line traffic control system running today on the Prague City Ring Road; patent NL 2018711 on signal-plan stabilisation; and four V2X message extensions proposed to ETSI ITS and the Car2Car Communication Consortium.

  • Methods used here
  • M5 SUMO · Vissim · Aimsun
  • M2 Multi-agent systems · fuzzy inference
  • M1 State-space models · MPC
  • M6 Impact assessment by microsimulation
  • All six groups →
Evidence in pictures

Twenty years in one arc

Three-level multi-agent traffic management architecture
Operators intervene at the knowledge level, not at the device level. Sensor and actuator level, data level, knowledge level — with intelligent agents negotiating control strategies where the human operator also intervenes. This separation is the direct ancestor of our human-in-the-loop digital twin architecture in Pillar 2. Source: LAMbDΛ.
Automated vehicle receiving GLOSA speed advice during a MAVEN field trial
Cooperation in the field. A DLR FASCarE automated vehicle receives signal phase and timing information and adopts the recommended approach speed automatically. Source: LAMbDΛ, MAVEN project material.
Flow-occupancy diagram under different variable speed limit settings
Why harmonisation buys capacity. The flow–occupancy relationship shifts with the posted limit, from Czech motorway sensor data — the empirical basis of the deployed line control system. Source: LAMbDΛ.
GLOMODO system architecture diagram
Decision support inside a control centre. GLOMODO feeds three modules — traffic quality, economic benefit and a mental model of the city — for Prague's Main Traffic Control Centre, over 108 strategic detectors. Labels in Czech. Source: LAMbDΛ.

A declared direction with no results yet

Human–machine interaction and trust in automated transport: handover in semi-automated systems, in-vehicle AR transparency, external signalling towards pedestrians and cyclists, and long-term trust formation. We have the driving simulator and eye-tracking infrastructure and no findings to report — which is precisely why we are looking for a partner on it.

The evidence base

Projects behind this pillar

ProjectFunder and identifierPeriodOur rolePillar

Publications

Every entry links to its DOI. Open-access items are marked and can be read without a subscription.

    All publications All projects
    Collaboration

    Who we want to hear from

    We are looking forWhat we would contributeTarget instruments
    Cities and control centres preparing for mixed automated traffic Impact assessment, algorithm design, before-and-after evaluation methodology, simulation of local conditions Horizon Europe Cluster 5 · CCAM Partnership
    CCAM pilot and demonstration consortia Evaluation work package leadership; modelling of pilot sites; queue and emissions estimation at low penetration CCAM Partnership · national testbeds
    Roadside ITS and traffic controller suppliers Negotiation-based control, signal-plan stabilisation, cooperative queue estimation Bilateral R&D · TA ČR TREND · Eurostars
    Human factors groups working on trust in automation A driving simulator with eye tracking and a declared agenda on handover, in-vehicle AR and external signalling Horizon Europe Cluster 5 · CCAM Partnership
    How we work with partners lambda@fd.cvut.cz