Volkswagen: Quantum Traffic Optimization in Lisbon

The Problem

Traffic optimization is a combinatorial nightmare. Given many vehicles and many possible routes, finding the best assignment grows explosively with scale. Classical computers approximate it; quantum annealers are designed to search exactly this kind of energy landscape.

Volkswagen had been researching quantum routing for several years before Lisbon. In 2017 the company published a research paper with D-Wave showing how traffic flow for taxis in Beijing could be formulated for a quantum annealer using historical GPS data. The Lisbon project was the next step: moving from offline research to a live system on real streets.

The Lisbon Pilot

During the Web Summit technology conference in November 2019, Volkswagen ran what it described as the world’s first pilot project for quantum-optimized public transport routing, together with the Lisbon transport operator CARRIS.

The documented facts of the deployment:

• Nine MAN buses operated by CARRIS took part.
• The buses served 26 stops across four routes, including a link between the Web Summit venue and the Marques de Pombal traffic node.
• The system had two components: a passenger flow prediction layer built on anonymized geo-coordinates and passenger volume data, and a quantum routing algorithm running on a D-Wave quantum computer that calculated an individual route for each bus in near real time.
• The goal was for each bus to be routed around developing congestion before it formed, so that the buses themselves would not add to traffic jams.
• Partners included D-Wave (quantum hardware access), PTV Group (traffic data analytics and city modeling), and Hexad (the driver navigation app).

D-Wave’s then CEO Vern Brownell said the project was “among the first that we know of to make production use of a quantum computer.”

How a Problem Like This Maps to a Quantum Annealer

D-Wave machines solve QUBO problems: Quadratic Unconstrained Binary Optimization. Each decision becomes a binary variable, and costs and constraints are encoded as linear and quadratic terms.

For vehicle routing, a natural formulation asks: should vehicle X take route Y? The objective minimizes total congestion, and penalty terms enforce that each vehicle takes exactly one route.

The snippet below is a simplified illustration of how such a problem can be formulated with D-Wave’s open-source Ocean tools. It is not Volkswagen’s code and does not reproduce the Lisbon system; it shows the general shape of a QUBO routing model.

# Simplified illustration of a QUBO routing formulation
import dimod

bqm = dimod.BinaryQuadraticModel('BINARY')

# For each vehicle-route pair: add a variable with its congestion cost
for vehicle in vehicles:
    for route in routes[vehicle]:
        bqm.add_variable(f'x_{vehicle}_{route}', cost(vehicle, route))

# Penalize assigning the same vehicle to multiple routes
for vehicle in vehicles:
    route_vars = [f'x_{vehicle}_{r}' for r in routes[vehicle]]
    for i, r1 in enumerate(route_vars):
        for r2 in route_vars[i+1:]:
            bqm.add_interaction(r1, r2, penalty_weight)

In a hybrid setup like Volkswagen’s, the classical side handles data ingestion, prediction, and the driver app, while the quantum processor handles only the core optimization step.

What the Pilot Did and Did Not Show

The pilot demonstrated that a quantum annealer could sit inside a live routing loop for a real city service during a major event. That is a meaningful systems-engineering result: data pipelines, the quantum API, and the driver-facing app all had to work together in real time.

What the public record does not contain is a quantitative result. Volkswagen did not publish measured travel-time savings versus a classical baseline, and no quantum advantage claim was made. The project was a proof of concept, and no permanent deployment in Lisbon or elsewhere has been announced since.

Technical Takeaways

• Problem type: Quantum annealing targets combinatorial optimization problems that can be mapped to a QUBO or Ising form, and urban routing fits naturally.
• Hybrid by necessity: The quantum processor solved only the optimization core; everything else was classical.
• Pilot, not production line: Even one of the most cited industrial quantum projects of its era was a time-boxed demonstration, which is a useful calibration for how mature this field actually is.

Sources

• Volkswagen optimizes traffic flow with quantum computers (Volkswagen Group press release, 2019)
• Traffic Flow Optimization Using a Quantum Annealer (Neukart et al., Frontiers in ICT, 2017)