In 2020 the public sector transport agency responsible for Sydney’s buses, rail and roads - Transport for NSW - began a groundbreaking engagement in the quantum computing sector with Q-CTRL.
Transport approached Q-CTRL with an interest in understanding how they - as an organization with heavy computational challenges - could put quantum computing to work for them. Q-CTRL took on the challenge and has focused on getting TfNSW quantum ready via a combination of quantum professional services and quantum control infrastructure software development.
Transport had a challenge
Many problems in managing transport services are computationally challenging - from timetabling of intersecting modes of transport to meet known patterns of demand through to dynamically routing vehicles to meet changing traffic patterns. These problems quickly grow intractable even for advanced computing tools as the number of “interacting” modes of transport, vehicles, connection points, and stops grows.
Data scientists at TfNSW already engage with some of the most advanced computational techniques available - from analytic approaches to traffic optimization through to deployment of machine learning in parsing troves of user data. The organization is known internationally for having a future-looking perspective and embracing new technologies that can deliver better value for their customers.
So with this mindset, TfNSW posed a simple sounding question to the team at Q-CTRL - could quantum computers eventually be relevant to the specific challenges the agency faces on a day to day level?
This was a rare opportunity for some of our leading transport innovators and quantum computing experts to come together to tackle complex transport network management and congestion problems. We could see all trains, busses, ferries, trams and motorways essentially ‘talking to each other’ to find out where customers are and deploy resources where needed.
Putting Q-CTRL’s expert team to work
The TfNSW request presented a perfect case study in how Q-CTRL could combine its applications-focused expertise with its unique quantum control infrastructure software to deliver real value to customers.
Q-CTRL operates the world’s largest team of expert quantum control engineers. And while the team’s core focus is on the development of quantum control tools and techniques for quantum computing and quantum sensing, they maintain deep expertise in the application of quantum technology to end-user problems.
To approach TfNSW’s question the Q-CTRL team leveraged its deep bench of technical talent. Senior Quantum Control Engineer Dr. Chris Bentley has a background in both quantum computing and transport optimization. With his insights across both domains he was able to identify an opportunity for the application of quantum optimization algorithms to the specific problem of Mobility as a Service.
In this paradigm, travel journeys start and end at individual locations for each passenger. Fleets of small or medium-sized vehicles can transport passengers over these journey legs, and aim to provide short waiting and travel times. Achieving this requires computational optimization of vehicle routes and passenger assignment. And as the number of passengers and locations grows so does the computational complexity of this problem.
Fortunately, Dr. Bentley saw a path to leverage quantum optimization algorithms for this task.
A path to quantum advantage
Through his experience across both the quantum and transport domains. Dr. Bentley identified a new resource-efficient algorithmic compilation which enables modelling of transport networks on near term quantum computers using fewer computational resources than otherwise expected. The next step was to develop new software enabling the team to understand likely performance and identify when useful problems might come within reach.
First, the team built a custom circuit simulator in Boulder Opal that lets you simulate multiqubit algorithms subject to the real noise endemic to near-term quantum computer hardware. With this tool the team demonstrated that the target MaaS problem could really be solved by a quantum computer, even in the presence of imperfections.
Next, the team used these tools to answer how to deliver useful performance when today's systems are so small and error-prone. The team addressed a prototype but practically relevant single-vehicle MaaS network problem with 3-5 customer locations; algorithmic mapping and the use of error-robust quantum logic gates could deliver useful solutions with a success probability of >99.9% (a very high value), over 100X better than previously known approaches. The benefits could be enormous in circumstances when real hardware errors were limiting.
Probability of success in finding optimal route
|Error model||Standard approach||Q-CTRL robust||Q-CTRL advantage|
|Poor hardware calibration||1.26 %||>99.9 %||~300X|
|Random error in each operation||75.72 %||>99.9 %||~240X|
Where to from there?
The million dollar question is when will quantum computers deliver useful value for Transport. The team’s analysis showed that with the 1000-qubit machines predicted to arrive by 2023, real problems of interest to Transport could be encoded using Q-CTRL’s new resource efficient mappings. And so long as larger networks can be broken into these constituent parts, a path to real quantum advantage in large-scale network optimization looks to be possible.
The next step, of course, is making those quantum computers perform well enough - for that you can definitely use Q-CTRL’s quantum firmware.
The team at Q-CTRL is currently building a custom software package for Transport for NSW, to deliver the ability to simulate and optimize quantum algorithm performance for mobility-as-a-service problems. It’s all part of our objective to make quantum technology useful for our partners and clients.