Quantum

Where robotics meets quantum computing

Australian physicists say they have adapted techniques from autonomous vehicles and robotics to efficiently assess the performance of quantum devices.

An ion trap at the Sydney Quantum Control Laboratory used in the research. Credit: University of Sydney

Writing in the journal Quantum Information, a University of Sydney team reports that its new approach has been shown experimentally to outperform simplistic characterisation of these environments by a factor of three, with a much higher result for more complex simulated environments.

Lead author Riddhi Gupta says one of the barriers to developing quantum computing systems to practical scale is overcoming the imperfections of hardware.

Qubits – the basic units of quantum technology – are highly sensitive to disturbance from their environments, such as electromagnetic noise, and exhibit performance variations that reduce their usefulness.

To address this, Gupta and colleagues took techniques from classical estimation used in robotics and adapted them to improve hardware performance. This is achieved through the efficient automation of processes that map both the environment of and performance variations across large quantum devices.

“Our idea was to adapt algorithms used in robotics that map the environment and place an object relative to other objects in their estimated terrain,” she says.

“We effectively use some qubits in the device as sensors to help understand the classical terrain in which other qubits are processing information.”

In robotics, Gupta says, machines rely on simultaneous localisation and mapping (SLAM) algorithms. Devices like robotic vacuum cleaners are continuously mapping their environments then estimating their location within that environment in order to move.

The difficulty with adapting SLAM algorithms to quantum systems is that if you measure, or characterise, the performance of a single qubit, you destroy its quantum information.

Gupta has developed an adaptive algorithm that measures the performance of one qubit and uses that information to estimate the capabilities of nearby qubits.

“We have called this Noise Mapping for Quantum Architectures.,” she says. “Rather than estimate the classical environment for each and every qubit, we are able to automate the process, reducing the number of measurements and qubits required, which speeds up the whole process.”

Co-author Michael J Biercuk, director of the Sydney Quantum Control Laboratory, says the work demonstrates that state-of-the-art knowledge in robotics can directly shape the future of quantum computing.

“This was a first step to unify concepts from these two fields, and we see a very bright future for the continued development of quantum control engineering.”

Leave a Reply

Your email address will not be published. Required fields are marked *