BMW and Zapata have applied quantum computing to optimise vehicle production. The companies created generative AI techniques that have been tested in multiple BMW plants. Tests have shown that these new techniques are out-performing current assembly line solvers.

The two organisations are part of MIT’s Centre for Quantum Engineering (CQE) Consortium. The Consortium helps support the work of MIT students and researchers.

The companies developed simulations using Zapata’s Generator-Enhanced Optimisation to optimise vehicle production schedule across multiple plants. In many cases, GEO outperformed state-of-the-art solvers in minimising assembly line idle time while maintaining monthly vehicle production targets. The work was done on Zapata’s Orquestra software platform.

“The problem that BMW presented to our team is an excellent quantum computing use case that addresses an incredibly complex, real-world challenge of commercial interest,” said Dr William D Oliver, Professor of Electrical Engineering and Computer Science and of Physics at MIT, and Director of The Centre for Quantum Engineering. “That’s precisely why we created our Quantum Science and Engineering Consortium (QSEC) – connecting the best and brightest from the academic landscape with industry partners to solve real-world problems.”

“At BMW, we’re always looking for new, innovative ways to drive operational efficiency at our manufacturing plants,” said Marcin Ziolkowski, Emerging Technologies Manager at BMW Group. “As you might imagine, optimising our production schedule is an incredibly complex and unique challenge. There are a wide range of possible configurations and a high number of constraints, including varying production rates between shops, a discrete set of shift schedules, and the need to prevent overflows and shortages in the buffers between steps in the manufacturing process. This initiative aligns perfectly with MIT’s broader research and educational mission. Working with Zapata and CQE, we were able to prove that GEO outperforms other techniques in production planning.”

“We ran roughly a million optimisation runs cycling through dozens of various algorithms, problem configurations and optimiser solutions to benchmark their performance against each other,” said Yudong Cao, CTO and co-founder at Zapata Computing. “GEO uses quantum or quantum-inspired generative machine learning models to learn from and improve upon the results generated by classical solvers. As we worked on new ways to address this challenge, we kept MIT’s mission front-of-mind – ensuring that we were helping to advance knowledge and educate students in science, technology, and other areas of scholarship that will best serve the nation and the world in the 21st century.”