Foretellix and Voxel51 said they have launched a joint solution designed to improve the validation of autonomous vehicle systems by transforming real-world driving data into high-fidelity three-dimensional simulations.
The integration combines Foretellix’s Physical AI toolchain with Voxel51’s visual AI data platform, enabling automated data curation, auditing, neural reconstruction and scenario variation within a single workflow. The companies said the approach addresses a growing challenge in autonomous vehicle development, where end-to-end AI architectures require large volumes of high-quality data that is costly and time-consuming to collect.
See also: China Autonomous-Driving Developer CiDi Opens IPO Subscription in Hong Kong
Under the joint setup, real-world drive logs are ingested and analysed to identify gaps in operational design domains. Voxel51’s platform audits the data to detect issues such as sensor misalignment or calibration errors before it is reconstructed into 3D scenes using neural techniques. Foretellix then generates controlled scenario variations and synthetic sensor data to support closed-loop simulation and validation.
“Safety is the foundation of Physical AI,” said Ziv Binyamini, chief executive and co-founder of Foretellix, adding that the collaboration brings together real-world grounding with controllable scenario variation. Brian Moore, co-founder and chief executive of Voxel51, said data quality becomes “mission-critical” as datasets grow, and that the joint solution helps engineering teams validate autonomous systems more efficiently.
The companies said the solution allows developers to scale testing by filling data gaps with realistic synthetic scenarios while maintaining strong data integrity, supporting safer and faster deployment of autonomous driving technologies.