AutV

AutV is a modular software framework for autonomous vehicles designed to standardize the interfaces between sensing, perception, planning, and control components. The project aims to enable cross-platform deployment by providing a common data model, middleware layer, and reference implementations that can run on different hardware configurations and operating systems. AutV supports sensor abstraction, allowing lidar, radar, cameras, and other inputs to be integrated through a uniform interface, while its perception pipeline produces a world state used by planners to generate safe trajectories. The architecture emphasizes safety, verifiability, and real-time performance, with facilities for runtime monitoring, fault isolation, and deterministic scheduling.

The framework includes a set of core modules: sensor manager, perception and mapping, world modeling, behavior and trajectory planning, control, and a safety and compliance layer. A serialization standard, the AutV Data Model, defines message formats for inter-module communication, enabling reproducible testing and simulation. AutV is designed for both on-vehicle execution and cloud-assisted workflows, supporting offline training, simulation, and over-the-air updates.

Development is coordinated by the AutV Consortium, a collaborative community with representation from automotive manufacturers, suppliers, and academic institutions. The project follows open standards principles and maintains an emphasis on reproducibility, code quality, and security. Adoption varies by region and use case, from research trials to production deployments in limited domains.

See also: autonomous vehicle, software architecture for autonomous systems, ROS, AUTOSAR.