日本語 
TIER IV has introduced new software stacks for Level 4 autonomous driving built around data centric AI. The stacks are released through Autoware, the open source autonomous driving platform widely used by developers and automakers. The goal is simple. Expand where and how Level 4 vehicles can actually operate.
Software is programmed with the very intention of being hardware independent, implying automotive manufactures are liberated from particular chips or sensor configurations. It supports different system on chip architectures and sensor configurations depending on the vehicle design. Using TIER IV’s MLOps platform alongside the stack, developers can keep retraining and improving AI models using new driving data.
Two system approaches are included. One combines perception AI and planning AI using diffusion models to track changes around the vehicle and generate driving decisions that mimic human behavior. The second is an end to end system that handles perception, planning and control in one learning pipeline.
Also Read: Nissan, Uber, and Wayve Partner to Launch Robotaxi Pilot in Tokyo
TIER IV is now testing these capabilities across three locations with academic partners. Trials are running in Tokyo with The University of Tokyo, in Pittsburgh with Carnegie Mellon University, and in Munich with Technical University of Munich. The experiments test various vehicle types under different urban traffic conditions to measure their actual performance results.
English