#Japan #Chiyoda #T2 #PFN #MIRU2026

T2 and PFN's Collaborative Research Presentation at MIRU2026

T2, based in Chiyoda, Tokyo, is set to make a significant contribution to the field of autonomous transportation as it collaborates with Preferred Networks (PFN) to present their groundbreaking research on autonomous trucks at the upcoming MIRU2026 symposium. This prestigious event is scheduled for August 3, 2026, in Nagasaki Prefecture, Japan.

Under the leadership of T2's CEO, Masatomo Kumabe, the organization is preparing to revolutionize freight transport with the launch of level 4 autonomous truck services by the fiscal year 2027. To achieve this ambitious goal, T2 has been focusing on enhancing 3D object recognition accuracy on highways by utilizing an advanced array of sensors, including cameras and LiDAR systems. Since 2024, T2 has implemented practical demonstration runs using level 2 autonomous trucks to transport goods for partner companies, further refining their technology through extensive data collection.

Despite these advancements, the challenges of recognizing rare scene occurrences—such as pedestrians or falling objects—remain. These scenarios can significantly increase the risk of accidents, and the frequency of their occurrence on highways is rather low. Consequently, acquiring sufficient real-world driving data to prepare for such situations poses a significant hurdle. Additionally, due to the longer braking distances associated with large logistics trucks, early recognition of pedestrians from afar is crucial. Unfortunately, existing real-world data often falls short in addressing these challenges.

To overcome these obstacles, T2 and PFN plan to unveil a novel methodology at MIRU2026 that seamlessly synthesizes virtual object data into actual driving scenarios without creating visual inconsistencies. Preliminary results from experiments conducted by both companies indicate that training AI systems with this synthesized rare scene data can significantly enhance pedestrian recognition accuracy. The practical application of this technique could allow for more timely and precise recognition of rare events, such as pedestrian appearances on highways, which in turn is anticipated to greatly improve the overall safety of autonomous driving systems.

Following the outcomes from MIRU2026, T2 aims to further develop recognition models capable of handling rare scenes, which are rarely encountered during standard driving situations, as they work towards achieving level 4 autonomous driving.

About T2

Founded on August 30, 2022, T2 is committed to developing autonomous driving systems and offering freight transport services utilizing level 4 autonomous trucks. They are also involved in a variety of related service businesses. The company’s headquarters is located at the Hibiya International Building in Chiyoda, Tokyo.

For more information, visit the official T2 website at T2 Auto. Follow them on social media platforms such as X and note for the latest updates.