"Autonomy Must Deliver the Same Efficiency Regardless of the Driver" - Free Access - 07-09-2026 - Commercial Vehicles - Interview - Article Activate our intelligent search to find suitable subject content or patents. Select sections of text to find matching patents with Artificial Intelligence. powered by Select sections of text to find additional relevant content using AI-assisted search. With its AI-based world model, Inceptio Technology aims to bring fully automated SAE Level 4 trucks into series production. In an interview with ATZheavy duty, founder and CEO Julian Zheren Ma explains how the system works, what efficiency gains are possible, and which regulatory requirements are still missing. ATZheavy duty:Inceptio Technology is taking an AI-based approach to fully automated trucks (SAE L4) that relies on a "world model". How is this model structured? Ma: Our world model combines three closely interlinked functions: sensor synthesis, physics-aware dynamic simulation, and behavioral agent modeling of other road users. It generates realistic camera and lidar data, simulates the driving behavior of a heavy truck according to the laws of physics, and models the interactive reactions of other road users. These functions are integrated into a generative pipeline capable of reconstructing and expanding real-world driving scenarios, including complete lidar point clouds. The model is grounded in Inceptio’s Real-world Operation Scenario Library (ROSL), which provides statistically representative data from large-scale commercial operations. How is integration with electric mobility handled, and how do you coordinate range management? Our autonomy stack operates largely independently of the powertrain and is already in use on diesel, LNG, and electric platforms. For energy management, we have developed FEAD (Fuel Efficient Autonomous Driving). The system proactively optimizes speed, driving maneuvers, and behavior on inclines and declines. The strategy reduces energy losses and can be directly applied to electric vehicles – including range prediction, regenerative