Generative methods for perception, prediction and planning. A paradigm shift in the development of autonomous driving functions.
A paradigm shift in the development of autonomous driving functions.
The project nxtAIM – NXT GEN AI METHODS leverages advancements in the development of generative AI methods and will, for the first time introduce bidirectional information flow in the chain of effects – a paradigm shift in development that promises massive improvements in automated driving.
through an inexhaustible reservoir of data for offline testing, validation, training, and online error detection.
through the ability to deconstruct and recombine semantic information and extend the Operational Design Domain (ODD) by targeted scenario and sensor data generation.
through online verification, plausibility checks of individual processing steps in in the chain of effects during operation and understanding of latent representations.
The nxtAIM project aims to realise pioneering AI developments in the field of autonomous driving over the next three years.
Insight
nxtAIM uses generative learning methods to enhance simulated road scenes photorealistically, thereby creating a larger dataset for training and validating semantic segmentation networks.
One sees a street scene from the ego perspective of a vehicle. This sequence was created using the CARLA simulation tool. The environment and the sequence of the scene can be defined. Image annotations are generated automatically. In nxtAIM research, simulated images from the synthetic domain are adapted to the domain of the real world
News
Consortium
The joint project NXT GEN AI METHODS – nxtAIM, launched at the beginning of 2024 and initiated by the VDA Leitinitiative Autonomous and Connected Driving, will conduct joint research over the next three years. 20 project partners are involved in the project, which is being funded by the Federal Ministry for Economic Affairs and Climate Action with 27 million euros. The partners of the consortium include OEMs, suppliers, technology providers, and research bodies.
