DeepFusion AI is a perceptive sensor fusion system company that integrates 4D imaging radar, cameras, and LiDAR using advanced deep learning technology, providing solutions to various industries.

​

Efficient sensor fusion of optical and radio-frequency sensors has always been emphasized in autonomous driving perception. While deep learning for optical sensors (cameras, LiDAR) is becoming widespread, deep learning for radio-frequency sensors (4D imaging radar, X-band radar) remains limited due to the lack of a standardized dataset.

​

To solve this, we developed a virtual 4D imaging radar environment to build a pre-trained model. This model achieved accuracy comparable to LiDAR and has been successfully deployed in actual vehicles.

​

We fine-tuned the pre-trained model by capturing additional datasets from 4D imaging radars installed on actual vehicles, creating a 4D imaging radar deep learning model. We have completed and announced the world's only commercialized model, demonstrating accuracy over 40% higher than competitors.

​

We are leveraging this technology to apply it to robot taxis, unmanned surface vehicles (USVs), and unmanned ground vehicles (UGVs), establishing it as a core perception sensor fusion system.