Imec aims to develop a multiple sensor fusion platform for high speed, time critical and computation-intensive applications such as autonomous cars, smart homes and industrial robotics.
Our sensor fusion platform will combine data from multiple radars and multi-modal sensors (radars and cameras, for example) with pre-processing techniques and machine learning to detect and classify objects. The radar will recognize objects and patterns using a database of signatures (for bicycles and pedestrians, for example).
Our data fusion is based on deep learning principles. While it is critical for sensors to recognize certain patterns immediately, some of the non-time critical “learning” of sensors can happen through the feeding of information from the cloud and edge nodes of the network.