Besides driving innovation behind the modern technologies, semiconductor IPs perform an important
role of accelerating conventional processing tasks and thus essentially enabling the modern ecosystem of
electronic devices and systems [6]. This kind of technology is the basis for all modern trends, like,
artificial intelligence, edge and fog computing, smart cities, autonomous driving, Industry4.0, precision
agriculture, etc.
Our team has multiple silicon IP developments at different TRLs (Technology Readiness Levels). During
the SilHouse project we will concentrate on the most “ready” core – lens distortion correction and
general image transformation. Other cores will be presented briefly just to outline the long-term strategy
and ambition of our team.
1) Lens distortion correction and general image transformation IP core
Challenge. Many modern applications such as autonomous driving, cooperative industrial robotics,
unmanned aerial vehicles have challenging requirements concerning the responsiveness and power
consumption. This IP core addresses this by migrating frequently applied image processing tasks from
software to the silicon hardware, essentially offloading processors, improving responsiveness and
reducing power consumption.
Technology. Virtually every image processing task starts with correction procedure for distortions which
are introduced by lenses, then images are often used for fusion or reconstruction purposes, which means
that they must be aligned with the virtual coordinate system or with other sensors. This is achieved by
performing image transformation. These are the exact tasks performed by the EDI lens distortion
correction and general image transformation IP core.
2) Occlusion attentive stereo-vision IP core (follow up product in the design house portfolio).
The second core considered for commercialization (most probably not during the SilHouse project) is our
stereo-vision IP core which obtains 3D information from a pair of stereo-images via
pixel-correspondence algorithms. This IP core can be used in any use-case where the system must perceive
the 3D environment, but it is especially effective in UAV use-case due to high demands for energy efficiency.
3) High dynamic range optical flow sensor IP.
This technology is currently at a very low technology readiness level and is not considered for
commercialization in SilHouse project. Nevertheless, it highlights the general direction of our team. The
optical flow sensors are very good candidates for the realization of very demanding collision avoidance
and localization algorithms, especially suitable for indoor UAVs (Unmanned Aerial Vehicle). At the
moment the optical flow sensors have a very low dynamic range, which means that the same sensor
cannot be used at different distances from the viewed plane. EDI team already does research in this
direction in H2020 ECSEL project – COMP4DRONES “Framework of key enabling technologies for
safe and autonomous drones’ applications” ( https://www.comp4drones.eu/ ).