A universal spectral imaging sensor platform for industry, agriculture and autonomous driving

The HyperImage project is the first initiative to develop a universal, fast, modular and cost-effective spectral image sensing technology platform suitable for both short- and long-range imaging applications. 

HyperImage will make use of innovative photonic components including electrically tunable liquid lenses, pixel shifters, fast steering mirrors, and configurable SMD style IR micro-emitter arrays and it will integrate these components into innovative, high-performance multi- and hyperspectral snapshot and line scan cameras.

AI machine learning algorithms will translate spectral image data into relevant functional products properties and detect and classify objects for more accurate decision making in these different industrial use cases:

      • Quality control in manufacturing of high power electronics;
      • Crop growth monitoring for fully automated vertical farming of salads, herbs and microgreens;
      • Spectral image based vision and navigation in off-road autonomous driving; 
      • Light-weight, high-resolution hyperspectral vision system for unmanned geo-surveillance drones.

The technology platform will be complemented with a cloud-based spectral image analysis platform and reference data repository that enables users to continuously improve image analysis accuracy and prediction models.


Robotnik is responsible for the off-road autonomous driving scenario. We will provide the RB-CAR vehicle to validate spectral-imaging based computer vision for off-road autonomous driving for search & rescue and infrastructure surveillance use-cases. 

Today, autonomous driving is focused to work on public roads where the environment is more or less uniform and controlled in terms of infrastructure (terrain level, definition of curves, the road cleaning…). However, in off-road scenarios the challenges are even greater. Not only other driving vehicles must be taken into account, also the level of the terrain, the vegetation around the way and the possible obstacles. Nowadays, no autonomous vehicles in the market offer this functionality, because state-of-the-art image and distance sensor solutions (RGB cameras, LIDAR, RADAR) are not able to distinguish between crossable and non-crossable obstacles and holes. 

Hyperspectral cameras have not been used before towards enabling navigation of the autonomous vehicles. With this technology, at the end of the project, the RB-CAR from Robotnik will be able to:

      • Detect terrain holes and gaps and avoid them in autonomous mode.
      • Enabling adaptability of the velocity of the vehicle to the inclination detected. (Move in autonomous mode a slope of 5% up and down);
      • Identify and differentiate crossable plants from trees that can be a serious obstacle. (Cross plants during autonomous navigation and avoid 2 or more trees).

In summary, within the context of the HYPERIMAGE project, Robotnik will be able to perform autonomous navigation in unknown terrains (e.g. no marked pathways) at all weather conditions detecting holes, distinguishing crossable and non-crossable plants to showcasing and moving up to 15 km/h full-autonomous operation.

mobile robot

Project partners