ESA Open Invitation to Tender AO9976
Open Date: 26/07/2019
Closing Date: 27/09/2019 13:00:00

Status: ISSUED
Reference Nr.: 19.1ED.08
Prog. Ref.: Technology Developme
Budget Ref.: E/0901-01 – Technology Developme
Special Prov.: BE+DK+FR+DE+IT+NL+ES+SE+CH+GB+IE+AT+NO+FI+PT+GR+LU+CZ+RO+PL+EE+HU
Tender Type: C
Price Range: 200-500 KEURO
Products: Satellites & Probes / On-board Data Management / On Board Data Management / Payload Data Handling Units
Technology Domains: Onboard Data Systems / Payload Data Processing / System Technologies for Payload Data Processing / Onboard Data Systems / Onboard Data Management / System / Space System Software / Earth Observation Payload Data Exploitation / Data and Information Processing and Exploitation / Space System Software / Earth Observation Payload Data Exploitation / Applications and Services
Establishment: ESTEC
Directorate: Directorate of Tech, Eng. & Quality
Department: Electrical Department
Division: Data Syst & Microelectronics Division
Contract Officer: Casini, Gian Lorenzo
Industrial Policy Measure: N/A – Not apply
Last Update Date: 26/07/2019
Update Reason: Tender issue

This activity will analyse if current state-of-the-art information extraction of payload data (hyper- and multispectral visual and infrared imaging; etc.) can be applied in embedded systems to enable new methods in data selection optimisation, advanced autonomy, acquisition optimisation and in general to increase quality (in terms of content) and quantity of acquisition. Examples of information extraction applications include: land area type designation; detection of objects for tracking such as ship and airplane detection over ocean areas and natural disaster detection.The recent developments of Artificial deep Neural Network (ANN)-based algorithms for terrestrial applications (such as convolutional neural networks), have shown a significant improvement in information extraction algorithms, such as image classification. The suitability of using such algorithms for on-board data for payload imaging data information extraction and object classification shall be analysed to enable information extraction to levels not possible with traditional approaches.Tasks:- Methods for how algorithms can be executed on space hardware shall be analysed- Evaluation of how existing models and frameworks could be adapted- The potential need for extending available algorithms to multi- and hyper-spectral imagers typical for space applications, should be addressed.- Promising applications selected based on the availability and maturity of the algorithms as well as on the bases of their applicability on board.- Trade-off and selection of an enabling information extraction on-board processing chain shall be done.- For ANN, the necessary hardware accelerators shall be identified as well as the partitioning of tasks between hardware and software and the overall approach for training the network.- Priority schemes for further processing of the images (e.g. Selection of region of interest and image compression) and downlinking, based on the information extracted shall also be included in the activity including their impact on payload data operational aspects.- Selected chain and algorithm to be bread-boarded, evaluated and benchmarked for demonstration.