Programmable Systems for Intelligence in Automobiles
Period: May 2018 – April 2021
Total budget: ~51.55M Eiro
Project Leader: Herbert Roedig, Infineon Technologies AG
Partners: 60
Countries (14): Germany, Italy, Netherlands, Israel, Austria, Spain, Turkey, Greece, Finland, Romania, Lithuania, Belgium, Latvia, Sweden.
The ambition of PRYSTINE is to strengthen and to extend traditional core competencies of the European industry, research and universities in smart mobility and in particular the electronic component and systems and cyber-physical systems domains. PRYSTINE’s target is to realize Fail-operational Urban Surround perceptION (FUSION) which is based on robust Radar and LiDAR sensor fusion and control functions in order to enable safe automated driving in urban and rural environments. Therefore, PRYSTINE’s high-level goals are:
1. Enhanced reliability and performance, reduced cost and power of FUSION components
2. Dependable embedded control by co-integration of signal processing and AI approaches for FUSION
3. Optimized E/E architecture enabling FUSION-based automated vehicles
4. Fail-operational systems for urban and rural environments based on FUSION
PRYSTINE will deliver (a) fail-operational sensor-fusion framework on component level, (b) dependable embedded E/ architectures, and (c) safety compliant integration of Artificial Intelligence (AI) approaches for object recognition, scene understanding, and decision making within automotive applications. The resulting reference FUSION hardware/software architectures and reliable components for autonomous systems will be validated in in 22 industrial demonstrators, such as:
1. Fail-operational autonomous driving platform
2. An electrical and highly automated commercial truck equipped with new FUSION components (such as LiDAR, Radar,
camera systems, safety controllers) for advanced perception
3. Highly connected passenger car anticipating traffic situations
4. Sensor fusion in human-machine interfaces for fail-operational control transition in highly automated vehicles
PRYSTINE’s well-balanced, value chain oriented consortium, is composed of 60 project partners from 14 different European
and non-European countries, including leading automotive OEMs, semiconductor companies, technology partners, and
research institutes.
EDI will develop and implement (in hybrid CPU/SoC/GPU) advanced AI based fail-operational algorithms for sensor data (e.g. LiDAR, stereo camera, radar) fusion, object detection and classification, motion prediction and decision making. Special emphasis will be placed on enhanced precision, reliability, safety and reduced costs. The sensor data will be processed using deep learning approach. The convolutional neural networks will be implemented and trained for object detection in video and LiDAR data. The capabilities of neural networks to fuse different kinds of sensors will be further researched and used in the proposed system. In order to train the machine learning models, EDI will adapt existing datasets and create additional labeled data. Scenarios with faulty sensors will be included in the training of detection models making the system fail-operational. EDI will prove concept of variable data acquisition approach by developing and validating VDA unit. This will lead to optimized and fail-safe system operation, due to computational resource redistribution. This will be done by analyzing vehicles motion dynamics and environment state, in such a way defining the most appropriable data source for sampling. Besides that, EDI will also analyze and use different approaches for efficient implementation and use of available resources in selected embedded platform.
More informtion: www.prystine.eu
Participating scientists
