InnoHPC - Success story West University of Timisoara and Continental Automotive Romania

09-10-2019

Continental Automotive Romania - manufactures and markets electronic automotive safety systems for the automotive industry in Romania. The company was formerly known as Siemens VDO Romania SRL. The company was founded in 2000 and is based in Timisoara, Romania. Continental Automotive Romania s.r.l. operates as a subsidiary of Continental Automotive GmbH.

Business challenge that has potential for benefiting from HPC was to calculate key performance indicators for eye and face gaze in order to detect the attention and drowsiness of the driver for reducing the number of accidents.

Continental Automotive Romania: Calculate key performance indicators for eye and face gaze in order to detect the attention and drowsiness of the driver for reducing the number of accidents, responsible – Mihail Gaianu.

The proposed solution of Continental Automotive Romania in terms of enable HPC is to asure HPC infrastructure for data processing and the necessary hardware storage. On the other hand one of the needs was to assure specialists for labeling procedures of obtained images form car interior camera for the proposed pilot project. To be more explicit Interior Cameras can be used for Driver Monitoring or Gesture Recognition. To understand what the driver exactly needs is crucial to offer a real Holistic Human-Machine Interface experience. Therefore, our Interior Camera gathers the current condition of drivers. It can be used for Driver Monitoring or Gesture Recognition (Figure 1). Driver Monitoring enables gaze target recognition, face recognition or drowsiness classification. Gesture Recognition provides for a natural interaction and joy of use. Standard swipe gestures or custom specific gestures can be implemented.

Technical Information:

  • Driver monitoring & gesture recognition
  • Operates in complete darkness by using infrared illumination (wavelength 850 or 940 nm)
  • Automatic extrinsic calibration
  • Driver Monitoring: Scalable from mono to multi-camera system
  • Gesture control
  • Minimizes driver distraction - meant that drivers had to take their hands off the steering wheel or take their eyes off the road. This also prevents the driver from unintentionally starting gesture-based control by means of their usual everyday gestures, and thus making unwanted selections.


In terms of the proposed project description and needs are mandatory almost 3 years for testing, obtaining accurate algorithms in terms of face-recognition of drivers, involving more then 10 specialists from automotive industry, and 10 specialists from academia to obtain expected outcomes of the proposed project.

The future steps for proposed project from Continental Automotive are the following: Need of Petabyte, Expandable to 8.6PB, Raw Data Storage, in a Single Rack.

From the perspective of InnoHPC Lab testing the following tested steps were performed: First contact was established via InnoHPC project events and communication was ongoing since then. SME representative registered at InnoHPC LAB portal, being able to access all the materials available. SME got in contact with the LP of the project and received the consultancy about their business problem and some tentative conclusions were drawn along with proposal on how to continue the work.

During the process an issue of IPR was raised, both parties considered to sign the NDA that was available at InnoHPC LAB platform but in the end the NDA was not signed mainly due to the level of cooperation that was established in the end. In case the business problem would be developed and elaborated further on, both parties agreed the NDA would have been in place. 

Programme co-funded by European Union funds (ERDF, IPA, ENI)