DC7. Data fusion of traffic, behaviour & infrastructure for holistic driver assistance

Aristotelis Tsoutsanis

Education
M.Sc. in Computer Science: Computer Graphics and Vision, Machine Learning and Analytics, Robotics from Technical University of Munich
About
Experienced Machine Learning Engineer specializing in Computer Vision and Deep Learning. At PreciTaste in Munich, contributed to developing and enhancing deep learning pipelines for real-time vision AI systems in the fast food industry, focusing on data preprocessing, model training, and deployment. As a contributor to Smartfield, the world’s first AI-managed autonomous crop field, developed at Technical University of Munich, implemented advanced yield forecast models that optimized fertilizer and growth regulator applications, achieving precise predictions and improved yield outcomes.
Passionate about leveraging vision-based technologies to drive innovative solutions across industries.

Hosts: NTUA & OSEVEN

Objectives:

  • To exploit multi-parametric data for the creation of a holistic AI framework for road safety-related driver evaluations
  • To define appropriate traffic, behaviour and infrastructure parameters or other KPIs to be collected and used in the models
  • To create new AI algorithms harmonising the selected parameters to comparable datasets
  • To integrate the developed AI algorithms to telematics-based applications focusing on driver assistance and support

Expected results:

  • New knowledge on integration and harmonisation of traffic, behaviour and infrastructure big data parameters
  • A functional AI framework incorporating these elements based on driver telematics with transferability evaluations, that will lead to the development of new and seamless road safety solutions
  • A case-study utilising driver telematics in an urban area, with actionable results, compatible with the vision and activities of OSeven.
  • The related framework can be used to develop an app providing personalised recommendations to drivers based on their trip/route habits

Planned secondment(s):

  • RHDHV-NL, Purpose: to test the created holistic algorithm in real conditions and new networks and explore connectivity with existing Dutch Talking Traffic ITS-chain systems