During the project, we will build an experimental road vehicle platform, the central element of which is an attractive mid-range electric vehicle equipped with a sensor and control system. Tasks of the vehicle platform: road data collection of vehicle and environmental data, design and execution of experiments on test track and road, demonstration of functions on test track and road without intervention. With this system, we can enter the development areas of autonomous vehicle control without delay and respond directly to innovation needs.

The main topics of the research area: research related to sensor fusion, global and local route planners, research related to communication architectures:

  • Research on machine learning methods of environmental perception. Sensor fusion of radar, lidar and camera using classical and machine learning methods. Research on situation assessment methods.
  • Research on route planning algorithms for autonomous vehicles and mobile robots. Research on local procedures that take into account both global and dynamic objects, with special regard to the balance between optimization and runtime.
  • Start research on vehicle-vehicle and vehicle-infrastructure communication (V2X). Explore data collection opportunities in V2X and 5G networks.

Further development of the existing vehicle platforms at Széchenyi University:

  • Expansion of the sensor system of the converted passenger car. Development of possibilities for the further development of the central data collection and processing computer. Assessing the communication, data storage and computing needs of machine learning and the expanded sensor set, and developing proposals taking into account current state-of-art technologies.
  • Fine-tuning vehicle platform functions in a relevant test environment.

aa

Other research and development tasks:

  • Creation of Lanelet2 maps containing global routes (modeling of Lanelet2 at the Zala University Track).
  • Commissioning of the LIO SAM (Tightly-coupled LIDAR Inertial Odometry via Smoothing and Mapping) procedure on the Nissan Leaf vehicle and testing at the SZE Campus.
  • Pavement segmentation artificial neural network replacement.
  • Optimization of artificial neural networks by Bayes method.
  • Introduction of a point cloud-based object detection artificial neural network.
  • Replacing an artificial neural network with a camera image-based object detector.
  • Application of TensorRT based networks.
  • Bench recognition and free space algorithm. New path recognition algorithm. Blind spot filter on both sides of the car.