We have already successfully carried out numerous research projects and are currently working on future-oriented projects. Here we provide an overview of current research projects and their project objectives as well as our tasks as DH electronics.

1. Reinforcement learning-based traffic optimisation and impact and benefit assessment for urban regions
   The project focuses on inner-city traffic optimization via reinforcement learning (RL). In order to create the necessary data basis for RL-based optimisation and further applications in the field of the smart city, on the one hand a self-sufficient sensor solution is being developed that is capable of comprehensively recording conditions at traffic junctions and, on the other hand, a cloud-based solution for data use is being developed that forms the basis for data management, standardized interfaces, simulation and machine learning (ML). In the project, a realistic traffic simulation of a city in Bavaria will be extended and an AI-based prediction for traffic models will be developed, which will form the basis for RL-based traffic optimization. The aim is to achieve a significant improvement in traffic flow for the road users under consideration on the basis of innovative algorithms and novel approaches. Furthermore, benefit and consequence assessments are to be made possible, which predict the technological change (e.g. autonomous vehicles) with regard to future traffic potential, as well as address potentials of a smart city. Finally, a test run of the sensor solution and optimization algorithms in the city's real traffic system is planned.
   
   In this project, DH electronics is responsible as a system integrator for the overall solution, which is made up of the components of the individual project participants and in particular the sensor system. DH electronics contributes comprehensive know-how in embedded systems to the project and offers the suitable platform with a system on module. In addition, DH has the opportunity to gain initial experience in the area of "self-sufficient energy supply" using photovoltaics and battery technology as storage.
    
2. Advanced Deep-Learning System Control

   The efficient use of energy is an essential prerequisite for the economic operation of plants and machinery. A large savings potential lies in particular in the operational optimization of refrigeration systems. However, this potential is only insufficiently utilised because the operating state of refrigeration systems cannot usually be fully recorded due to the complex thermodynamic relationships. 

   The aim of the project is to develop an innovative method for the optimal control of incompletely known systems or processes and to implement it in an edge controller. The approach is based on deep learning technology and is being tested using commercial refrigeration systems as an example. The controller has a new type of hardware that optimally supports the neural algorithm.