Current state-of-the-art approaches in Artificial Intelligence (AI) require large amounts of training data to solve industrial challenges. It is worth mentioning that the performance is highly dependent on the size and quality of the training data. While large datasets of real-world training data exist or can easily be generated for some problem domains, in robotics data is particularly hard to acquire due to the needed effort, time and hardware resources, or due to safety precautions.

To this end, the TUM team under the AI4DI project is developing simulation environments in which AI agents can be virtually-trained to control robots to execute industry-relevant challenges. To provide a high level of realism in simulated physics and rendering, we use high-end physics simulators and game development engines. In the video, we demonstrate how the virtual environment can be used to train an AI agent for solving a robotic manipulation task using reinforcement learning.

The benefits of fast and safe training in simulation come with the drawback of limited transferability of the simulation-trained AI agents to control real robots. This is still a major challenge due to the mismatch between the simulation and the real-world environment. In the next step of our research, we are working on approaches that can reduce the effects of the so-called reality gap and improve the transferability to real robots.

  • AI4DI Industry Sectors and Applications

    AI4DI's mission is bringing AI from the cloud to the edge and making Europe a leader in silicon-born AI by advancing Moore's law and accelerating edge processing adoption in different industries through reference demonstrators.
  • Automotive

    • AI-based logistics solutions
    • Assembly Process Optimisation
    • Autonomous Reconfigurable Battery Systems
    • Virtual AI Training Platforms
    • Autonomous Mobile Robotic Agents
    • Predictive Maintenance via Digital Twins
  • Semiconductor

    • AI-based Failure Modes and Effects Analysis Assistants
    • Neural Networks for predicting critical 3D Dimensions in MEMS inertial Sensors
    • Machine Vision Systems developed in the Wafer Inspection Production Line
    • Semiconductor Wafer Fault Classifications
    • Automatic Inspection of Scanning Electron Microscope Cross-Section Images for Technology Verification
    • Anomaly Detection on Wire Bond Process Trace Data
    • Optical Inspection
    • Wood Machinery with the Perception of the Surrounding Environment
    • Intelligent Robot Applications
  • Food and Beverage

    • AI-based environmental monitoring
    • Autonomous Environment-Aware Quality Control Systems for Champagne Production
    • Production Process Optimisation and Predictive Maintenance for Soybeans Manufacturing
  • Transportation

    • Mobility-As-a-Service Development of AI-based Fleet Management for Supporting Multimodal Transport.

Make Europe the leader in Silicon-Born-AI for accelerated edge processing

Accelerate the Artificial Intelligence (AI) adaptation to serve European priorities in the digitizing of the industry
Maximize the benefits of Moore’s Law and More Moore, and revive Moore’s Law beyond the current technology
Deployment plan showing how to develop and valorise the AI technology
Build AI community in Europe which is complementary with other initiatives
Build and sustain dynamic AI technology an ecosystem in Europe ensuring ethical clean an trusted AI for safety critical real time applications
Project Partners
Month Duration
M Funding
Upcoming Publications