Infrastructure & Facilities
IT Infrastructure
BHT has secured substantial funding for computer science from BMFTR, BMWE, EU, and DFG FIP programs to establish a cluster for in-memory databases, large-scale data processing, and large GPU systems. In addition, through the BMWi-funded SOOFI project, BHT has access to more than 1,000 NVIDIA B200 GPUs for training the next generation of foundation models. This enables BHT to independently conduct experiments with large foundation models and agents in the field of machine learning.
The local infrastructure includes CPU nodes, each equipped with 1 TB of RAM, as well as 16 × 50 TB of storage for a distributed CephOS file system. Furthermore, the university operates a high-performance DELL SAN; GPU nodes based on NVIDIA B200, H200, A100, and V100 architectures; and corresponding InfiniBand networking technology with hot- and cold-aisle containment.
Similarly, HTW Berlin established a high-performance cluster for ML research and teaching (via BMBF funding). It includes a JupyterHub with A30-GPUs for fast prototyping and data exploration, a local LLM setup with H100-GPUs, as well as a SLURM-Cluster with A100-GPUs for multi-GPU training of models and dealing with larger data sets.
The Human.VR.Lab
Common infrastructural equipment for the applications (R1-R3)
The Human.VR.Lab, founded using 1.2 million euros of BMBF funding, is an open research laboratory under the umbrella of the BHT. Here, both researchers and interested parties can interact with VR/AR technologies and robotic systems, and experience, test, and investigate future human-centered technology in various (virtual) applications. The Human.VR.Lab consists of a motion tracking space (8x4m) with 8 workstations and a small assembly area, as well as a adjoining VR learning room where training courses and workshops can also take place. The following systems are part of the Human.VR.Lab and can be used by the project:
- Optical motion tracking system: 24 cameras for marker-less and marker-based tracking.
- Electromyography system: 16-channel for acquisition and analysis of muscle activities.
- FDM pressure measuring plate for gait analysis of humans and robots.
- Class sets of VR/AR technologies with newest VR HMDs for multi-user applications.
The Human.VR.Lab is complemented by VITALab.Mobile. This is a mobile VR/AR laboratory for case studies, field studies, and clinical trials to evaluate novel virtual therapies and scientific research. The lab is built into a truck that can reach a wide range of different user groups in a variety of locations. The focus of this interdisciplinary lab is to create a user-friendly and realistic research and interaction environment for diagnostics and therapy in a medical-therapeutic context.
The overall equipment of the Human.VR.Lab makes it possible to measure and analyze demonstration tasks. Scientists and PhD can work closely together, exchange ideas. The lab as a hub of interdisciplinary exchange brings stakeholders together and improves scientific outcome.
Robotics Infrastructure - House of Robotics
The House of Robotics at BHT brings together three complementary research laboratories: the Neurorobotics Research Laboratory (NRL), the Compliant Robotics Laboratory (CoRoLab), and the Soft Interactive Robotics Laboratory (SIRo Lab). Within this shared research environment, APPL-FM activities are conducted by CoRoLab and SIRo Lab, while the NRL complements the broader robotics expertise of the House of Robotics. Together, CoRoLab and SIRo Lab provide expertise in compliant locomotion, soft and dexterous manipulation, teleoperation, robot learning, and physical human–robot interaction. Their infrastructure includes advanced humanoid and mobile robots such as Booster K1, Digit, and TIAGo Pro, collaborative manipulators, and in-house visuo-haptic interfaces as well as the SIRo Soft Hand developed within Appl-FM. This infrastructure enables APPL-FM research on learning, planning, and executing complex behaviors on real robotic systems.





















