Tilburg Robotics :: Dexterous Manipulation & Deep Reinforcement Learning (DRL)

Tilburg Robotics is a small Dutch robotics company (closely affiliated with Tilburg University’s AI for Robotics Lab) that develops low-cost, open-source robotic hardware for research, particularly in dexterous manipulation and Deep Reinforcement Learning (DRL). The company focuses on making advanced robotic platforms more accessible to research groups and developers who cannot afford expensive commercial solutions like the Shadow Hand or Allegro Hand.

Their main product is the Tilburg Hand, a 16 degrees-of-freedom (DoF) dexterous robotic hand designed specifically for research in embodied AI and manipulation tasks. It uses 16 Dynamixel XL330-M288 smart servos, industrial 3D-printed parts (Multi Jet Fusion), and a simple USB interface. The hand is roughly twice the size of a human hand (palm ~13x10cm), weighs under 1kg, and provides position, velocity, and current-based torque feedback. It is available as a fully assembled and tested unit for approximately €5,000 (incl. EU VAT) or €4,132 (excl. VAT for international buyers). A pair (left + right) is offered at a discount.

Key features include:

• Open-source software (Python library with planned C++ support) and documentation available on GitHub.
• URDF and MuJoCo models for simulation.
• Easy integration with ROS and camera-based hand pose tracking.
• Customizable fingertips and wrist interfaces for different robot arms.
• Strong focus on affordability to democratize dexterous robotics research.

The Tilburg Hand is positioned as a practical, cost-effective platform for studying complex manipulation, grasping, and reinforcement learning. The company also encourages community contributions by hosting open-source code and trained models from users. While small (effectively a 1-person operation or very small team), it combines academic research expertise with practical hardware development.