Manipulation Lab at KIT-H2T

The humanoid robot ARMAR-6 demonstrates its cognitive and sensorimotor capabilities in the context of autonomous handling of unknown objects using an exemplary decontamination scenario. In this scenario, plant components from a collection of unknown objects in a box are recognized, grasped and visually measured. For this purpose, the robot uses its visual and haptic perception as well as its context-specific domain knowledge. In a next step, the objects are to be cleaned and finally deposited in a second box. The robot operates in a domain with an area of more than 30m². In addition, there is a control station for controlling and monitoring the robot and an area for the automatic decontamination process.

ARMAR-6 during mobile manipulation for decontamination of radioactive plant components © KIT

ARMAR-6 is a humanoid robot that interacts with and proactively assists humans. It is designed to assist technicians with maintenance and repair tasks in industrial environments. With its human-like body, it can use tools created for humans, such as drills or hammers. Its artificial intelligence allows it to independently take on tasks in the context of industrial equipment maintenance, recognize the need for help from a human cooperation partner and offer its assistance. A major challenge here is that ARMAR-6 is confronted with lots of possible situations that cannot be foreseen at the time of its programming. In ROBDEKON, the robot serves as a demonstrator to evaluate methods for decontaminating plant components.

Precise gripping operation with ARMAR-6. During the execution of gripping and manipulation tasks, the operation is continuously checked for success and errors. © KIT

The KIT-H2Ts manipulation lab offers:

  • Humanoid robot as demonstrator for tasks in human-hostile environments
  • Autonomous capabilities of grasping and mobile manipulation
  • Learning from human observation and own experience
  • Intuitive interaction and cooperation with humans
  • Holistic architecture to integrate methods of AI, computer science and mechatronics