We describe the design concepts of the modular humanoid robot Myon, which can be disassembled and reassembled during runtime. The body parts are fully autonomous in a threefold sense: they all possess their own energy supply, processing power, and a neural network topology which allows for stand-alone operation of single limbs. The robot has especially been designed for robustness and easy maintenance. It exhibits a combination of an endoskeleton with an exoskeleton, the latter of which can manually be detached without the need for technical equipment. One of the essential parts is a novel flange which firmly connects the body parts mechanically, whilst at the same time relaying the power supply lines and sensorimotor signals. We also address the details of the antagonistic and compliant actuation system which not only protects the gears against high impact forces but also enables biologically inspired joint control.

No abstract

This paper concerns the construction and function of a test foot with a passive toe joint. This was developed within the scope of the EU project ALEAR (Artificial Language Evolution on Autonomous Robots) together with a test leg for the humanoid robot "M-Series" to test different function principles. In particular the sensor types are exemplified which are used in this foot. Further, the sensor principles, the electronic subsequent treatment of the sensor signals as well as their provision for the connected bus system are explained. Concluding, the experiences which were won with the present test construction are highlighted. The outlook is given on the planned foot which should be used in the humanoid "M-Series" robot.

Recovering balance from unknown disturbances can be considered a complex sensorimotor ability of humanoid robots. We present so-called Cognitive Sensorimotor Loops (CSLs), discuss their properties and show how they can be used for motion generation and balance recovery on robots. Their behavioral abilities will be demonstrated on a single robot leg controlled by CSLs and we will show that a complex stand-up motion can emerge from the interplay of independent joint controllers. Furthermore, we explain how CSLs can be used to help a robot to adapt to changing slopes and recover balance after disturbance.

No abstract

Humanoid robots are complex systems that require considerable processing power. This applies both for lowlevel sensorimotor loops, as well as for image processing and higher level deliberative algorithms. We present the distributed architecture DISTAL which is able to provide the processing power of large neural networks without relying on a central processor. The architecture successfully copes with runtime-metamorphoses of modular robots, such as the humanoid robot MYON, the body parts of which can be detached and reattached during runtime. We detail the implementation of DISTAL on 32-bit ARM RISC processors, describe the underlying neural byte-code (NBC) of neurons and synapses, and also depict the graphical application software BRAINDESIGNER which releases the user from program coding.

No abstract