A new versatile hydraulically powered quadruped robot ( HyQ ) has been developed to serve as a platform to study not only highly dynamic motions, such as running and jumping, but also careful navigation over very rough terrain. HyQ stands 1 m tall, weighs roughly 90 kg, and features 12 torque-controlled joints powered by a combination of hydraulic and electric actuators. The hydraulic actuation permits the robot to perform powerful and dynamic motions that are hard to achieve with more traditional electrically actuated robots. This paper describes design and specifications of the robot and presents details on the hardware of the quadruped platform, such as the mechanical design of the four articulated legs and of the torso frame, and the configuration of the hydraulic power system. Results from the first walking experiments are presented, along with test studies using a previously built prototype leg.

We presented and evaluated an approach based on HMM, GMR, and dynamical systems to allow robots to acquire new skills by imitation. Using HMM allowed us to get rid of the explicit time dependency that was considered in our previous work [12], by encapsulating precedence information within the statistical representation. In the context of separated learning and reproduction processes, this novel formulation was systematically evaluated with respect to our previous approach, LWR [20], LWPR [21], and DMPs [13]. We finally presented applications on different kinds of robots to highlight the flexibility of the proposed approach in three different learning by imitation scenarios.

Variable stiffness actuators (VSAs) are complex mechatronic devices that are developed to build passively compliant, robust, and dexterous robots. Numerous different hardware designs have been developed in the past two decades to address various demands on their functionality. This review paper gives a guide to the design process from the analysis of the desired tasks identifying the relevant attributes and their influence on the selection of different components such as motors, sensors, and springs. The influence on the performance of different principles to generate the passive compliance and the variation of the stiffness are investigated. Furthermore, the design contradictions during the engineering process are explained in order to find the best suiting solution for the given purpose. With this in mind, the topics of output power, potential energy capacity, stiffness range, efficiency, and accuracy are discussed. Finally, the dependencies of control, models, sensor setup, and sensor quality are addressed.

Robots are becoming safe and smart enough to work alongside people not only on manufacturing production lines, but also in spaces such as houses, museums, or hospitals. This can be significantly exploited in situations in which a human needs the help of another person to perform a task, because a robot may take the role of the helper. In this sense, a human and the robotic assistant may cooperatively carry out a variety of tasks, therefore requiring the robot to communicate with the person, understand his/her needs, and behave accordingly. To achieve this, we propose a framework for a user to teach a robot collaborative skills from demonstrations. We mainly focus on tasks involving physical contact with the user, in which not only position, but also force sensing and compliance become highly relevant. Specifically, we present an approach that combines probabilistic learning, dynamical systems, and stiffness estimation to encode the robot behavior along the task. Our method allows a robot to learn not only trajectory following skills, but also impedance behaviors. To show the functionality and flexibility of our approach, two different testbeds are used: a transportation task and a collaborative table assembly.