Cognitive Robotics

The NeuTouch ITN aims at improving artificial tactile systems, by training a new generation of researchers that study how human and animal’s tactile systems work, develop a new type of technology that is based on the same principles, and use this technology for building robots that can help humans in daily tasks and artific

In the project REBA+, funded within DFG priority program "Autonomous Learning", we develop, implement and evaluate rich extensions of a robot's body schema, along with learning algorithms that use these representations as strong priors in order to enable rapid and autonomous usage of tools and a flexible coping with novel mechanical linkages between the body, the grasped tool and target objects.

Which relations exist between properties of animals or people and their kinematic patterns? For example, can we tell, who performed a hand-over of which kind of object under which conditions just by looking at the sequence of joint angles? We try to find answers to these questions by employing a 3D motion tracking system.

The SARAFun project has been formed to enable a non-expert user to integrate a new bi-manual assembly task on a robot in less than a day. This will be accomplished by augmenting the robot with cutting edge sensory and cognitive abilities as well as reasoning abilities required to plan and execute an assembly task.

Together with the advances in building anthropomorphic robot hands, we are facing the question of how to dexterously control such complex robots with up to 20 degrees of freedom in up to five fingers and a wrist. Implementing fixed grasp and manipulation programs does not lead to satisfying results. In our work, we propagate a manifold representation of such movements recorded from human demonstration. The main idea is to construct manifolds embedded in the finger joint angle space which represent the subspace of hand postures associated with a specific manipulation movement.

The spatio-temporal contact pattern during manipulation is a valuable source of information about object identity and object state, especially in uncertain environments. Using a bimanual robot manipulator setup with two 256 "pixel" touch sensor arrays, the present project is creating a "haptic pattern database" and investigates machine learning techniques to analyse the information contents of different haptic features and to extract identity and state information from haptic patterns. A closely connected goal are dynamic control strategies for contact movements with deformable or plastic objects, such as clay.

Alignment is not restricted to support language, dialog functions and the execution of speech production. Similar mechanisms support the cooperative execution of more general actions. Besides the extension of alignment into the action domain, we hypothese that the formation of alignment is adaptive: Repetitions of actions facilitate the alignment and its adaptation is important for acquiring team expertise. In our opinion, adaptation and adaptive alignment paves the way for smooth and effective common acting.

While language provides us with a concise code capturing much of the movement complexity of our mouth, we still lack a comparable representation for the movement of our hands. This project aims to create a database of human hand interaction patterns from a variety of multimodal data sources. An associated goal is to develop methods for the clustering of captured trajectory data into physics-based models of manual interaction. We hope that the resulting database can make a contribution towards a better grounding of control strategies for anthropomorphic robot hands and develop for robotics a similar utility as the WordNet database has for linguistics.

Priming of relevant motor degrees of freedom to achieve rapid alignment of motor actions can be conceptualised as the rapid selection of low-dimensional action manifolds that capture the essential motor degrees of freedom. The present project investigates the construction of such manifolds from training data and how observed action trajectories can be decomposed into traversals of manifolds from a previously acquired repertoire. To this end we focus on manual actions of an anthropomorphic hand and combine Unsupervised Kernel Regression (UKR, a recent statistical learning method) with Competitive Layer Models (CLM, a recurrent neural network architecture) to solve the tasks of manifold construction and dynamic action segmentation.

What principles enable rapid and adaptive alignment in coordination?

This project investigates Gestalt principles and their generalization from the perceptual into the action/cooperation domain for modeling adaptive alignment and its functional replication in human-robot cooperation. Departing from learning algorithms for dynamic Gestalt formation in layered recurrent networks (Competitive Layer Model CLM), we develop a hybrid, hierarchical architecture for adaptive alignment in cooperation that integrates elements from connectionist and symbol-based representations. We evaluate its performance in a human-robot cooperation scenario involving two anthropomorphic hands mounted on a bimanual robot platform.