Former Research Assistant

Current Project

Robots get deployed in more and more applications. The robots used can have very different structures, reaching from 3-DoF serial chain manipulators to complex humanoid systems. Their capabilities can vary strongly as well. If several robots are available to perform a task, the question arises, which one is able to perform the task best. Of course, different aspects can play an important role as, e.g., perceptual capabilities or manipulation capabilities provided by the structure of the robot.

The focus of my work is the analysis of manipulator structures in task-specific contexts. The analysis is performed with respect to efficient control. In an ideal case, the joints of the manipulator should not only move slowly and smoothly, but the workload should also be distributed equally among all joints. It is important, that the analysis is applicable to any arbitrary manipulator, independently of its number of DoF. It has to be possible to analyze a manipulator with one or multiple broken joint(s) as well. Hence, we provide an inter- and intra-robot comparison with respect to efficient control (Petsch and Burschka, IEEE ICRA 2013). An exemplary scene is depicted in the Figure below (Petsch and Burschka, IEEE ICRA 2013).

The analysis is built on the abstract task representation, which has been developed in previous work (see past project below). The abstract representation of the robot's capabilities and the abstract representation of the task allow an advanced analysis in contrast to, e.g., a comparison of task execution times along pre-defined paths.

In order to enable such an analysis of manipulator structures, an advanced estimation of inverse kinematics is required. It has to be possible to estimate the inverse kinematics for any arbitrary serial chain manipulator at a single point and along entire paths. Moreover, the inverse kinematics for human-like robotic hands should be estimated as well. The required method needs to be able to overcome local minima and, furthermore, it has to work without any pre-knowledge about possible solutions regarding, e.g., advantageous initial configurations, incremental methods or partial derivatives. If several solutions exist, they have to be found. If no solution is present, the configuration which moves the robot's end-effector closest to the desired position should be returned. We provided such an estimation of inverse kinematics of arbitrary serial chain manipulators and human-like robotic hands (Petsch and Burschka, IEEE/ASME AIM 2013).

We extended also our previous representation of manipulation-relevant object properties to dexterous manipulations. Such manipulations require a much more detailed knowledge. Most of the times, the aim of a manipulation is a defined goal state of the objects in the environment. For example, objects could be placed on each other in a certain manner. A more complex example are threads, which are supposed to be moved in a manner, such that a knot is tied. Hence, we developed a representation which focuses on the relations of objects in the environment (Petsch and Burschka, IEEE/ASME AIM 2013).

Moreover, the previously developed abstract task representation was utilized in the context of path optimization. The abstract representation consists of the characteristic properties of the task. These properties need to be considered for the choice of the path. However, the final choice of a path is not limited to one pre-defined path. It can be chosen from a set of paths fulfilling the characteristic manipulation properties. The resulting freedom can be utilized for path optimization with respect to efficient control. The joints of the manipulator should move slowly and smoothly. Furthermore, the workload should be distributed equally among all joints. The Elastic Power Path concept has been developed for such a path optimization in the context of abstractly represented tasks (Petsch and Burschka, IEEE ICRA 2013). An scene with exemplary path alternatives is shown in the Figure below (Petsch and Burschka, IEEE ICRA 2013).

Past Project

In our daily life we are usually grasping objects very often. Most of the times we are not even thinking about questions like "How can I grasp an object?" or "What can I do with a cup?". But if we transfer the task of grasping an object to a robot, its complexity becomes clear.

My work within the EU-Project "Emergence of Cognitive Grasping through Introspection, Emulation and Surprise" (GRASP) dealt with the acquisition of manipulation-relevant object knowledge.

The robot observes a human demonstrator, who is manipulating an object. But how does the robot know what to observe in the whole scene? First, the robot has to identify the object candidates, which can be manipulated. The human selects one object for the manipulation by grasping it. Now, the robot knows the interesting object. The 6DoF-trajectory of this object can be observed with a vision-based system. It is important to notice our object-centric point of view: It enables the estimation of the spatio-temporal object properties for manipulation tasks (Petsch and Burschka, IEEE ICRA 2010). An exemplary observation of a human is illustrated in the Figure above (Petsch and Burschka, IEEE ICRA 2010_). Read more_.

After the extraction of the manipulation-relevant object properties, the question of knowledge representation arises. The representation influences the future abilities of the robot: First, the robot should be able to reuse its knowledge. Second, the knowledge has to be extendable, since the robot can be confronted with new situations. A further important aspect is the question who or what influences the manipulation-relevant object properties. The properties are, of course, not only set by the object and its state, but also by the environment. Therefore, the functionality of an object is also influenced by its environment (Petsch and Burschka, IEEE IROS 2011). An exemplary scene is depicted in the Figure above (Petsch and Burschka, IEEE IROS 2011). Read more.

Teaching

Exercise Robotik WS 2011/12

Exercise Robotik WS 2012/13

Publications