Research
Research Topics:
ASPOGAMOThe research project 'Sensor-based, Automatic Analysis of Football Games' is an ambitious, mid-term research project that studies the automation of these tasks. The main objectives of the project are (1) the investigation of novel computational mechanisms that enable computer systems to recognize intentional activities, (2) the development of an integrated software system to automate game interpretation and analysis, and (3) the demonstration of the impact of automated game analysis on application areas, such as sport science, football coaching, and sports entertainment.
|
|
|
|
CogitoA key challenge for the next generation of autonomous robots is the reliable and efficient accomplishment of prolonged, complex, and dynamically changing tasks in the real world.
One of the most promising approaches to realizing these capabilities is the plan-based approach to robot control. In the plan-based approach, robots produce control actions by generating, maintaining, and executing plans that are tailored for the robots' respective tasks. Plans are robot control programs that a robot can not only execute but also reason about and manipulate. Thus a plan-based controller is able to manage and adapt the robot's intended course of action --- the plan --- while executing it and can thereby better achieve complex and changing goals. The use of plans enables these robots to flexibly interleave complex and interacting tasks, exploit opportunities, quickly plan their courses of action, and, if necessary, revise their intended activities.
One of the grand visions in the area of plan-based robot control is the realization of general autonomous robot control programs that can adapt themselves to the environments they are to operate in and to the distribution of complex tasks they are to perform. An instance of this grand vision is a pre-programmed household robot that knows how to clean a kitchen, how to operate a dishwasher, and so on. Being installed in a new environment it specializes its general plans to the specifics of the household and learns to manage the specific agenda of household chorus that is given to it. The robot also has to learn about the pitfalls of its tasks and its environment and avoid them through foresight. Our research field is still far away from realizing such competent robotic agents.
|
AwareKitchenIntelligent sensor equipped environments can be of much greater help if they are capable of recognizing the actions and activities of their users, and inferring their intentions.
Understanding human activities and characterizing them into expressive and detailed activity models is one of the key issues of today's current pervasive computing systems. If such a system could recognize and understand automatically its user's behavior, it could interact in a more efficient and friendly manner. Unfortunately, the current model construction techniques are based on supervised learning and require specifications from their human counterparts, such as labeling the acquired sensor data. Our vision is to build technical cognitive systems that create and use models in a straightforward manner, by combining already existing online information with the system's context history.
|
|
|
|
MudisSince existing methods for human-machine interaction are often
unintuitive, a lot of time is required for humans to adapt to the
operation of a specific machine. In contrast, the MuDiS project aims at
granting machines the ability to adapt to typical human behavior. The goal
of the project is the development of a multimodal dialog system that
considers various human communication channels such as facial expressions,
spoken language and gestures for human-machine interaction. We perform
experiments to determine the requirements for robots to interact with
humans in an intuitive way. Insights gained from these human-human
experiments are applied to the human-machine interface to grant robots the
capability of participating in simple, every-day dialogs in various
environments. To tackle this challenge, we unite researchers from diverse
scientific areas, such as computer science, electrical engineering and
psychology to reflect the interdisciplinary character of the project.
|
CoPThe project aims at the unification of vision-based sensing in the CoP (Cognitive Perception) in the learning and planning system. On the one hand, CoP manages the interpretation of different kinds of sensors and on the other hand it automatically acquires and maintains the knowledge about the world and objects in the world. CoP selects sensors and sensor interpretation algorithms based on their expected utility. To this end, CoP learns and improves intersensor and inter-algorithmic models for method seleciton from experience. Especially the vision system, the major sensor we use, provides several automatic model improving techniques. Improved models accelerate the perception process and provide more robust results.
|
|
|
|
EnvModWe focus on point cloud based representation and reasoning techniques for building accurate and meaningful 3D maps for mobile robots in both indoor and outdoor environments. One of our main application and deployment scenario is the Assistive Kitchen. However, all our methods were carefully crafted with generality in mind, therefore they have been also successfully applied to outdoor urban, aerial, and underwater datasets.
|
KnowRob / Knowledge 4 CoTeSysThe project targets at building knowledge representation and processing systems for mobile robots by combining description logics knowledge bases with data mining, (self-) observation modules and imported knowledge from the World Wide Web.
|
|
|
|
PARAThis projects develops plan-based control mechanisms for human-robot interaction, where the robot assists the person in everyday tasks and adapts to the person's abilities, expectations and preferences. The joint human-robot plan is represented explicitly in the robot program. To opimize the joint execution, we develop methods for representing and learning models of a person's capabilities, expectations and preferences. We apply this research in different domains, especially in the context of elderly care, where a robot assistant can enable elderly people to live independent, who currently depend on the help of others.
|
CogMaShCogMaSh aims to create a manufacturing system that achieves similar levels of flexibility,
robustness and improvement through experience as found in human machine shops.
In the production of prototypes, customized products and small or mid-size series, human workers
with their problem solving abilities, experience and cognitive capabilities are still the single way to provide the required flexibility, adaptability and reliability.
|
|
|
|
CogManThe CogMan project (1) develops computational and control models of pick-and-place tasks in the context of
everyday manipulation activities in human environments, (2) implements the model into a control system for
the kitchen scenario, and (3) empirically analyzes the impact of this control model on the flexibility, robustness,
adaptability, and naturality of the robot behavior.
|
Face Image AnalysisAs robots emerge from their classical domain - factories - to be included in every day life, they need to gain new abilities besides those needed in manufacturing. They need not only to support humans, but also be able to socialize with their users to enhance the interactant experience and allow for social bonding. Recent progress in the field of Computer Vision allows intuitive interaction via gesture or facial expressions between humans and technical systems. Recent research aims at enabling machines to utilize communication channels natural to human beings, such as gesture or facial expressions. Humans interpret emotion from video and audio information and heavily rely on this information during every-day communication. Therefore, knowledge about human behavior, intention, and emotion is necessary to construct convenient human-machine interaction mechanisms. The human face provides much of the information that is passed between humans in every-day communication. Although most of this information is passed on a subconscious level, we still rely on the interaction partner's facial expression to determine emotional state or attention to form a prediction of his or her reaction.
|
|
|
|
MeMoManWe are developing new computational models and a system for accurate measurement of human motion. Our primary goal is to develop markerless vision-based tracking algorithms for use with the industry-proven anthropometric human model RAMSIS (in collaboration with the TUM Ergonomics Department/Faculty of Mechanical Engineering). By providing RAMSIS with markerless tracking capabilities, we open up new fields of application in ergonomic studies and industrial design. On the other hand, we believe that a far-developed, flexible and accurate model such as RAMSIS is beneficial for human motion tracking given the ergonomic expertise that has affected its design.
|
ProbCogProbCog designs and implements a statistical relational learning
system that supports efficient learning and inference in relational
domains, focusing mainly on probabilistic logical models. Based on a
common data model, we integrate various representation formalisms and
have successfully extended the expressiveness and practical
applicability of existing approaches. ProbCog thus provides a coherent
probabilistic framework that enables cognitive technical systems to deal
with a high degree of uncertainty and complexity in a variety of
application contexts, including adaptive plan generation for
under-specified tasks and heuristic generation.
|
|
|
|
Highspeed DartboardThe main objective of this team semester project was to offer all participating students a highly detailed look at an interdisciplinary project work (mechanics, electronics, software). Furthermore they were able to get to know the importance of the interoperability of all these disciplins with the means of a practical example.
A technical realization was done by integration of image processing techniques for the trajectory calculation of all found darts, controlling strategies for an exact positioning of the dartboard.
|
AgiloRobotic soccer has become a standard 'real-world' testbed for autonomous multi robot control. In robot soccer (mid-size league) two teams of four autonomous robots --- one goal keeper and three field players --- play soccer against each other. The soccer field is four by nine meters big. The key characteristics of mid-size robot soccer are that the robots are completely autonomous. Consequently, all sensing and all action selection is done onboard of the individual robots. Skillful play requires our robots to recognize objects, such as other robots, field lines, and goals, and even entire game situations.
In the AGILO project we investigate how probabilistic visuomotoric autonomous robot controllers that are capable of learning can meet these challenges.
The AGILO robot controllers employ game state estimation and situated action selection based on automatically learned control mechanisms.
|
|
MuJoVisionThe Multi-Joint Vision project aims at providing image data obtained simultaneously from a large number of cameras in real time for further processing in Computer Vision applications. It involves the integration of a flexibly expandable architecture and interface for distributed Computer Vision applications. The project is part of a Multi-Joint Action scenario developed for the CoTeSys cluster of excellence, and the data obtained is used to generate tracking information for Human-Computer Interaction and monitoring tasks.
|
|
CRAMThe main scientific goal of the proposed project is to build the
CRAM (Cognitive Robot Abstract Machine) as a software toolbox for the
design, the implementation, and the deployment of cognition-enabled
autonomous robots performing everyday manipulation activities. CRAM
provides a language for programming such cognitive control
systems. This language includes data structures, primitive statements
and control structures that are specifically designed to enable and
support mobile manipulation as well as cognition-enabled control.
CRAM is needed because a robot performing everyday manipulation tasks
must continually decide on its course of action and on how actions have
to be performed. Even seemingly simple tasks such as picking up
an object from a table require complex decision making. To pick up an
object, the robot must decide where to stand in order to reach the
object, which hand(s) to use, how to reach for it, which grasp type to
apply, where to grasp, how much grasp force to apply, how to lift the
object, how much force to apply to lift it, where to hold the object,
and how to hold it. The decision problems are even more complex
because many decisions depend on the task context, which requires the
robot to take many factors into account to achieve the best
performance or at least a performance that is good enough.
|
|
|
|
Naive Physics ReasoningIn this project we investigate a simulation-based approach to naive
physics prediction in the context of autonomous robot everyday
manipulation. We identify the abstractions underlying typical
first-order axiomatizations as the key obstacles for making valid
naive physics predictions. We propose that naive physics reasoning
should not be performed based on abstractions but rather based on
detailed physical simulations.
This idea is realized as a naive physics reasoning system for
autonomous manipulation robots that translates naive physics
problems into parametrized physical simulation tasks, that logs the
data structures and states traversed in simulation, and translates
the logged data back into symbolic time-interval-based first-order
representations.
|
