Send article to a friendPrint

Perception and Cognition

Head of Department: Dr. Ing. Michael Suppa

The DLR crawler explores and maps autonomously its environment . Fiels of competences .

The Department of Perception und Cognition develops solutions to enable a wholistic
perception of robotic systems using the two predominant robot senses vision and tactility.
Perception and cognition are elementary components for tele operated, and autonomous robot systems, due to the fact that they enable reactive behavior, attention-based control and interpretation of situations.

The research areas cover

• Innovative sensor concepts
• Sensor data processing and analysis in realtime, e.g. stereo reconstruction, object tracking, and 3D-reconstruction
• Optical navigation and localization of mobile, autonomous systems
• Processing and analysis of EMG and forcetorque data
• 3D environmental modeling using point-, volume-, and surface-based representations and derived functional models
• Generation of digital terrain models and fusion of 3D-models of different dimensions
• Data fusion, data classification and learning algorithms
• Object recognition, pose estimation and scene interpretation
• Real-time 3D-visualisation of large data sets, e.g. terrain rendering, augmentation and auto stereoscopy

The development of integrated solutions for robotic perception from the data acquisition
to the application is a core competence of the department. Sensor-based applications such
as tactile and visual exploration and navigation, photo realism, further augmentation of real
scenes for tele robotics as well as derivation of solutions for automation in the
industry are research topics. The methods for data interpretation and task-oriented
scene comprehension are based on the modeling process and allow for efficient exploration
for different missions in space, aerospace and terrestrial scenarios.

The department is networked worldwide as part of national and international cooperations.
Additionally, close industrial contacts exist, accounting for the increased industrial demand
for intelligent sensor data processing.