CoViS (Cognitive Vision Software) is a C++ computer vision library framework. Its core functionality is the extraction of an Early Cognitive Vision inspired scene representation. Based on this additional processes (e.g., motion estimation, model accumulation, creation of higher level features) are provided.

CoViS is supposed to be used as an Early cognitive vision level which bridges the gap between early vision pixel wise processing and cognitive vision tasks and which is in particular useful for bootstrapping processes in cognitive robots. It is the basis for all applications developed by SDU within the PACO-PLUS project.

CoViS is made available under a BSD license. Access is currently by personal request since we are still improving the documentation and demo programs. We expect a full public release by fall 2010.

Provides estimation of the hand pose in terms of orientation and joint angles. It's based on classification of a monocular view against a database of hand pose synthetic images generated with Poser software.

Provides real-time estimation of the hand pose in terms of orientation with respect to the camera and joint angles. It's based on classification of a monocular view against a database of hand pose synthetic images generated with Poser software. After that classification, the best candidates are compared with previous estimations to avoid glitches. The library also shows which is the closest match from the database. More information about the system can be found in:

Javier Romero, Hedvig Kjellstrom, Danica Kragic. Hands in Action: Real-Time 3D Reconstruction of Hands in Interaction with Objects. In ICRA 2010

SPOAC provides a network interface and a software framework for symbolic execution of OACs on a robot. It contains a network interface to a symbolic planner, an OAC database server and a GUI through which all three parts can be controlled and monitored.

he high-level domain definitions we have developed during PACO-PLUS will be made available as part of a collection of benchmark planning problems. This collection with include descriptions of the planning domains for the SDU robot/vision system and the KIT ARMAR robot. It will also include examples of simple dialogue problems. The domain description files will be provided primarily as PKS input files, but also converted to PDDL where possible.

Inventor Physics API (IPSA) is an extension of the Open Inventor toolkit with physics objects using the ODE library for simulating rigid body dynamics. IPSA allows the generation of 3D animations and simulations by considering physical attributes such as shape, mass, gravitation, friction etc. The simulations may be programmed directly using the Open Inventor extension classes or more preferably by writing an Open Inventor Scene description file.

Several examples for both methods are provided with the code.

This library facilitates the exchange of messages between the high-level planning components (e.g., action planner and plan execution monitor) and other modules in a multi-component software system. The library is divided into three main parts: (i) a lightweight network transport mechanism, based on sockets, providing a range of communication services (i.e., point-to-point, subscription, and broadcast messaging), (ii) a predefined message exchange protocol for requesting and delivering high-level plans to particular components in the system, and (iii) interface code for operating with the PKS planner, and for using plans generated by other external services.