Grants: finished » Grant Dziekański 2012: DisCODe MDE

PT napisał (23.12.2011):

INTO 2.0.0.beta2 released

For once a soft packet that doesn't suck

Today, we released a shiny new beta of Into, the machine vision and AI platform of the future. What could possibly be a better way of spending your Christmas than coding parallel image processing applications in C++ or JavaScript? This is the first version publicly available to anyone without registration. Grab your copy at http://intopii.com/into/download/. Happy coding!

Release notes

The most notable new feature in 2.0.0beta2 is a new remote object framework. It allows one to map QObjects directly to URIs in a web server, which makes web browser integration a no-brainer. The remote object framework also provides a completely network-transparent signal/slot mechanism and allows calling remote functions with no "brokers" or "stub compilers" many Corba developers hate deeply. Beta2 comes with a new OCR plug-in and over 200 bug fixes and improvements in performance, stability, and usability.

New features

- Remote object framework
- Property protection levels in PiiOperation
- Thread-safe read/write of properties in PiiDefaultOperation
- OCR plug-in. Handwritten digit recognition implemented.
- An almost complete rewrite of the linear algebra library. New SVD and PCA algorithms.
- Circular Hough transform
- Multi-class boosting and an operation interface to boosting algorithms
- Blocking learning to PiiClassifierOperation
- Solution algorithm to the linear assignment problem
- Platform-neutral microsecond-resolution timer (on Linux/OS X, no Windows support yet)
- Support for template classes in JavaScript API
- JavaScript wrappers to QColor, PiiAbstractSocket
- New "synchronized" construct for easy mutual exclusion.
- Prebuilt MSVC binaries.

Improvements

- Better OpenCV integration, including bug fixes
- Remarkably improved camera driver interface with a few new implementations
- PiiAsyncCall now accepts parameters for asynchronous functions
- Scale invariance to shape context descriptors
- Region-of-interest support to adaptive thresholding
- New composition mode in PiiMatrixComposer
- ... and about 70 other minor changes

Changed functionality

- PiiVariant now uses qint64 as a long integer type.
- PiiDefaultOperation::Simple was renamed "NonThreaded"
- PiiWaitingIODevice no longer exists. The functionality was moved to PiiSocketDevice
- PiiBackgroundExtractor property names and output name
- PiiImageFileWriter::stripPath is now false by default
- Renamed PiiNetworkClient::readTimeout to connectionTimeout

Bug fixes

- PiiOperationCompound::stop() now works in "Starting" state.
- PiiOperationCompound subclasses now work in JavaScript
- Never-ending on-line training in PiiClassifierOperation.
- Potential synchronization error when stopping a configuration with many parallel flows at different flow levels.
- ... and more than 150 others, many of them platform-specific.

My best wishes to the new year.

PT napisał (7.04.2012):

RTMaps - konkurencja DisCODe

Ci, co widzieli prezentacje dot. PROTEUS'a to pewnie znają, ale mam nadziej

http://www.youtube.com/watch?feature=player_embedded&v=HBxFq04S91g
http://intempora.com/technology/rtmaps-software

Wrzucam też garść pomysłów na to, skąd czerpać wiedzę (tak na start):

Component-based robotics engineering z bricks
http://www.best-of-robotics.org/publications

Hmmm, kiedyś byłem na takiej stronie (ba, miałem ją nawet wydrukowaną, ale to było w czasach FraDIA) o robotycznych middleware...
Niestety to nie to: http://en.wikipedia.org/wiki/Robotics_middleware

Tam było takie zestawienie zrębów i ich porównanie, naprawdę spore... hmmmm

Hmmm, coś mi świta, że to było związane z RoSta:

http://wiki.robot-standards.org/index.php/Main_Page

No i nie warto zapominać o SmartSOFT

http://zafh-servicerobotik.de/ULM/en/publikationen.php?exp=

Znalazłem ciekawy artykułu dot. oprogramowania (w szczególności zastosowaniu agentów):
"Robotics software frameworks for multi-agent robotic systems development"

http://www.sciencedirect.com/science/article/pii/S0921889012000322

Tutaj taki tekst opisu z grantu europejskiego, oni badają to od sprawy hardware, natomiast wyartykułowali kilka ciekawych myśli, które de facto są i naszą motywacją... ;)

http://ec.europa.eu/information_society/events/cf/ictpd12/item-display.cfm?id=8637

Ubiquitous Complex Event Processing in Exocortex Applications

The uCepCortex project develops the concept of an artificial cognitive system that integrates multiple sources of information, including from specialized sensors and software agents in the cloud, and acts as an exocortex for human users. It combines a high degree of autonomy with sophisticated human interface techniques to ensure that the human is kept in the loop and exercises control, while greatly extending his/her effective cognitive range by off-loading processes to the cognitive system.

The uCepCortex approach utilizes the emerging disciplines of Ubiquitous Complex Event Processing and Event-Driven Process Management to implement a human-inspired cognitive model that complements human abilities and can detect, correlate, filter, enrich, process and learn from millions of events per second from arbitrary event types, and that implements flexible (re-) actions or processes in rapidly changing scenarios.

It includes the development of low-cost smart sensors that may be either worn or distributed in the environment, including some extending senses beyond human abilities, that implement cognitive processes including attention, collision avoidance, recognition, anomaly detection and detection of emotional cues. These components use rapid, parallel and reconfigurable computational hardware (FPGA)or Numenta's Grok to implement the artificial cognitive architecture. The exocortex includes rich information flow to the human, and limited control flow from the human. System to human interfacing includes rapid, high-bandwidth augmented sensory input via a belt device, and linguistic/visual feedback for symbolic information.

We investigate how the brain responds to exocortex interaction using emerging Brain-Computer Interface (BCI) technologies, and uCepCortex learning algorithms will support co-adaptation of the exocortex and user. Included demonstrators will explore improved quality of life. Ethical, legal and social issues will be evaluated, consistent with the recommendations of the fBNCI roadmap. The uCepCortex project will use value-sensitive design to ensure that the emerging technology also meets the moral requirements of users.

We will in addition investigate how an autonomous robot or Cyborg would be started with a basic modeling of its special matrix according to what this Artificial Cognitive System (ACS) respectively Cyborg should do and how it would autonomously learn and enhance its matrix or world model. We will also explore how a swarm of robots would mutually exchange information according to the theory of quantum information. Such a swarm would remodel the swarm's matrix more and more complex, according to the Entropy or the second law of thermodynamics. Question is whether such autonomous robot swarms would build a parallel, own "reality" and act or "live" in it, and how the mankind would guarantee the control for a symbiotic coexistence and cooperation.