The KogMo-RTDB interfaces can be regarded as part of an outer framework for interprocess communication. The actual inner framework used within a cognitive module heavily depends on the engineering discipline it comes from: for the control level, this might be a robotics or control framework, at an artificial intelligence level an interference machine might be the choice.
- The license has been changed from LGPL to Apache License 2.0 (2011-12-31)
- The code has been moved to GitHub (2011-12-19)
- Real-time Database as central interface for interprocess communication (event trigger available)
- Data exchange between both real-time and non-real-time modules
- Data consistency guaranteed by non-blocking read/write protocol
- History for temporal decoupling of slower modules with longer cycle times
- Comprehensive data logging using an interference-free method
- Hard real-time capable with specified hard- and software platform
- Reference platform for hard real-time:
GNU/Linux (CentOS 4) with Kernel 22.214.171.124 and Xenomai 2.4.4
- Full functionality with soft real-time:
C compiler and a POSIX compatible system (uses pthreads, mutexes, condvars, message queues, shared memory)
- Minimal requirements (no real-time):
shared memory (e.g. tmpfs on /dev/shm/)
emulates mutexes with polling and atomic instructions on x86, arm and mips
- Successfully running on embedded systems:
- Linksys NSLU2 running SlugOS 4.8-beta (compiled little endian with GNU-EABI)
- Asus WL-500GPv2 running OpenWRT 8.09 (Kamikaze, BRCM-2.4) (see this german book on embedded linux and this article)
- Cognitive Automobiles (Transregional Collaborative Research Center 28 funded by Deutsche Forschungsgemeinschaft)
- The autonomous robot car AnnieWAY(photo) at the DARPA Urban Challenge 2007
- Cognitive Factory and Cognitive Household (Cognition for Technical Systems)
LicenseThe source code on this site is published under the Apache License Version 2.0. You must retain the copyright notice of the original authors of the KogMo-RTDB. (Other licensing models on request).
If you use the KogMo-RTDB in your own projects, I'd be pleased if you tell me about it.
If you agree, I'll add a link on http://www.kogmo-rtdb.de.
If you publish about projects using the KogMo-RTDB, please cite one of my papers and send me a note. The first paper on the KogMo-RTDB was the IEEE Intelligent Vehicles 2007 paper:
- Matthias Goebl und Georg Färber. A Real-Time-capable Hard- and Software Architecture for Joint Image and Knowledge Processing in Cognitive Automobiles. In Proc. IEEE Intelligent Vehicles Symposium, pages 734-740, IEEE Press, Juni 2007.