2009 E&RT Projects

CSCI P545 Embedded & Real-Time Computing

Tue Oct 25 15:34:54 EDT 2011 [SDJ]

ERTS Projects – Fall 2011

These are preliminary project suggestions. More details will be added as time permits. Meanwhile, if you are interested in undertaking one of these projects, let the Instructor know. Most of the projects listed have been undertaken before, and the goal is to advance that prior work. Keep in mind that finishing the Lab Project is more important than doing a project, we have just six weeks left in the semester, and the weather will get much worse.

Propose your own project. Any project will be considered that increases your experience with embedded development. It is very helpful to write a project description, rather than telling the Instructor, "I have an idea about something to do." Part of accepting a project proposal is believing you have thought things through. If you are considering starting a project that will continue under independent study credit, consult with the instructor.
Improve your Lab Project (Recommended for most students). There will be many ways to improve on the basic capabilities of the Lab Project. Some examples:
- Demonstrate more complex avoidance maneuvers.
- Responding appropriately to "impossible" avoidance scenarios.
- Performing alternative maneuvers.
Higher planning capabilities. Prior students have explored more sophisticated treatments of path planning, including
- Splining, the traditional engineering approach.
- Force vectors, a classsic AI approach.
- Swarming, a cellular, incremental approach.
- There are many other approaches to planning that have yet to be tried.
Speed Control (recommended for students specializing in embedded systems). Add a reliable speed control component. We have fragments of one, but more experimentation is required.
Tactical Maneuvers. Special modes for special circumstances, such as driving through a gate, parking, etc. See the project description [PDF]
Vision. Ben and Shiva are working on adding a vision system to ERTS. I'm sure they could use some help. We have accumulated several vision functions over the years that need revisions to deploy on the vision systems. For instance, most of these functions were developed before we made the decision to use OpenCV as the standard vision library.
CartFS Frameworks. Students have done projects implementing CartFS in other languages, including Java, C/C++, Scheme and Matlab. The next step is to re-implement the Lab project in those languages and in those frameworks. This has been done in Java and Scheme, but is not well tested.
Proxy Components. Experiments developed beyond the reach of SyncFS can be integrated by running a proxy component to serve as an interface. For example, a socket component running in the /tmp/cartfs realm would accept a stream from outside parse or construct the JSON representation, and synchronize through SyncFS. Something similar can be done with TCP/IP channels, wireless connections, and even through the internet.
Component Migration. The Python programs in the Lab Project are prototypes that can be implemented at lower levels, in C/C++ for example, and embedded more deeply, in the VCM for example. Similarly, SyncFS files are prototpye communication channels that can be implemented at lower levels, as sockets, pipes, shared-memory regions, and so forth. The beginning of a project is doing this by hand, with the aim of developing a systematic implementation methodology.
More to come ...