Robocup 2010: Notebook
1/31/2010
Today I continue to examine BHuman's codebase and document my findings. I hope to finish this by today and start looking in detail at the walk enging code. Alright, all done with initial look through the code! Tommorow I hope to work on 2 main things: 1. Plan the next few steps of working on the walk. 2. Look through BHumans Simulator interface and walk engine in detail.
2/1/2010
One of the questions which came up today is where do BHuman use DCM in their walk. This will be the first issue I investigate today and a good segue into reading about the walk engine in more detail.
In answer to the question, it seems BHuman uses ALMemoryFastAccess rather than the DCMProxy. This is a bit unexpected because all the documentation about DCM details examples using the DCM proxy and doesn't mention the ALMemoryFastAccess class.
2/2/2010
Met today with Luis Sentis to talk about designing an omni-directional walk. He feels that it is quite possible and even has some code snippets to generate the necessary patterns. With his aid, as well as Michael and Todd's previous experience, it should be quite possible to make progress. Adam and I now need to focus on mastering DCM to lay the groundwork for the higher level walking.
2/3/2010
Today I'm working on getting Adam's DCM code to compile in UT's framework. I hope to finish this soon, but its hard to gauge exactly how many errors there are and how long each will take to fix. C++ is not my strongest language so it takes me a while to fix each error.
2/4/2010
I continue to work on getting the DCM code to compile. The issue is complicated by the fact that we may be using conflicting versions of NaoQi. I'm beginning to think it may be simpler to just rewrite his code and slowly integrate into a codebase I know is working.
2/5/2010
Code compilation was completed late yesterday. I also fixed the wireless problem on the laptop I'm using. At this stage, Adam and I need the robots to continue the testing process. Because they're not getting back until next week, we'll take a few days off.
2/8/2010
Michael and Todd back from Paris as of yesterday with news of a new SDK release. This breaks the Webots functionality, leaving the team without a simulator. This puts priority on getting the BHuman simulator to interface with UT's code. As of now, this will be my primary project.
2/9/2010
So far the Simulator has evaded my attempts to find a clean interface. The code is pretty devlishly tricky or at least was written in non obvious ways. Today I hope to continue looking with whatever time I can spare.