aO(N²)

OpenTLD is an open-source visual tracking algorithm by Zdenek Kalal. He released it earlier this year as open-source, and there have been efforts to port the algorithm from the original MATLAB to a C/C++ only implementation.

Lytro has released their light field camera, where you can set the focus AFTER taking the photos. The camera records light as more than a single-valued pixel - it records intensity as well as the vector direction of the light. Focusing is done on the software rather than the optics, giving it a lot more data than traditional photos. This holds the potential for faster photo taking as well as improved low light performance (however, the company doesn't seem to have demonstrated low light performance yet).

I haven't been using the Robotic Arm that much lately due to the hassle of getting and installing new batteries. However, since I'm going to make a presentation about the robotic arm tomorrow, I've modified it to take most of its power through a wall socket DC adapter.

This is just a personal update. I've recently posted the PS3 sixaxis video and a placeholder for the tutorial, however time is quite limited for me these days. In addition to having a full-time job, I'm currently taking two online courses: Stanford University's Introduction to Artificial Intelligence and Machine Learning.

Hopefully this will help me build better algorithms as well as cooler projects, and even at the beginning I'm learning some things from the Machine Learning course that I think I can already use in some code.

Update: demo video now up!

Here is the tutorial in using the PS3 sixaxis controller with the OWI/Maplin robotic arm. It's less impressive than the voice control, however it's still useful in some ways as the sixaxis controller is an excellent controller.

On Friday I managed to enable my PS3 sixaxis controller to work with the robot arm! It's not as impressive as the voice control, but the code might come useful to me someday. The PS3 controller uses bluetooth and acts like a joystick HID device, and I used pygame to help me process the input events.

Additionally, using this I can actually move several joints at the same time. I'll be posting a short tutorial, code and a demo video as well soon!

Here is part 3 for the robot arm with voice control tutorial. Click here for the first part or here for the second part.

Now that the output has been filtered and tested for confidence, the sentence can now be processed to move the arm:

Here is part 2 of the robotic arm tutorial. Click here for the first part.

Running julius

If you've managed to perform Part 1 of the tutorial successfully, then in the command line in the 'voxforge/auto' directory, run:

Julius should then wait for microphone input, and if you speak into the microphone, for example 'elbow up' then it should print output similar to the following:

UPDATE: Demo video now up!
NB: I will take my time creating this tutorial, as my typing speed is slower due to my wrist injury. I will be able to type much faster when my plaster comes off in about 2 weeks' time.

At 2am last Tuesday, I managed to get it working! Using HTK, Julius, libusb and a python script, I have enabled voice control for my maplin robot arm. Full details, plus a demo video coming soon!