Working on challenging research projects, among other things, is also a great opportunity to learn. As I learn new things I get more inspired and that inspiration drives me to apply new ideas to even more projects. This sweetening spiral often makes me ache to work on some of these ideas, but alas only twenty-four hours a day with no sleep are not enough to work on all the projects that I would like to pursue.

In this page, I describe a few future projects that I would like to share with the World Wide Web and some projects in their infancy that deserve an honorable mention.

Using Wiimotes for Psychophysical Investigations

Date: Future
Collaborators: Looking for Any

Whenever you research an interactive technology, you need to validate your results with a user study. Unfortunately performing a good, meaningful user study is very challenging. Even just finding a sufficient number of people willing to participate is not easy. And then, how are you going to do it? Are you just going to have them fill a survey and do an ANOVA test or Duncan’s MRT to make your clearly insufficient data look good on your next publication? Sadly only large industries can afford to put together a substantial focus group and perform user studies that are scientifically meaningful. While we have to live with the fact that our user studies can only involve a small group of people, we*can* improve our means! Many quantitative studies in psychology use devices that can "measure" human subjects directly during an experiment. For instance, the saccadic movements of the eye are often used as an indicator of human response to external stimuli, Can we do something similar for computer graphics and HCI?

Another related problem is to design technologies that try mirror human behavior, such AI for autonomous agents, and validate whether your model is good enough. Again this is a critical problem in the domain of computer graphics as we need to devise technologies that look pleasing and appeal to a large audience.

One possible solution for both problems is to build a platform that can measure meaningful quantitative data directly from human users. What we need to do this is a technology that is cheap, easy to use, and widely available so that a large number of researchers may use it to exchange their findings. I believe that the popular Wii Remote (Wiimote) is a device that addresses many of these requirements and thus may be a useful tool for this kind of research.  In fact, it may sound ridiculous to many scientist to use a wiimote --a toy-- to tackle a "serious" problem, yet if we were to devise some custom hardware, nobody would be able to use it and we will not be able to raise awareness among researchers on this fundamental issue. We need a popular device to get this started!

I have already put together a few existing libraries and got an API running that may serve as a starting point for this work, but I am still looking for someone, possibly a motivated graduate student, that is willing to collaborate with me on this project.


Fast Hardware Accelerated Stereo Vision

Date: Infant
Collaborators: Undisclosed

I am collaborating with several other researchers in the academia and industry to study a hardware architecture that could accelerate stereo vision for applications in robotics and autonomous vehicles. Our goal is to go beyond what can be done today with accelerated techniques based on correlation.