Selected Publications

This paper presents a six degree of freedom position regression CNN (convolutional neural network) based on Google's Inception-V4 CNN. This network is then evaluated quantitatively and compared to previous state-of-the-art position regression CNNs. Our model achieves a 22% and 51% relative improvement compared to previous state-of-the-art methods for position and orientation accuracy respectively. A modular system for integrating our model into probabilistic localization algorithms for accurate kidnap resolution and global metric initialization in real-time is also introduced and evaluated. This modular system is able to globally initialize 85% of the time in under 70ms. If the robot is allowed to rotate in place and capture multiple views, this rises to 95%.
In URAI

Recent & Upcoming Talks

Recent Posts

More Posts

Behavioral interviews are about showing your interviewer how you respond in a variety of situations. But beyond that, they help give the interviewer a sense of who you are and if you have a good “culture fit” with the company. At a really high level then, behavioral interviews are about people. It came as a bit of a shock then to hear about the interview experience of a few colleagues who encountered the video interview process created by HireVue.

CONTINUE READING

If another OS (also *nix based) is installed alongside arch, grub will likely get messed up on the arch side of things. Grub will see the arch linux boot option, however, upon startup it will throw a kernel panic with the error: Not syncing: VFS Unable to mount root fs on unknown-block. This is caused by a bug in the way some other linux distros generate the grub.cfg file. If you have the above error, your grub.

CONTINUE READING

So, it has been some time since my last post… Before I get into the fun stuff (well… fun for me anyways), let me give you a bit of a life update! Having already finished 1/4th of my masters, I can say with confidence that I am really happy that I decided to continue on and get an advanced degree. I’m able to work with world-class researchers on amazing topics doing work that has the potential to drastically improve people’s lives.

CONTINUE READING

As I groggily walked out of my dorm and into the crisp Fall air this morning it became readily apparent that Winter was indeed on its way. It is that extra little bite to the air that wasn’t present a few days ago. Midterms have come and gone uneventfully which in my mind is a good thing. I did find out a rather crazy fact about tests for undergraduates here at KAIST though… apparently, most tests give you as much time as you want.

CONTINUE READING

It has taken a while for me to really feel settled in here, but I think I’ve finally crossed that line and I’m starting to feel really blessed to be here. The last couple weeks have been an absolute blast and given the fact that I basically get to build cool things every day, so should the rest of my time here. As far as classes go… I can’t really pick out many differences so far compared to the structure of similar graduate level courses in the US.

CONTINUE READING

Projects

Value Iteration Networks

PyTorch Implementation of Value Iteration Networks including full gridworld environments.

Inception V4

Keras implementation of Google’s Inception-V4 architecture (includes Keras compatible pre-trained weights).

Touch Location Inference (Side Channel Attack)

Web based touch location inference (SCA) on mobile devices using the accelerometer and gyroscope.

3D Object Reconstruction

Monocular 3D static object reconstruction using shape from silhouette techniques.

Contact