HeartLander: Miniature mobile robot for minimally invasive cardiac therapy

One of the most promising application areas for robotics and more specifically miniature and nano robotics is in medicine. Whether the tiny robots are specifically designed to deliver medications or directly attack viruses, their usefulness in prolonging our lives and eliminating the pain and suffering of disease is indisputable. I am always happy to read about recent advances in medical robotics that bring us one step closer to such devices. This post is about HeartLander, a miniature medical robot under development at CMU's Robotics Institute; the robot is designed for performing minimally invasive cardiac therapy.

So how does the robot work?

Basically, a surgeon creates a small incision on the patient's chest. Using a pair of forceps, the surgeon places the robot directly on the beating heart. Using a joystick, he can then guide the robot delivering medicine directly to affected areas, helping to attach pacemaker electrodes or even assisting with specialized techniques for curing arrhythmia. The worm-like robot moves using an ingenious mechanism driven by miniature ultrasonic piezoelectric motors.

Although the robot is still a proof of concept, the CMU research team has been able to demonstrate its use on a pig's beating heart (see the video at the end of this post.) The team still has to work out a number of issues such as the development of wireless remote control mechanism in order to eliminate the reliance on a tether whose stiffness causes problems with locomotion. The same tether is used to supply energy to HeartLander although a future production version would most likely utilize an on-board battery. This is an excellent and very promising research project and I am looking forward to the next generation of HeartLander.

Note: The image and video are copyright CMU.

Siftables: From wireless sensor networks to tangible user interfaces

Interacting with a computer using a keyboard and mouse is really starting to get old. I recently purchased a Tablet netbook and I have found the touch screen interface a pleasure to use over using the mouse; coupled with Vista's excellent handwriting recognition, the tablet has become my number one machine for use daily. But how about new paradigms for human-computer interaction that are not just a small change over how we interact with computers today but a completely new one?

Say hello to Siftables, the creation of MIT Media Lab graduate student David Merrill and his colleagues. Siftables are small blocks with a microprocessor, an LCD display screen, and the ability to sense other Siftables around them.

Siftables applies technology and methodology from wireless sensor networks to tangible user interfaces. Siftables are independent, compact devices with sensing, graphical display, and wireless communication capabilities. They can be physically manipulated as a group to interact with digital information and media. Siftables can be used to implement any number of gestural interaction languages and HCI applications.

It is very difficult to appreciate how innovative Siftables are to human-computer interaction from a textual description so I encourage you to watch the following video of David's TED conference presentation of his work. It is truly an amazing idea!

Shaving digitally

Have you ever browsed old photographs only to have wondered how much better you could have looked had that bushy beard not been hiding your face at the time? Or are you contemplating shaving your beard of many years but you would like to see the final result before the event just in case? It is now possible to satisfy you curiosity using a new image-based shaving technique developed at CMU's Robotics Institute.

Virtual shaving is achieved by modeling human faces in images as a set of layers that can be separated and manipulated at will; in fact, the method is not only good for shaving but also for adding a beard or even transferring a beard from one photo to another. If you find that you are losing your hair, you could potentially use this method to virtually enhance your photograph with a full set of luscious hair.

The CMU researchers utilized a machine learning method to predict what a person's face looks like under a beard making it possible to reconstruct features not visible in the original image. Using a large database of faces with and without beards (properly labeled for supervised learning,) the researchers learned two subspaces (one for each class of faces) and a model of the differences between the two subspaces. These are later used to transform a bearded face image to a non-bearded one and vice versa. Below is an example beard removal from the published paper.

The result shown above was obtained after training with a small number of images; the total number of images used was about 1200, i.e., a very small number for machine learning to be the most effective. As the authors proposed in their Eurographics 2008 paper, a larger set of images available for training should provide a large improvement on the final result. Finally, the same method is general enough to be used for other digital image manipulation tasks such as removing one's glasses.

IBM Sequoia supercomputer to be the fastest ever

In the world of supercomputers, IBM is easily the number one manufacturer. Last year, we wrote about the RoadRunner supercomputer which topped the charts as the fastest machine available for scientific computation housed at the Los Alamos Laboratory and most notably running PetaVision a piece of software used to simulate the human visual cortex. A couple of days ago, IBM funded by the US department of energy announced that they are building a new supercomputer that will be more 15-20 times faster than any other supercomputer in use today. Sequoia is slated to become the new king among the list of the top 500 supercomputers in use today throughout the world. The new computer will be housed at the US Department of Energy’s Lawrence Livermore National Laboratory in California.

The supercomputer named Sequoia will go online in 2012; yes, it takes a long time to build these computers. As it is commonly done, Sequoia will be constructed using thousands of CPUs connected together and working collectively to solve large scientific problems. According to information scattered around the Web, Sequoia will have some 1.6 million processors and the computational capacity of 20 petaflops that is 20 quadrillion calculations per second. When finished, the computer will occupy 3,422 square feet of space and require 6 megawatts of energy to operate (this apparently is far less than what supercomputers require today when measured with respect to the number of calculations performed against the energy used.) The actual cost of the computer has not been disclosed.

The computer will be used run massive simulations to keep track of the health of the country's nuclear stock; moreover, meteorologists will use Sequoia to more accurately predict the weather (maybe for once we can get some accurate forecasts,) study the cosmos and the human genome.