Posts Tagged ‘simulation’

Dreaming Robot Can’t Wake Up

ZaZa can't wake from dreams of being splashed.

Do androids dream of electric sheep? Apparently not, according to researchers at the Interdisciplinary Technology Institute in Boston. “Most of ZaZa’s dreams are about getting splashed.” Whether that’s the result of oversight or foresight, ZaZa’s vulnerability to liquids has been a boon to the scientists at ITI from the beginning of this uncanny project.

As part of their research into the relationship between dreaming and memory formation, ZaZa was created to be a learning robot. Scientists expose her to new stimuli and information every day and then assess her recollection over time. According to one researcher, ZaZa came in contact with an uncovered cup of coffee during the first week of the study, and immediately afterward, her dreams became less like randomized input logs and more recognizably dream-like. “After the incident, we instructed her to avoid all moisture in the future. We expected to see the event reflected in her dreams but we’re still surprised by the extent of its impact.”

Because organic brains are still far more complex than even the most advanced computers, scientists at ITI had to overcome major hurdles while designing the project. In order to construct a useful model of a human mind, they had to give ZaZa several ‘brains.’ In addition to the central unit in her chest, she has a computer to regulate and monitor each of her six sensor-types, another to coordinate the senses and simulate short term memory and recall, and a ninth computer dedicated to communication and dreaming. One would expect a robot brain composed of so many computers to be cumbersome and awkward, but ZaZa is surprisingly small; about the size of a kindergartner. Because ITI’s dream research requires that ZaZa be able to move around and interact with scientists, they took advantage of existing, inexpensive broadband mobile technology rather than reinventing the wheel for the project. As a result, little ZaZa is completely wireless and has a remote brain.

Wifi is only one of the technological advances that researchers at ITI have incorporated into the design of their dream-bot.  To give ZaZa the ability to learn like a human, they applied developments in self-organizing computer networks, simulated cognition and artificial intelligence, and even language acquisition and physical creativity. “The ZaZa Project is really a collaborative effort between ITI and dozens of other institutions. The individual advances made in their labs are brought together in ours.”

Considerable effort went into ZaZa’s outward design, as well. In order to inspire more ‘life-like’ dreams, they’ve equipped her with the social skills necessary to recognize human emotions and respond appropriately. For day-to-day interactions between ZaZa and researchers to be as normal as possible, they’ve even given her human mannerisms and appearance. One scientist said, “When all her systems are functioning optimally, you could almost forget she isn’t someone’s little girl.”

Of course, any system as complex as ZaZa’s is bound to malfunction at times. Because most of her brains are located outside her body, she slips into standby mode whenever the local wifi signal drops and must be woken manually. If even one of her computers crashes, scientists must shut her down completely and repeat the day’s research from the beginning. Simple physical problems, like replacing worn sensors, can be dealt with more easily because ZaZa doesn’t technically feel pain. However, every time she gets wet – a month after the coffee incident, poor ventilation in another lab at ITI triggered the fire sprinklers – ZaZa’s body suffers catastrophic failure and must be rebuilt.

When everything goes according to plan, ZaZa is still only awake for eight hours a day, five days a week. “ZaZa can’t be left unattended while she’s awake, so she has to dream while we’re all home on nights and weekends,” explained the project’s lead scientist. That’s perfect for their research because it means that during periods without major malfunctions, they still acquire enough dream logs to make up for the time they spend rebuilding and repairing her systems.

Scientists are naturally reluctant to offer much speculation about the results of this study so early in the project, but many researchers are already planning future studies involving ZaZa and conceiving new robots based upon her design. One such project has already been green-lighted by ITI, but the only details scientists would divulge about it were that the next generation of ‘dream-bot’ will be adult-sized to accommodate internal, self-contained computer brains. Also, unlike ZaZa, who spends most hours lying under a tarp unable to wake from dreams about getting splashed, their next prototype will be able to swim if necessary, and may rest, but never sleep.

Fully Functional and Anatomically Correct

A few months ago, my sci-fi short, Pieces of You, was published in M-Brane SF magazine. It’s a fairly straightforward coming of age story about a boy and his adoptive android guardian. I knew when I wrote it that I didn’t have everything right about robots, and of course I doubt humans will ever replace existing foster care systems with highly sophisticated machinery. But given recent developments in human simulation, I won’t be shocked to find myself sharing an apartment with a social robot of some kind in the next few years. Even if it is just a cantankerous Roomba.

While doing research for another android guardian story I came across some useful information. As in human colleges, a lot of funding goes toward figuring out how to make better robot athletes, although in this case it’s a good thing because it gives us another reason and perspective from which to consider human and animal locomotion.

But do people really want their appliances to be smarter and more independent than a cat-friendly auto-vac? Yes, especially older people. And if future robots can augment our ability to take care of ourselves, might they also help us take better care of each other? As with engineering robotic football all-stars, developing robots that can simulate emotion is useful because it gives us a new angle for examining how we function.

Maybe house-bots will never amount to more than personal or medical assistants and glorified mechanical pets. And really, if we want much more than that from the machines in our homes, we should probably take the opportunity to re-evaluate the human social structures that make automating our most intimate affairs seem necessary.

Fair warning, the video below is Not Safe For Work!