(Press-News.org) ITHACA, N.Y. – Robots are getting smarter, but they still need step-by-step instructions for tasks they haven't performed before. Before you can tell your household robot "Make me a bowl of ramen noodles," you'll have to teach it how to do that. Since we're not all computer programmers, we'd prefer to give those instructions in English, just as we'd lay out a task for a child.
But human language can be ambiguous, and some instructors forget to mention important details. Suppose you told your household robot how to prepare ramen noodles, but forgot to mention heating the water or tell it where the stove is.
In his Robot Learning Lab, Ashutosh Saxena, assistant professor of computer science at Cornell University, is teaching robots to understand instructions in natural language from various speakers, account for missing information, and adapt to the environment at hand.
Saxena and graduate students Dipendra K. Misra and Jaeyong Sung will describe their methods at the Robotics: Science and Systems conference at the University of California, Berkeley, July 12-16.
Video and abstract available at http://tellmedave.cs.cornell.edu
The robot may have a built-in programming language with commands like find (pan); grasp (pan); carry (pan, water tap); fill (pan, water); carry (pan, stove) and so on. Saxena's software translates human sentences, such as "Fill a pan with water, put it on the stove, heat the water. When it's boiling, add the noodles." into robot language. Notice that you didn't say, "Turn on the stove." The robot has to be smart enough to fill in that missing step.
Saxena's robot, equipped with a 3-D camera, scans its environment and identifies the objects in it, using computer vision software previously developed in Saxena's lab. The robot has been trained to associate objects with their capabilities: A pan can be poured into or poured from; stoves can have other objects set on them, and can heat things. So the robot can identify the pan, locate the water faucet and stove and incorporate that information into its procedure. If you tell it to "heat water" it can use the stove or the microwave, depending on which is available. And it can carry out the same actions tomorrow if you've moved the pan, or even moved the robot to a different kitchen.
Other workers have attacked these problems by giving a robot a set of templates for common actions and chewing up sentences one word at a time. Saxena's research group uses techniques computer scientists call "machine learning" to train the robot's computer brain to associate entire commands with flexibly defined actions. The computer is fed animated video simulations of the action –- created by humans in a process similar to playing a video game – accompanied by recorded voice commands from several different speakers.
The computer stores the combination of many similar commands as a flexible pattern that can match many variations, so when it hears "Take the pot to the stove," "Carry the pot to the stove," "Put the pot on the stove," "Go to the stove and heat the pot" and so on, it calculates the probability of a match with what it has heard before, and if the probability is high enough, it declares a match. A similarly fuzzy version of the video simulation supplies a plan for the action: Wherever the sink and the stove are, the path can be matched to the recorded action of carrying the pot of water from one to the other.
Of course the robot still doesn't get it right all the time. To test, the researchers gave instructions for preparing ramen noodles and for making affogato – an Italian dessert combining coffee and ice cream: "Take some coffee in a cup. Add ice cream of your choice. Finally, add raspberry syrup to the mixture."
The robot performed correctly up to 64 percent of the time even when the commands were varied or the environment was changed, and it was able to fill in missing steps. That was three to four times better than previous methods, the researchers reported, but "There is still room for improvement."
You can teach a simulated robot to perform a kitchen task at the "Tell me Dave" website, and your input there will become part of a crowdsourced library of instructions for the Cornell robots. Aditya Jami, visiting researcher at Cornell, is helping Tell Me Dave to scale the library to millions of examples. "With crowdsourcing at such a scale, robots will learn at a much faster rate," Saxena said.
INFORMATION:
Cornell University has television, ISDN and dedicated Skype/Google+ Hangout studios available for media interviews.
Video: Robot can be programmed by casually talking to it
2014-06-23
ELSE PRESS RELEASES FROM THIS DATE:
Biologists find 'missing link' in the production of protein factories in cells
2014-06-23
Biologists at UC San Diego have found the "missing link" in the chemical system that enables animal cells to produce ribosomes—the thousands of protein "factories" contained within each cell that manufacture all of the proteins needed to build tissue and sustain life.
Their discovery, detailed in the June 23 issue of the journal Genes & Development, will not only force a revision of basic textbooks on molecular biology, but also provide scientists with a better understanding of how to limit uncontrolled cell growth, such as cancer, that might be regulated by controlling ...
The brain's balancing act
2014-06-22
Researchers at the University of California, San Diego School of Medicine have discovered a fundamental mechanism by which the brain maintains its internal balance. The mechanism, described in the June 22 advanced online publication of the journal Nature, involves the brain's most basic inner wiring and the processes that control whether a neuron relays information to other neurons or suppresses the transmission of information.
Specifically, the scientists have shown that there is a constant ratio between the total amount of pro-firing stimulation that a neuron receives ...
Mysterious 'Magic Island' appears on Saturn moon
2014-06-22
ITHACA, N.Y. – Now you don't see it. Now, you do.
And now you don't see it again. Astronomers have discovered a bright, mysterious geologic object – where one never existed – on Cassini mission radar images of Ligeia Mare, the second-largest sea on Saturn's moon Titan. Scientifically speaking, this spot is considered a "transient feature," but the astronomers have playfully dubbed it "Magic Island."
Reporting in the journal Nature Geoscience June 22, the scientists say this may be the first observation of dynamic, geological processes in Titan's northern hemisphere. ...
Regional weather extremes linked to atmospheric variations
2014-06-22
Variations in high-altitude wind patterns expose particular parts of Europe, Asia and the US to different extreme weather conditions, a new study has shown.
Changes to air flow patterns around the Northern Hemisphere are a major influence on prolonged bouts of unseasonal weather – whether it be hot, cold, wet or dry.
The high altitude winds normally blow from west to east around the planet, but do not follow a straight path. The flow meanders to the north and south, in a wave-like path.
These wave patterns are responsible for sucking either warm air from the tropics, ...
Architecture of signaling proteins enhances knowledge of key receptors
2014-06-22
DURHAM, N.C. -- A team of scientists from Duke Medicine, the University of Michigan and Stanford University has determined the underlying architecture of a cellular signaling complex involved in the body's response to stimuli such as light and pain.
This complex, consisting of a human cell surface receptor and its regulatory protein, reveals a two-step mechanism that has been hypothesized previously but not directly documented.
The findings, reported on June 22, 2014, in the journal Nature, provide structural images of a G-protein coupled receptor (GPCR) in action.
"It ...
Study shows greater potential for solar power
2014-06-22
Concentrating solar power (CSP) could supply a large fraction of the power supply in a decarbonized energy system, shows a new study of the technology and its potential practical application.
Concentrating solar power (CSP) could supply a substantial amount of current energy demand, according to the study published in the journal Nature Climate Change. In the Mediterranean region, for example, the study shows that a connected CSP system could provide 70-80% of current electricity demand, at no extra cost compared to gas-fired power plants. That percentage is similar to ...
Microenvironment of hematopoietic stem cells can be a target for myeloproliferative disorders
2014-06-22
The discovery of a new therapeutic target for certain kinds of myeloproliferative disease is, without doubt, good news. This is precisely the discovery made by the Stem Cell Physiopathology group at the CNIC (the Spanish National Cardiovascular Research Center), led by Dr. Simón Méndez–Ferrer. The team has shown that the microenvironment that controls hematopoietic stem cells can be targeted for the treatment of a set of disorders called myeloproliferative neoplasias, the most prominent of which are chronic myelomonocytic leukemia (CMML), juvenile myelomonocytic leukemia ...
Family of proteins plays key role in cellular pump dynamics
2014-06-22
Case Western Reserve University scientists have discovered how a family of proteins — cation diffusion facilitators (CDFs) — regulates an important cellular cycle where a cell's energy generated is converted to necessary cellular functions. The finding has the potential to inform future research aimed at identifying ways to ensure the process works as designed and, if successful, could lead to significant breakthroughs in the treatment of Parkinson's, chronic liver disease and heart disease.
The results of this research were posted online June 22 by the journal Nature ...
Evidence found for the Higgs boson direct decay into fermions
2014-06-22
For the first time, scientists from the CMS experiment on the Large Hadron Collider (LHC) at CERN have succeeded in finding evidence for the direct decay of the Higgs boson into fermions. Previously, the Higgs particle could only be detected through its decay into bosons. "This is a major step forwards," explains Professor Vincenzo Chiochia from the University of Zurich's Physics Institute, whose group was involved in analyzing the data. "We now know that the Higgs particle can decay into both bosons and fermions, which means we can exclude certain theories predicting that ...
Protons power protein portal to push zinc out of cells
2014-06-22
Researchers at The Johns Hopkins University report they have deciphered the inner workings of a protein called YiiP that prevents the lethal buildup of zinc inside bacteria. They say understanding YiiP's movements will help in the design of drugs aimed at modifying the behavior of ZnT proteins, eight human proteins that are similar to YiiP, which play important roles in hormone secretion and in signaling between neurons.
Certain mutations in one of them, ZnT8, have been associated with an increased susceptibility to type 2 diabetes, but mutations that destroy its function ...