(Press-News.org) CAMBRIDGE, Mass. — Object recognition is one of the most widely studied problems in computer vision. But a robot that manipulates objects in the world needs to do more than just recognize them; it also needs to understand their orientation. Is that mug right-side up or upside-down? And which direction is its handle facing?
To improve robots' ability to gauge object orientation, Jared Glover, a graduate student in MIT's Department of Electrical Engineering and Computer Science, is exploiting a statistical construct called the Bingham distribution. In a paper they're presenting in November at the International Conference on Intelligent Robots and Systems, Glover and MIT alumna Sanja Popovic '12, MEng '13, who is now at Google, describes a new robot-vision algorithm, based on the Bingham distribution, that is 15 percent better than its best competitor at identifying familiar objects in cluttered scenes.
That algorithm, however, is for analyzing high-quality visual data in familiar settings. Because the Bingham distribution is a tool for reasoning probabilistically, it promises even greater advantages in contexts where information is patchy or unreliable. In ongoing work, Glover is using Bingham distributions to analyze the orientation of pingpong balls in flight, as part of a broader project to teach robots to play pingpong. In cases where visual information is particularly poor, his algorithm offers an improvement of more than 50 percent over the best alternatives.
"Alignment is key to many problems in robotics, from object-detection and tracking to mapping," Glover says. "And ambiguity is really the central challenge to getting good alignments in highly cluttered scenes, like inside a refrigerator or in a drawer. That's why the Bingham distribution seems to be a useful tool, because it allows the algorithm to get more information out of each ambiguous, local feature."
Because Bingham distributions are so central to his work, Glover has also developed a suite of software tools that greatly speed up calculations involving them. The software is freely available online, for other researchers to use.
In the rotation
One reason the Bingham distribution is so useful for robot vision is that it provides a way to combine information from different sources. Generally, determining an object's orientation entails trying to superimpose a geometric model of the object over visual data captured by a camera — in the case of Glover's work, a Microsoft Kinect camera, which captures a 2-D color image together with information about the distance of the color patches.
For simplicity's sake, imagine that the object is a tetrahedron, and the geometric model consists of four points marking the tetrahedron's four corners. Imagine, too, that software has identified four locations in an image where color or depth values change abruptly — likely to be the corners of an object. Is it a tetrahedron?
The problem, then, boils down to taking two sets of points — the model and the object — and determining whether one can be superimposed on the other. Most algorithms, Glover's included, will take a first stab at aligning the points. In the case of the tetrahedron, assume that, after that provisional alignment, every point in the model is near a point in the object, but not perfectly coincident with it.
If both sets of points in fact describe the same object, then they can be aligned by rotating one of them around the right axis. For any given pair of points — one from the model and one from the object — it's possible to calculate the probability that rotating one point by a particular angle around a particular axis will align it with the other. The problem is that the same rotation might move other pairs of points farther away from each other.
Glover was able to show, however, that the rotation probabilities for any given pair of points can be described as a Bingham distribution, which means that they can be combined into a single, cumulative Bingham distribution. That allows Glover and Popovic's algorithm to explore possible rotations in a principled way, quickly converging on the one that provides the best fit between points.
Big umbrella
Moreover, in the same way that the Bingham distribution can combine the probabilities for each pair of points into a single probability, it can also incorporate probabilities from other sources of information — such as estimates of the curvature of objects' surfaces. The current version of Glover and Popovic's algorithm integrates point-rotation probabilities with several other such probabilities.
In experiments involving visual data about particularly cluttered scenes — depicting the kinds of environments in which a household robot would operate — Glover's algorithm had about the same false-positive rate as the best existing algorithm: About 84 percent of its object identifications were correct, versus 83 percent for the competition. But it was able to identify a significantly higher percentage of the objects in the scenes — 73 percent versus 64 percent. Glover argues that that difference is because of his algorithm's better ability to determine object orientations.
He also believes that additional sources of information could improve the algorithm's performance even further. For instance, the Bingham distribution could also incorporate statistical information about particular objects — that, say, a coffee cup may be upside-down or right-side up, but it will very rarely be found at a diagonal angle.
Indeed, it's because of the Bingham distribution's flexibility that Glover considers it such a promising tool for robotics research. "You can spend your whole PhD programming a robot to find tables and chairs and cups and things like that, but there aren't really a lot of general-purpose tools," Glover says. "With bigger problems, like estimating relationships between objects and their attributes and dealing with things that are somewhat ambiguous, we're really not anywhere near where we need to be. And until we can do that, I really think that robots are going to be very limited."
###
END
Cold Spring Harbor, NY -- Think of the smell of freshly baking bread. There is something in that smell, without any other cues – visual or tactile – that steers you toward the bakery. On the flip side, there may be a smell, for instance that of fresh fish, that may not appeal to you. If you haven't eaten a morsel of food in three days, of course, a fishy odor might seem a good deal more attractive.
How, then, does this work? What underlying biological mechanisms account for our seemingly instant, almost unconscious ability to determine how attractive (or repulsive) ...
A study by Simon Fraser University researchers has found seniors in long-term care facilities are at high risk of head injuries – nearly 40 per cent of those who fall experience head impact.
The researchers studied video footage of 227 falls among 133 residents at a local long-term care facility. They found 37 per cent of falling residents struck their heads upon falling, and hit the ground – most often, linoleum or tile flooring – more than 60 per cent of the time. The researchers conclude: "By any measure, this is an alarmingly high prevalence."
More should be done, ...
Small farms and businesses may be the unintended victims of legislation aimed at cutting the federal budget by eliminating certain sets of local and county-based economic data, according to a group of economists.
"This local data is really what we use in our lab," said Stephan Goetz, professor of agricultural economics and regional economics, Penn State, and director of the Northeast Regional Center for Rural Development. "And, at the end of the day, we're using this information to try to understand how our world is changing."
The researchers, who report their findings ...
What makes some people more prone to wedded bliss or sorrow than others? Researchers at UC Berkeley and Northwestern University have found a major clue in our DNA. A gene involved in the regulation of serotonin can predict how much our emotions affect our relationships, according to a new study that may be the first to link genetics, emotions, and marital satisfaction. The study was conducted at UC Berkeley.
"An enduring mystery is, what makes one spouse so attuned to the emotional climate in a marriage, and another so oblivious?" said UC Berkeley psychologist Robert ...
CAMBRIDGE, Mass-- At first glance, Mars' clouds might easily be mistaken for those on Earth: Images of the Martian sky, taken by NASA's Opportunity rover, depict gauzy, high-altitude wisps, similar to our cirrus clouds. Given what scientists know about the Red Planet's atmosphere, these clouds likely consist of either carbon dioxide or water-based ice crystals. But it's difficult to know the precise conditions that give rise to such clouds without sampling directly from a Martian cloud.
Researchers at MIT have now done the next-best thing: They've recreated Mars-like ...
PROVIDENCE, R.I. [Brown University] — Home-delivered meals bring not only food to seniors but also the opportunity to remain in their homes. A new study by Brown University public health researchers projects that if every U.S. state in the lower 48 expanded the number of seniors receiving meals by just 1 percent, 1,722 more Medicaid recipients avoid living in a nursing home and most states would experience a net annual savings from implementing the expansion.
Pennsylvania would see the greatest net savings – $5.7 million – as Medicaid costs for nursing home care dropped ...
LA JOLLA, CA—October 7, 2013—Chemists at The Scripps Research Institute (TSRI) have devised a new technique for connecting drug molecules to antibodies to make advanced therapies.
Antibody-drug conjugates, as they're called, are the basis of new therapies on the market that use the target-recognizing ability of antibodies to deliver drug payloads to specific cell types—for example, to deliver toxic chemotherapy drugs to cancer cells while sparing most healthy cells. The new technique allows drug developers to forge more stable conjugates than are possible with current ...
Philadelphia, PA, October 7, 2013 – The recognition of a causal link between mutations in BRCA1 and BRCA2 genes and increased risk of developing breast and ovarian cancer has intensified the demand for genetic testing. Identifying mutations in these large genes by conventional methods can be time consuming and costly. A report in the November issue of the Journal of Molecular Diagnostics describes a new technique using second-generation sequencing technology that is as sensitive as the standard methodology but has the potential to improve the efficiency and productivity ...
Scientists from the Wildlife Conservation Society and the Kenyan Marine and Fisheries Research Institute have achieved a milestone in Africa: they've helped build a better fish trap, one that keeps valuable fish in while letting undersized juvenile fish and non-target species out.
By modifying conventional African basket traps with escape gaps, the marine researchers have proven that the new traps catch larger fish, allow more undersized and non-target fish to escape, increase profits, and—most importantly—minimize the impact of fishing on coastal reef systems. The findings, ...
Researchers from the University of Illinois at Urbana-Champaign have developed a new approach with applications in materials development for energy capture and storage and for optoelectronic materials.
According to Charles Schroeder, an assistant professor in the Department of Chemical and Biomolecular Engineering, the results show that peptide precursor materials can be aligned and oriented during their assembly into polypeptides using tailored flows in microfluidic devices.
The research was a collaboration between the labs of Schroeder and William Wilson, a research ...
