Artificial Intelligence Development

News

Eye of the beholder: Improving the human-robot connection

Researchers are programming robots to communicate with people using human-like body language and cues, an important step toward bringing robots into homes.

Researchers at the University of British Columbia enlisted the help of a human-friendly robot named Charlie to study the simple task of handing an object to a person. Past research has shown that people have difficulty figuring out when to reach out and take an object from a robot because robots fail to provide appropriate nonverbal cues.

"We hand things to other people multiple times a day and we do it seamlessly," says AJung Moon, a PhD student in the Department of Mechanical Engineering. "Getting this to work between a robot and a person is really important if we want robots to be helpful in fetching us things in our homes or at work."

Moon and her colleagues studied what people do with their heads, necks and eyes when they hand water bottles to one another. They then tested three variations of this interaction with Charlie and the 102 study participants.

Video available at: http://youtu.be/5AQ-E3njViw

Programming the robot to use eye gaze as a nonverbal cue made the handover more fluid. Researchers found that people reached out to take the water bottle sooner in scenarios where the robot moved its head to look at the area where it would hand over the water bottle or looked to the handover location and then up at the person to make eye contact.

"We want the robot to communicate using the cues that people already recognize," says Moon. "This is key to interacting with a robot in a safe and friendly manner."

This paper won best paper at the IEEE International Conference on Human-Robot Interaction.

Story Source:

The above story is based on materials provided by University of British Columbia. Note: Materials may be edited for content and length.

 

Scalable, universal quantum computer? Quantum information processed with system comprising optical photon and trapped atom

When it comes to recognizing complex patterns or to decoding encrypted messages, conventional computers reach their limits. A whole new quality in the communication and processing of data is expected from a technology that exploits the special properties of quantum particles such as superposition and entanglement. Scientists around the world pursue a variety of different concepts towards the development of such a quantum computer. Prof. Gerhard Rempe, Director at the MPQ and head of the Quantum Dynamics Division, follows the strategy of combining two rather dissimilar techniques: quantum communication using photons, and information processing using stationary atoms. His team has now for the first time realized a quantum logic gate between a single photon and a single atom. The development of this hybrid device could be a milestone on the path to a scalable and universal quantum computer.

Any modern computer operates according to a mathematical principle that was developed by German polymath Gottfried Wilhelm Leibniz more than 300 years ago: information can be encoded in the binary system and processed via the application of logic operators. Logic gates are based on this principle. They deterministically generate output signals for any combination of input signals according to a so-called truth table. Nowadays, computers contain many millions of logic gates in the form of electronic circuits.

In the experiment described here, the binary states 0 and 1 are represented by the two spin orientations of an atom (upwards or downwards), and by two polarization states of an optical photon (left or right circular), respectively. In contrast to classical bits, these "quantum bits" can be in a coherent superposition of both states. In order to realize a quantum gate, the atom is trapped inside a cavity which is made of two high-finesse mirrors. The properties of the cavity are chosen in such a way that atom and cavity form a strongly coupled system. The light quanta are prepared as faint laser pulses containing less than one photon on average.

In a former experiment it has been shown that -- by a proper choice of parameters -- the light quanta are always reflected. What matters is the fact that for certain combinations of atomic and photonic input states the photons are reflected at the first mirror. For other combinations, however, they first enter the cavity, subsequently leaving it on the same path. Thereby, they experience a phase shift of 180 degrees. "This conditional phase shift is the prerequisite for the implementation of a truth table assigning output signals to any combination of input bits in a deterministic way, similar to a classical logic gate.," Dr. Stephan Ritter explains.

"In our experiment we measure both the polarization of the reflected photons and the spin orientation of the atom after the gate operation. At present, we achieve an efficiency of about 70%. By further improving the mirror parameters this value could be significantly improved," Andreas Reiserer says.

These measurements demonstrate that the hybrid atom-photon system can act as a classical logic gate. However, the true advantage of a quantum gate compared to a classical one is its ability to generate entangled states from separable input states. In order to test this specific behaviour, the scientists chose a combination of input bits that -- according to the rules of quantum mechanics -- must lead to an entangled state of atom and photon after the gate operation. Also in this case the gate mechanism worked as expected.

By successively sending two laser pulses onto the system the physicists could even achieve entanglement between the atom and two photons. By clever manipulation of the atom in a second step it was disentangled, leaving a pair of two entangled photons. "These measurements demonstrate the versatility of the gate mechanism that even provides an interaction between two photons," Norbert Kalb says. "The mechanism should also allow generating entangled cluster states that consist of the atom and several photons."

The development of this hybrid quantum logic gate could be a big step towards a universal quantum computer. "Quantum communication, using flying photons, and data processing with atoms or ions have been regarded as separate research fields so far," Prof. Gerhard Rempe says. "In our experiment we merge both techniques. In particular, our quantum gate could be easily implemented in a network in which atoms serve as stationary nodes for the storage of information, whereas photons transmit the information between these nodes, even over large distances. In this way we hope to contribute to the realization of a scalable quantum computer."

Story Source:

The above story is based on materials provided by Max Planck Institute of Quantum Optics. The original article was written by Olivia Meyer-Streng. Note: Materials may be edited for content and length.

 

'RoboClam' hits new depths as robotic digger

A digging robot inspired by the unique mechanisms employed by the Atlantic razor clam has been created by a group of researchers in the US.

The robot, dubbed RoboClam, is able to dig with extreme efficiency by transforming the surrounding soil from a solid into a liquid, and could have a variety of applications from anchoring underwater robots to subsea cable installation and mine neutralization.

The first results of its performance have been published today, 9 April, in IOP Publishing's journal Bioinspiration & Biomimetics

The Atlantic razor clam, Ensis directus, is a large species of mollusc found on the North American coast which has a remarkable ability to burrow quickly and deeply into wet sand, easily out-performing any human digger.

According to the researchers, from the Massachusetts Institute of Technology, the Atlantic razor clam should only be able to submerge itself into the soil by about 2 cm with the 10 N of force that it is able to exert. Instead, the razor clam is able to dig 70 cm into the soil.

The efficiency is achieved by the opening and closing of the clam's valves, or shells, which agitate the surrounding soil and turn it into a fluid, therefore creating less resistance and making it easier for the clam to move downwards through the soil.

The researchers have recreated the opening and closing valve mechanism of the razor clam by creating a control platform in the robot which consists of two pneumatic pistons that move a 9 cm-long effector mimicking the razor clam's valves.

The RoboClam is controlled using a "genetic algorithm," which continuously records, and then configures, a number of different variables as the robot is working. The idea is that desirable traits -- in this case digging behaviours -- continue to get expressed through each generation of configuration, forcing the RoboClam to evolve, just as an organism does, into an optimized digging machine.

In their study, the researchers performed over 300 tests of the RoboClam in the razor clam's natural environment in ocean mudflats off the coast of Gloucester, MA.

The results showed that the RoboClam could achieve the same fluidization of soil as the Atlantic razor clam and was able to dig into the soil nearly as efficiently.

Lead author of the research, Professor Amos Winter, of the Department of Mechanical Engineering, said: "We have demonstrated that the robot is able to burrow into the soil with the same energy and depth relationship as the animal. Moving through static soil requires burrowing energy that scales with the square of depth. By fluidizing soil and reducing drag, razor clams and the RoboClam can burrow with energy that scales linearly with depth."

"There are many applications where a small, lightweight, low-power, reversible anchor would be very valuable. At the moment we are working with an underwater robotics company, Bluefin Robotics, who produce vehicles that need to remain stationary in a current, and could therefore benefit from a small anchor."

A video of RoboClam in action can be viewed here http://www.youtube.com/watch?v=bztw9PUiRss.

Story Source:

The above story is based on materials provided by Institute of Physics. Note: Materials may be edited for content and length.

   

Cybersecurity researchers roll out a new heartbleed solution: Red Herring creates decoy servers, entraps, monitors hackers

As companies scrambled in recent days to address the latest cybersecurity bug known as Heartbleed, researchers at The University of Texas at Dallas had a solution that fixes the vulnerability, and also detects and entraps hackers who might be using it to steal sensitive data.

The advanced technique -- dubbed Red Herring -- was created by a team led by Dr. Kevin Hamlen, an associate professor of computer science in the Erik Jonsson School of Computer Science and Engineering. It automates the process of creating decoy servers, making hackers believe they have gained access to confidential, secure information, when in fact their deeds are being monitored, analyzed and traced back to the source.

"Our automated honeypot creates a fixed Web server that looks and acts exactly like the original -- but it's a trap," said Hamlen, a member of the UT Dallas Cyber Security Research and Education Institute (CSI). "The attackers think they are winning, but Red Herring basically keeps them on the hook longer so the server owner can track them and their activities. This is a way to discover what these nefarious individuals are trying to do, instead of just blocking what they are doing."

The Heartbleed bug affects about two-thirds of websites previously believed to be secure. These are websites that use the computer code library called OpenSSL to encrypt supposedly secure Internet connections that are used for sensitive purposes such as online banking and purchasing, sending and receiving emails, and remotely accessing work networks. Heartbleed became public last week.

In 2012, a new feature named Heartbeat was added to software primarily for slow Internet connections. Heartbeat allowed connections to be held open, even during idle time. A flaw in the implementation allowed confidential information to be passed through the connection, hence the name Heartbleed.

Even though Heartbleed is now in the process of being fixed, victims face the challenge of not knowing who may already be exploiting it to steal the information, and what information they may be going after. A common fix for this type of problem is to create a trap, a honeypot that lures and exposes attackers. Typically this can involve setting up another Web server somewhere else.

"There are all sorts of ad hoc solutions where people try to confuse the attacker by deploying fake servers, but our solution builds the trap into the real server so that attacks against the real server are detected and monitored," Hamlen said. "Our research idea can build this honeypot really quickly and reliably as new vulnerabilities are disclosed."

The Red Herring algorithm created by Hamlen automatically converts a patch -- code widely used to fix new vulnerabilities like Heartbleed -- into a honeypot that can catch the attacker at the same time.

"When Heartbleed came out, this was the perfect test of our prototype," Hamlen said.

Red Herring doesn't stop at being a decoy and blocker; it can also lead to catching the attacker. As the attacker thinks he or she is stealing data, an analyst is tracking the attack to find out what information the attacker is after, how the malicious code works and who is sending the code.

"In their original disclosure, security firm Codenomicon urged experts to start manually building honeypots for Heartbleed," Hamlen said. "Since we already had created algorithms to automate this process, we had a solution within hours."

When news of Heartbleed became public on April 8, software engineering doctoral student Frederico Araujo started researching the vulnerability and had implemented Red Herring by 2:30 a.m. April 9.

"I was very proud that he had taken the initiative before I'd even gotten to it," Hamlen said. "Normally, I personally would have started working on it sooner, but I'd been up all night grading papers the night before."

Story Source:

The above story is based on materials provided by University of Texas, Dallas. Note: Materials may be edited for content and length.

 

Website information on colon cancer too complex, fails to address key concerns, researcher finds

Popular web information on colorectal cancer is too difficult for most lay people to read and doesn't address the appropriate risks to and concerns of patients, a study by UT Southwestern Medical Center gastroenterologists suggests.

In a review of a dozen popular websites, most of the online patient education materials for colorectal cancer screening were written beyond the recommended sixth-grade reading level, while content on the sites failed to address key risks, as well as the barriers to and benefits of screening, according to the study's authors Dr. Deepak Agrawal, Assistant Professor of Internal Medicine, and Dr. Chenlu Tian, an Internal Medicine fellow.

"Today, the Internet often is the first point of contact between the patient and health-related information, even for patients with low literacy. In, thus, is a great opportunity for us to influence the decisions people make about their health and to steer them in the right direction. Informing patients is a physician's responsibility and we take this role seriously," said Dr. Agrawal, senior author and gastroenterologist in the division of Digestive and Liver Diseases at UT Southwestern.

The study, which appears in the journal Gastrointestinal Endoscopy and was conducted along with researchers from UT Arlington, is the first to analyze the appropriateness of internet-based materials in terms of health literacy.

Colorectal cancer remains the second leading cause of cancer death in the United States, despite effective screening tests, including the fecal occult blood test (FOBT), flexible sigmoidoscopy, and colonoscopy. Developing more appropriate and targeted patient education resources on colorectal cancer may improve patient understanding and promote screening, an important goal because U.S. colorectal cancer screening rates remain below target levels, the authors said.

The "Healthy People 2010" initiative proposed a colon cancer screening rate of at least 70.5 percent. Colon cancer screening rates remain under 50 percent for those with less than high school reading levels, well below the overall average of about 65 percent, according to the study.

Previous studies have shown that six of 10 people rely on the Internet when seeking information about colonoscopy screening. Yet, readability for 10 of the 12 sites reviewed -- all from reputable medical societies and considered likely referral sites for physicians -- surpassed the maximum recommended sixth-grade reading level.

In addition to being too hard to read, the researchers found, the sites failed to address key concerns, such as the risk of getting colon cancer, the chances of dying from colon cancer, and how easy would it be to get screening. The review found that only half of the sites discussed colorectal cancer risk in the general population and only a quarter specifically addressed patients at high risk, such as African Americans, smokers, patients with diabetes, and obese patients.

The sites also failed to adequately address other common barriers to screening. For example, less than 10 percent of the sites addressed embarrassment, a common concern, and only a quarter addressed pain associated with colonoscopy or the costs of the procedure. None specifically mentioned the need for colonoscopy when no symptoms are present. "It is important to add that reading information on a website should not be considered a substitute for consulting a physician," Dr. Agrawal said. "Internet information is best used as a supplement. With colon cancer screening, there are many options and each has its risks and benefits. An actual discussion with a physician would help patients choose the best option."

Story Source:

The above story is based on materials provided by UT Southwestern Medical Center. Note: Materials may be edited for content and length.

   

Page 9 of 28

Our Partners