This week’s list of data news highlights covers February 13-19, 2016 and includes articles about Ford’s new pothole detection and avoidance system and an algorithm that could help humans and robots work together more effectively.
1. Building an Earth Observation Satellite Network
The European Space Agency (ESA) has launched the Sentinel-3A satellite to support the European Union’s Copernicus Earth observation program, a satellite network to collect data on environmental conditions supporting applications ranging from studying climate change to improving shipping routes. Sentinel-3A is the third satellite ESA has launched for Copernicus, making the system capable of surveying the entire planet in just two days. When ESA launches the next satellite, Sentinel-3B, in 2017, Copernicus will be able to survey the planet in less than one day, providing researchers with much more granular and timely data.
2. Preventing Domestic Violence with Machine Learning
Researchers at the University of Pennsylvania have developed a machine learning algorithm capable of predicting the likelihood a person accused of domestic violence will reoffend if released. The researchers trained their algorithm on data from 28,000 domestic violence arraignments that took place from January 2007 to October 2011 and incorporated other factors such as age and prior charges. In the arraignment process, a judge decides whether or not to detain an offender, and 20 percent of offenders released go on to reoffend. After backtesting their algorithm, researchers were able to demonstrate that if judges were to consult their algorithm’s predictions, they could reduce this rate by half.
3. Teaching a Car to Avoid Potholes
Ford has developed a system of sensors capable of scanning the road ahead of a car for potholes and automatically adjust a wheel’s shock absorbers to prevent the tire from harshly hitting the bottom of the hole, reducing the damage potholes can do to cars. In a demonstration, the system was effective enough to allow a car to drive over a pothole without its tires causing damage to ping pong balls placed at the bottom of the hole. In the United States, pothole damage is estimated to cost drivers a total $3 billion in vehicle repairs per year.
4. Giving Data a Shelf-Life of 13.8 Billion Years
Researchers at the University of Southampton have created a new method for storing large amounts of data on small glass discs. Standard data storage mediums, such as CDs, rely on reflecting light to transmit information—whether or not light is reflected at a particular location constitutes two dimensions. The researchers were able to build small structures called nanogratings in glass discs that can convey information in five dimensions—the orientation of the nanograting, the strength of the reflection, and the nanograting’s location on the x, y, and z axes. Each 1-inch disc can hold 360 terabytes of data and, because glass is durable and chemically stable, the data will remain intact for up to 13.8 billion years even at extreme temperatures.
5. Charting the Course for the Future of Smart Manufacturing
The U.S. Secretary of Commerce, in conjunction with the Executive Office of the President and the National Science and Technology Council, has published the National Network for Manufacturing Innovation’s (NNMI) Strategic Plan, outlining the initiative’s priorities for the next three years. NNMI is a network of research institutions launched by President Obama in 2012 to accelerate the development and commercialization of smart manufacturing technologies through public-private partnerships.
6. Helping Humans and Robots Collaborate
Researchers at the Massachusetts Institute of Technology have developed an algorithm that can make communications between humans and robotic systems more efficient, which could increase the effectiveness of emergency response teams and other efforts that typically rely on robots. With multiple robotic systems working together, frequent communication can be very resource intensive as it requires systems to constantly process and report new information, and much of it may be overwhelming for humans. The researchers’ algorithm allows robotic systems to evaluate new information to determine if it is important enough to report it to others on the team, reducing unnecessary communication in a test environment by 60 percent without a significant decrease in performance.
7. Preventing Suicide on Social Media
Facebook has partnered with emotional support charity Samaritans to develop a new tool for its users in the United Kingdom to anonymously report worrisome content they think might indicate the person who posted the content is considering suicide. What a user’s post is flagged, Facebook will notify them that a friend is worried and provide options to reach out to a friend or connect to a Samaritans’ helpline. Facebook deployed the service in the UK after successful pilots in the United States and Australia last year.
8. Making GPS Accurate Enough for Self Driving Cars
Researchers at the University of California, Riverside have developed a new method of processing location data to improve global positioning system (GPS) accuracy from one meter to one centimeter. GPS normally relies on data from satellites and ground-based reference stations to determine the location of a device. These estimates can be made more accurate by using inertial data from additional sensors on-board the device, though combining these data sources requires complex calculations too demanding for consumer-grade navigation systems. However, researchers developed a series of algorithms that make these calculations substantially less difficult, making it possible for consumer devices and self-driving cars, which require precise location data, to take advantage of this more accurate system.
9. Sorting Through Scripts with Artificial Intelligence
ScriptHop, a startup launched by a former talent agency director, has launched a new artificial intelligence service that can analyze movie scripts and provide detailed character breakdowns in just four seconds, significantly faster than the four hours it would take a human. With ScriptHop, an actor or agent can quickly process scripts to identify particular character traits, such as characters of a specific age, height, or race, as well as screen time, without having to read through lengthy scripts.
10. Boosting Manufacturing with Supercomputing
The U.S. Department of Energy (DOE) has selected 10 projects proposed by private sector manufacturers to provide with access to supercomputers at U.S. national labs. As part of its High Performance Computing for Manufacturing (HPC4Mfg) program, the DOE will allow manufacturers to use supercomputers to run advanced modeling and analytics programs that would otherwise be too financially or technically demanding for them to carry out on their own. The projects all focus on improving manufacturing efficiency, product development, and reducing energy use.
Image: U.S. Department of Energy.