Abstract: "I will talk about three results that surprised me. First, I will show that the free configuration space of an elastic wire is path-connected, a result that makes easy a manipulation planning problem that was thought to be hard. Second, I will show a linear relationship between stimulation parameters, skin impedance, and sensation intensity in electrotactile stimulation. This result leads to algorithms that keep sensation intensity constant despite large variability in skin impedance, eliminating a longstanding barrier to practical use of electrotactile stimulation for sensory substitution and haptic feedback. Third, I will show you several obvious ways to use fiducial markers – which everybody knows will improve the performance of Structure-from-Motion (SfM) algorithms for vision-based 3D reconstruction – that work poorly. Then, I will show you a simple but less obvious way to use them that seems to work well. I will also talk about my experience teaching two engineering courses – one on robotics, one on control systems – to students incarcerated at Danville Correctional Center, an Illinois state prison. I will tell you why I did it and what I learned."

Google made a big leap into interactive entertainment on Tuesday as it announced an online video gaming platform that the company says represents the future of play. The streaming games service, called Stadia, allows users to run video games on sophisticated hardware maintained remotely by Google while directly controlling the action from their own devices over an Internet connection. Google's approach closely mirrors that of Netflix and on-demand video, a model that has since spread to e-books, music and live television. And it differs from existing platforms such as Valve's Steam, which require users to download and install a game before it can be played. At an industry conference in San Francisco, Google also debuted a console-style game controller designed to communicate with the new games service, which is built on top of the Google Chrome browser and runs from Google data centers located around the world.

Ford's first wave of autonomous vehicles will be produced at a new center in southeast Michigan, the company announced Wednesday, as part of a $900 million investment to reshape its manufacturing operations in the state. Workers there will begin installing self-driving technology into hybrid vehicles in 2021, the company said. "As we ramp up AV production, this plan allows us to adjust our investment spending to accommodate the pace of growth of this exciting new technology," Joe Hinrichs, Ford's president of global operations, said in a news release. "This new plan combines our core strength in mass manufacturing with the agility and leanness we've shown with our modification centers for specialty manufacturing." Ford's wider restructuring in Michigan is projected to create 900 jobs in the next four years.

Lyft is due to be the first giant tech startup to list its shares on the stock market this year -- and it has laid out all the roadblocks that could derail not only its own business, but the ridesharing industry itself. From dockless scooters and bicycles to self-driving cars, any number of transport modes Lyft has bet on could upend the ride-hailing business that it helped pioneer, according to documents Lyft filed as part of its IPO process. To the extent that Lyft is dependent on drivers, they also are a source of risk to the company, as well as potential regulation stemming from concern over increasingly crowded streets and curb space. Analysts said it was among the most candid assessments yet of the challenges facing the ride-hailing industry, which has historically employed as few as possible in its pursuit of the transportation market, instead depending on thousands of independent "driver-contractors" to support its business. The risks, which are mandatory for companies to detail as part of a stock market listing, are a kind of worst-case scenario for the business.

In the food industry, it seems, the robot revolution is well underway, with machines mastering skilled tasks that have always been performed by people. In Boston, robots have replaced chefs and are creating complex bowls of food for customers. In Prague, machines are displacing bartenders and servers using an app. Robots are even making the perfect loaf of bread these days, taking charge of an art that has remained in human hands for thousands of years. Now comes Briggo, a company that has created a fully automated, robotic brewing machine that can push out 100 cups of coffee in a single hour -- equaling the output of three to four baristas, according to the company.

Last fall, Howard Gordon, showrunner for the hit TV show "24" and co-creator of "Homeland," was in the middle of negotiating a deal with a newcomer to Hollywood: Apple. In a meeting with film executives from the iPhone maker, which had set aside an enormous budget to develop TV shows and movies, Gordon brought up rumors that Apple wanted sanitized projects -- a potential problem for his show about a pair of disillusioned military veterans who go on a killing spree. Would Apple buy the project, an adaptation of an Israeli show called "Nevelot," which translates directly to "Bastards," only to strip it of any violence and adult themes? "They were fairly adamant that it would not be a problem," Gordon said in an interview. "I said, 'Can you say that again into the microphone,'" Gordon said.

It might not be the first place you imagine when you think about robots. But in the Renaissance splendour of the Vatican, thousands of miles from Silicon Valley, scientists, ethicists and theologians gather to discuss the future of robotics. The ideas go to the heart of what it means to be human and could define future generations on the planet. The workshop, Roboethics: Humans, Machines and Health was hosted by The Pontifical Academy for Life. The Academy was created 25 years ago by Pope John Paul II in response to rapid changes in biomedicine.

We present an extension of the Kolmogorov-Smirnov (KS) two-sample test, which can be more sensitive to differences in the tails. Our test statistic is an integral probability metric (IPM) defined over a higher-order total variation ball, recovering the original KS test as its simplest case. We give an exact representer result for our IPM, which generalizes the fact that the original KS test statistic can be expressed in equivalent variational and CDF forms. For small enough orders ($k \leq 5$), we develop a linear-time algorithm for computing our higher-order KS test statistic; for all others ($k \geq 6$), we give a nearly linear-time approximation. We derive the asymptotic null distribution for our test, and show that our nearly linear-time approximation shares the same asymptotic null. Lastly, we complement our theory with numerical studies.

Online learning is the process of answering a sequence of questions based on the correct answers to the previous questions. It is studied in many research areas such as game theory, information theory and machine learning. There are two main components of online learning framework. First, the learning algorithm also known as the learner and second, the hypothesis class which is essentially a set of functions which learner uses to predict answers to the questions. Sometimes, this class contains some functions which have the capability to provide correct answers to the entire sequence of questions. This case is called realizable case. And when hypothesis class does not contain such functions is called unrealizable case. The goal of the learner, in both the cases, is to make as few mistakes as that could have been made by most powerful functions in hypothesis class over the entire sequence of questions. Performance of the learners is analysed by theoretical bounds on the number of mistakes made by them. This paper proposes three algorithms to improve the mistakes bound in the unrealizable case. Proposed algorithms perform highly better than the existing ones in the long run when most of the input sequences presented to the learner are likely to be realizable.

The goal in network state prediction (NSP) is to classify the global state (label) associated with features embedded in a graph. This graph structure encoding feature relationships is the key distinctive aspect of NSP compared to classical supervised learning. NSP arises in various applications: gene expression samples embedded in a protein-protein interaction (PPI) network, temporal snapshots of infrastructure or sensor networks, and fMRI coherence network samples from multiple subjects to name a few. Instances from these domains are typically ``wide'' (more features than samples), and thus, feature sub-selection is required for robust and generalizable prediction. How to best employ the network structure in order to learn succinct connected subgraphs encompassing the most discriminative features becomes a central challenge in NSP. Prior work employs connected subgraph sampling or graph smoothing within optimization frameworks, resulting in either large variance of quality or weak control over the connectivity of selected subgraphs. In this work we propose an optimization framework for discriminative subgraph learning (DSL) which simultaneously enforces (i) sparsity, (ii) connectivity and (iii) high discriminative power of the resulting subgraphs of features. Our optimization algorithm is a single-step solution for the NSP and the associated feature selection problem. It is rooted in the rich literature on maximal-margin optimization, spectral graph methods and sparse subspace self-representation. DSL simultaneously ensures solution interpretability and superior predictive power (up to 16% improvement in challenging instances compared to baselines), with execution times up to an hour for large instances.