AI that can follow a person seems like a simple enough task. It's certainly a simple thing to ask a human to do, but what if people or objects get in the way of the robot following behind a person? How do you navigate an environment that's in a constant state of change? About a year ago at a robotics conference TechCrunch held at UC Berkeley, AI startup founders explored solutions for common problems encountered when trying to automate construction projects. Tessa Lau, CEO of Dusty Robotics, called attention to the challenge of moving machines in an unstructured environment filled with people.

There is growing interest in data-driven weather prediction (DDWP), for example using convolutional neural networks such as U-NETs that are trained on data from models or reanalysis. Here, we propose 3 components to integrate with commonly used DDWP models in order to improve their physical consistency and forecast accuracy. These components are 1) a deep spatial transformer added to the latent space of the U-NETs to preserve a property called equivariance, which is related to correctly capturing rotations and scalings of features in spatio-temporal data, 2) a data-assimilation (DA) algorithm to ingest noisy observations and improve the initial conditions for next forecasts, and 3) a multi-time-step algorithm, which combines forecasts from DDWP models with different time steps through DA, improving the accuracy of forecasts at short intervals. To show the benefit/feasibility of each component, we use geopotential height at 500~hPa (Z500) from ERA5 reanalysis and examine the short-term forecast accuracy of specific setups of the DDWP framework. Results show that the equivariance-preserving networks (U-STNs) clearly outperform the U-NETs, for example improving the forecast skill by $45\%$. Using a sigma-point ensemble Kalman (SPEnKF) algorithm for DA and U-STN as the forward model, we show that stable, accurate DA cycles are achieved even with high observation noise. The DDWP+DA framework substantially benefits from large ($O(1000)$) ensembles that are inexpensively generated with the data-driven forward model in each DA cycle. The multi-time-step DDWP+DA framework also shows promises, e.g., it reduces the average error by factors of 2-3.

This story was originally published by Undark and is reproduced here as part of the Climate Desk collaboration. Deep in the Mojave Desert, 60 miles from the city of Barstow, is the Slash X Ranch Cafe, a former ranch where dirt bike riders and ATV adventurers can drink beer and eat burgers with fellow daredevils speeding across the desert. Displayed on a wall alongside trucker caps and taxidermy is a plaque that memorializes the 2004 DARPA Grand Challenge, a 142-mile race whose starting point was at Slash X Ranch Cafe. It was the first race in the world without human drivers. Instead, it featured the fever-dream inventions -- robotic motorcycles, monster Humvees -- of a handful of software engineers who were hellbent on creating fully autonomous vehicles and winning the million-dollar prize offered by the Defense Department's Defense Advanced Research Projects Agency.

The manifold Helmholtzian (1-Laplacian) operator $\Delta_1$ elegantly generalizes the Laplace-Beltrami operator to vector fields on a manifold $\mathcal M$. In this work, we propose the estimation of the manifold Helmholtzian from point cloud data by a weighted 1-Laplacian $\mathbf{\mathcal L}_1$. While higher order Laplacians ave been introduced and studied, this work is the first to present a graph Helmholtzian constructed from a simplicial complex as an estimator for the continuous operator in a non-parametric setting. Equipped with the geometric and topological information about $\mathcal M$, the Helmholtzian is a useful tool for the analysis of flows and vector fields on $\mathcal M$ via the Helmholtz-Hodge theorem. In addition, the $\mathbf{\mathcal L}_1$ allows the smoothing, prediction, and feature extraction of the flows. We demonstrate these possibilities on substantial sets of synthetic and real point cloud datasets with non-trivial topological structures; and provide theoretical results on the limit of $\mathbf{\mathcal L}_1$ to $\Delta_1$.

Ocean scientists are using robot submariness to detect barrels of toxic chemicals under the sea. Thousands of barrels of DDT and other substances are believed submerged in the Pacific Ocean near Los Angeles, but authorities aren't sure where or how many. To get an idea, researchers have launched two'underwater Roombas,' Remote Environmental Monitoring UnitS (REMUS) that can operate in waters ranging from 80 feet to about 20,000 feet. The vehicles take 12 hours to recharge, so while one is scanning the seafloor with its sonar the other is powering up and passing along its findings. Ocean scientists are using'underwater Roombas' to scan the ocean floor for barrels of toxic chemicals, including the banned pesticide DDT.

Civil liberties activists are suing a company that provides facial recognition services to law enforcement agencies and private companies around the world, contending that Clearview AI illegally stockpiled data on 3 billion people without their knowledge or permission. The lawsuit, filed in Alameda County Superior Court in the San Francisco bay area, says the New York company violates California's constitution and seeks a court order to bar it from collecting biometric information in California and requiring it to delete data on Californians. The lawsuit says the company has built "the most dangerous" facial recognition database in the nation, has fielded requests from more than 2,000 law enforcement agencies and private companies and has amassed a database nearly seven times larger than the FBI's. Separately, the Chicago Police Department stopped using the New York company's software last year after Clearview AI was sued in Cook County by the American Civil Liberties Union. The California lawsuit was filed by four activists and the groups Mijente and Norcal Resist.

Extracting interesting scenarios from real-world data as well as generating failure cases is important for the development and testing of autonomous systems. We propose efficient mechanisms to both characterize and generate testing scenarios using a state-of-the-art driving simulator. For any scenario, our method generates a set of possible driving paths and identifies all the possible safe driving trajectories that can be taken starting at different times, to compute metrics that quantify the complexity of the scenario. We use our method to characterize real driving data from the Next Generation Simulation (NGSIM) project, as well as adversarial scenarios generated in simulation. We rank the scenarios by defining metrics based on the complexity of avoiding accidents and provide insights into how the AV could have minimized the probability of incurring an accident. We demonstrate a strong correlation between the proposed metrics and human intuition.

While all eyes were on the dramatic descent of NASA's Perseverance rover last month, a team sent a robot into another alien world, one closer to home: the deep sea. With its towering undersea mountains, dramatic geological features, and unique creatures--many of which remain mysterious--the deep sea is the last uncharted environment on Earth. Sinking any intrepid explorer into blackened waters means facing freezing temperatures and crushing pressure. Ever listened to the sound of metal creaking under pressure? Without protection, puny electronic components in a robot don't have a chance.

San Francisco-based Ample announced a new battery charging technology that refuels electric vehicles from any automaker in just 10 minutes – three times faster than traditional systems. Using Modular Battery system, AI-powered robots remove the depleted battery and replace it with a fully charged unit – Ample says its batteries are like Lego-blocks that can accommodate any vehicle. Ample, started by Ex-Tesla and Google engineers, has constructed five battery swap stations in the San Francisco Bay Area, which can fit in two parking spots, specifically for Uber drivers. The technology comes as Tesla had promised deliver electric vehicle battery swapping stations in 2013, but the Elon Musk-owned company did not deliver - so the start-up moved to make it happen. 'Hopefully this is what convinces people finally that electric cars are the future,' Musk said, rallying a crowd at a splashy demo in 2013, also noting that battery of a Tesla Model S could be swapped in about 90 seconds. However, Musk said in 2015 that Tesla owners were not interested in swapping batteries and pulled the plug on pursuing a batter swapping station.

This silicone rubber robot can withstand the pressures in the ocean's deepest abyss A silicone robot has survived a journey to 10,900 metres below the ocean's surface in the Mariana trench, where the crushing pressure can implode all but the strongest enclosures. This device could lead to lighter and more nimble submersible designs. A team led by Guorui Li at Zhejiang University in China based the robot's design on snailfish, which have relatively delicate, soft bodies and are among the deepest living fish. They have been observed swimming at depths of more than 8000 metres. The submersible robot looks a bit a manta ray and is 22 centimetres long and 28 centimetres in wingspan.