Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog

Category: robotics/AI – Page 424

On socially compliant navigation: Researchers show how real-world RL-based finetuning can enable mobile robots to adapt on the fly to the behavior of humans, to obstacles, and other challenges associated with real-world navigation:

Abstract.

We propose an online reinforcement learning approach, SELFI, to fine-tune a control policy trained on model-based learning. In SELFI, we combine the best parts of data efficient model-based learning with flexible model-free reinforcement learning, alleviating both of their limitations. We formulate a combined objective: the objective of the model-based learning and the learned Q-value from model-free reinforcement learning. By maximizing this combined objective in the online learning process, we improve the performance of the pre-trained policy in a stable manner. Main takeaways from our method are.

Continue reading “SELFI: Autonomous Self-improvement with Reinforcement Learning for Social Navigation” | >

A team of engineers, physicists, and data scientists from Princeton University and the Princeton Plasma Physics Laboratory (PPPL) have used artificial intelligence (AI) to predict—and then avoid—the formation of a specific type of plasma instability in magnetic confinement fusion tokamaks. The researchers built and trained a model using past experimental data from operations at the DIII-D National Fusion Facility in San Diego, Calif., before proving through real-time experiments that their model could forecast so-called tearing mode instabilities up to 300 milliseconds in advance—enough time for an AI controller to adjust operating parameters and avoid a tear in the plasma that could potentially end the fusion reaction.

Read more | >

DENVER—()—Palantir Technologies Inc. (NYSE: PLTR) today announced that the Army Contracting Command – Aberdeen Proving Ground (ACC-APG) has awarded Palantir USG, Inc. — a wholly-owned subsidiary of Palantir Technologies Inc. — a prime agreement for the development and delivery of the Tactical Intelligence Targeting Access Node (TITAN) ground station system, the Army’s next-generation deep-sensing capability enabled by artificial intelligence and machine learning (AI/ML). The agreement, valued at $178.4 million, covers the development of 10 TITAN prototypes, including five Advanced and five Basic variants, as well as the integration of new critical technologies and the transition to fielding.

“This award demonstrates the Army’s leadership in acquiring and fielding the emerging technologies needed to bolster U.S. defense in this era of software-defined warfare. Building on Palantir’s years of experience bringing AI-enabled capabilities to warfighters, Palantir is now proud to deliver the Army’s first AI-defined vehicle” Post this

TITAN is a ground station that has access to Space, High Altitude, Aerial, and Terrestrial sensors to provide actionable targeting information for enhanced mission command and long range precision fires. Palantir’s TITAN solution is designed to maximize usability for Soldiers, incorporating tangible feedback and insights from Soldier touch points at every step of the development and configuration process. Building off Palantir’s prior work delivering AI capabilities for the warfighter, Palantir is deploying the Army’s first AI-defined vehicle.

Read more | >

“If you look at the brain chemically, it’s like a soup with a bunch of ingredients,” said Dr. Fan Lam.

Can we map the brain to show its behavior patterns when a patient is healthy and sick? This is what a recent study published in Nature Methods hopes to address as a team of researchers at the University of Illinois Urbana-Champaign used a $3 million grant obtained from the National Institute of Aging to develop a novel approach to mapping brain behavior when a patient is both healthy and sick. This study holds the potential to help researchers, medical professionals, and patients better understand how to treat diseases.

“If you look at the brain chemically, it’s like a soup with a bunch of ingredients,” said Dr. Fan Lam, who is an assistant professor of bioengineering at the University of Illinois Urbana-Champaign and a co-author on the study. “Understanding the biochemistry of the brain, how it organizes spatiotemporally, and how those chemical reactions support computing is critical to having a better idea of how the brain functions in health as well as during disease.”

For the study, the researchers used a type of technology called spatial omics and combined this with deep learning to produce 3D datasets to unveil the brain’s myriad of characteristics down to the molecular level. Through this, the team has developed a novel method in monitoring brain activity when a patient is both healthy and sick, including the ability to identify complex neurological diseases.

Continue reading “3D Molecular Maps of the Brain: Unveiling Complexity with Spatial Omics” | >

Three weeks after surpassing Google’s parent company Alphabet to become the fourth most valuable company in the world, NVIDIA has now overtaken Saudi Aramco in market value.

This makes the AI chipmaker the third most valuable company in the world and in the United States, trailing only behind Apple Inc. and Microsoft Corp.

At the close of Friday’s trading session, NVIDIA’s market capitalization stood at $2.06 trillion, marking its inaugural venture above the $2 trillion threshold.

Read more | >