Having trounced humans at everything from chess and Go, to StarCraft and Gran Turismo, artificial intelligence (AI) has raised its game and defeated world champions at a real-world sport.

The latest mortals to feel the sting of AI-induced defeat are three expert drone racers who were beaten by an algorithm that learned to fly a drone around a 3D race course at breakneck speeds without crashing. Or at least not crashing too often. — Read More

Empowering end-users to interactively teach robots to perform novel tasks is a crucial capability for their successful integration into real-world applications. For example, a user may want to teach a robot dog to perform a new trick, or teach a manipulator robot how to organize a lunch box based on user preferences. The recent advancements in large language models (LLMs) pre-trained on extensive internet data have shown a promising path towards achieving this goal. Indeed, researchers have explored diverse ways of leveraging LLMs for robotics, from step-by-step planning and goal-oriented dialogue to robot-code-writing agents.

While these methods impart new modes of compositional generalization, they focus on using language to link together new behaviors from an existing library of control primitives that are either manually engineered or learned a priori. Despite having internal knowledge about robot motions, LLMs struggle to directly output low-level robot commands due to the limited availability of relevant training data. As a result, the expression of these methods are bottlenecked by the breadth of the available primitives, the design of which often requires extensive expert knowledge or massive data collection.

In “Language to Rewards for Robotic Skill Synthesis”, we propose an approach to enable users to teach robots novel actions through natural language input.  — Read More

For decades, when people have imagined the distant future, they’ve almost always included a starring role for robots. Robots have been cast as dependable, helpful and even charming. Yet across those same decades, the technology has remained elusive — stuck in the imagined realm of science fiction.

Today, we’re introducing a new advancement in robotics that brings us closer to a future of helpful robots.  Robotics Transformer 2, or RT-2, is a first-of-its-kind vision-language-action (VLA) model. A Transformer-based model trained on text and images from the web, RT-2 can directly output robotic actions. Just like language models are trained on text from the web to learn general ideas and concepts, RT-2 transfers knowledge from web data to inform robot behavior.

In other words, RT-2 can speak robot. — Read More

The AI for Good global summit hosted by the U.N. tech agency invited a panel of robots and their creators to a press conference to answer questions from reporters.

At the AI for Good 2023 global summit, a panel of robots and their creators sat in front of the press to answer journalists’ questions on topics such as job automation, artificial intelligence (AI) leadership and collaboration with humans for a better future.

… Altogether nine robots were in attendance, including Sophia, who serves as the U.N. Development Program’s first robot innovation ambassador, a robot healthcare service provider named Grace and a rock star robot called Desdemona. — Read More

Last spring, engineers in Barcelona packed up the sperm-injecting robot they’d designed and sent it by DHL to New York City. They followed it to a clinic there, called New Hope Fertility Center, where they put the instrument back together, assembling a microscope, a mechanized needle, a tiny petri dish, and a laptop.

Then one of the engineers, with no real experience in fertility medicine, used a Sony PlayStation 5 controller to position a robotic needle. Eyeing a human egg through a camera, it then moved forward on its own, penetrating the egg and dropping off a single sperm cell. Altogether, the robot was used to fertilize more than a dozen eggs.

The result of the procedures, say the researchers, were healthy embryos—and now two baby girls, who they claim are the first people born after fertilization by a “robot.” Read More

“IT HAS AI NOW.”

As if robotic dogs weren’t creepy enough.

A team of programmers just equipped a Boston Dynamics robot dog with OpenAI’s ChatGPT and Google’s Text-to-Speech voice modulation, allowing it to literally speak to them and answer their questions. Read More