Lifeboat Foundation

Language is the most intuitive way for us to express how we feel and what we want. However, despite recent advancements in artificial intelligence, it is still very hard to control a robot using natural language instructions. Free-form commands such as “Robot, please go a little slower when you pass close to my TV” or “Stay far away from the swimming pool!” are hard to parse into actionable robot behaviors, and most human-robot interfaces today still rely on complex strategies such directly programming cost functions which define the desired behavior.

With our latest work, we attempt to change this reality through the introduction of “LaTTe: Language Trajectory Transformer”. LaTTe is a deep machine learning model that lets us send language commands to robots in an intuitive way with ease. When given an input sentence by the user, the model fuses it with camera images of objects that the robot observes in its surroundings, and outputs the desired robot behavior.

Continue reading “Just say the magic word: using language to program robots” »

face_with_colon_three circa 2020.

Watch any kid play with a tape measure and you’ll understand where researchers from the Seoul National University of Science and Technology found the inspiration for their new chameleon-like robot that can snag objects with its artificial tongue over 30 inches away in less than 600 milliseconds.

An instrument designed to detect seismic activity on Mars has just revealed an incredibly cool new ability, detecting faint tremors from meteorites impacting the red planet.

By combining data collected from the NASA’s Mars InSight lander with information from the Mars Reconnaissance Orbiter researchers have successfully linked those ground-shaking booms with freshly formed craters.

Not only does this help us understand the impact processes that continue to shape Martian geology, it demonstrates how collecting seismic data can reveal information beyond expected mission parameters. This may help inform future exploration of other worlds.

Continue reading “A Robot on Mars Detected The Tremors of Meteorites Hitting The Red Planet” »

With the advent of generative image AI, a new discipline is emerging: prompt engineering. Search engines like Krea provide inspiration.

Prompts are the short text descriptions that an image AI like DALL-E 2, Midjourney or Stable Diffusion uses to generate an image.

Often it is difficult to put the idea you have in mind into fitting words. Moreover, the AI may execute instructions differently than you want it to. And there are plenty of modifiers, for example for image cropping or style, that can lead to very different results for the otherwise same prompt.

Continue reading “AI image search engine finds great prompts for Stable Diffusion and co” »

ERNIE-ViLG, the new text-to-image AI developed by Baidu can generate images that show Chinese objects and celebrities more accurately than existing AIs. But a built-in censorship mechanism will filter out politically sensitive words.

Advancing the frontiers of machine learning.

Scientific publishers such as the American Association for Cancer Research (AACR) and Taylor & Francis have begun attempting to detect fraud in academic paper submissions with an AI image-checking program called Proofig, reports The Register. Proofig, a product of an Israeli firm of the same name, aims to help use “artificial intelligence, computer vision and image processing to review image integrity in scientific publications,” according to the company’s website.

During a trial that ran from January 2021 to May 2022, AACR used Proofig to screen 1,367 papers accepted for publication, according to The Register. Of those, 208 papers required author contact to clear up issues such as mistaken duplications, and four papers were withdrawn.

In particular, many journals need help detecting image duplication fraud in Western blots, which are a specific style of protein-detection imagery consisting of line segments of various widths. Subtle differences in a blot’s appearance can translate to dramatically different conclusions about test results, and many cases of academic fraud have seen unscrupulous researchers duplicate, crop, stretch, and rotate Western blots to make it appear like they have more (or different) data than they really do. Detecting duplicate images can be tedious work for human eyes, which is why some firms like Proofig and ImageTwin, a German firm, are attempting to automate the process.

Continue reading “AI software helps bust image fraud in academic papers” »

Ameca, a highly realistic android, has now been upgraded to include GPT-3, one of the largest neural networks and language prediction models.

Back in December 2021, UK-based Engineered Arts revealed what it described as “the most advanced android ever built” – a machine with strikingly lifelike motions and facial expressions. Since then, the company has been working to upgrade Ameca (as she is called) with speech and other capabilities.

In the video demonstration below, automated voice recognition has been combined with GPT-3, a large neural network and language prediction model that makes use of 175 billion parameters. This allows Ameca to recognise what people are saying and respond to questions. Before speaking, her output is fed to an online text-to-speech service, which generates the voice and visemes for lip sync timing.

The device can provide high-resolution 3D imaging of the neural network.

Researchers can now view the mouse brain through the skull thanks to a new holographic microscope. Led by Associate Director Choi Wonshik of the Center for Molecular Spectroscopy and Dynamics within the Institute for Basic Science, Professor Kim Moonseok of The Catholic University of Korea and Professor CHOI Myunghwan of Seoul National University developed a new type of holographic microscope.

The results were published in Science Advances on July 27.

Continue reading “A novel holographic microscope could image mouse brain through its skull” »