Connect with us

Biotech

Engineers Teach AI to Navigate Ocean With Minimal Energy

Published

on

Engineers at Caltech, ETH Zurich, and Harvard are developing an artificial intelligence (AI) that will allow autonomous drones to use ocean currents to aid their navigation, rather than fighting their way through them.

Engineers at Caltech, ETH Zurich, and Harvard are developing an artificial intelligence (AI) that will allow autonomous drones to use ocean currents to aid their navigation, rather than fighting their way through them.

“When we want robots to explore the deep ocean, especially in swarms, it’s almost impossible to control them with a joystick from 20,000 feet away at the surface. We also can’t feed them data about the local ocean currents they need to navigate because we can’t detect them from the surface. Instead, at a certain point we need ocean-borne drones to be able to make decisions about how to move for themselves,” says John O. Dabiri (MS ’03, PhD ’05), the Centennial Professor of Aeronautics and Mechanical Engineering and corresponding author of a paper about the research that was published by Nature Communications on December 8.

The AI’s performance was tested using computer simulations, but the team behind the effort has also developed a small palm-sized robot that runs the algorithm on a tiny computer chip that could power seaborne drones both on Earth and other planets. The goal would be to create an autonomous system to monitor the condition of the planet’s oceans, for example using the algorithm in combination with prosthetics they previously developed to help jellyfish swim faster and on command. Fully mechanical robots running the algorithm could even explore oceans on other worlds, such as Enceladus or Europa.

In either scenario, drones would need to be able to make decisions on their own about where to go and the most efficient way to get there. To do so, they will likely only have data that they can gather themselves–information about the water currents they are currently experiencing.

To tackle this challenge, researchers turned to reinforcement learning (RL) networks. Compared to conventional neural networks, reinforcement learning networks do not train on a static data set but rather train as fast as they can collect experience. This scheme allows them to exist on much smaller computers–for the purposes of this project, the team wrote software that can be installed and run on a Teensy–a 2.4-by-0.7-inch microcontroller that anyone can buy for less than $30 on Amazon and only uses about a half watt of power.

Using a computer simulation in which flow past an obstacle in water created several vortices moving in opposite directions, the team taught the AI to navigate in such a way that it took advantage of low-velocity regions in the wake of the vortices to coast to the target location with minimal power used. To aid its navigation, the simulated swimmer only had access to information about the water currents at its immediate location, yet it soon learned how to exploit the vortices to coast toward the desired target. In a physical robot, the AI would similarly only have access to information that could be gathered from an onboard gyroscope and accelerometer, which are both relatively small and low-cost sensors for a robotic platform.

This kind of navigation is analogous to the way eagles and hawks ride thermals in the air, extracting energy from air currents to maneuver to a desired location with the minimum energy expended. Surprisingly, the researchers discovered that their reinforcement learning algorithm could learn navigation strategies that are even more effective than those thought to be used by real fish in the ocean.

“We were initially just hoping the AI could compete with navigation strategies already found in real swimming animals, so we were surprised to see it learn even more effective methods by exploiting repeated trials on the computer,” says Dabiri.

The technology is still in its infancy: currently, the team would like to test the AI on each different type of flow disturbance it would possibly encounter on a mission in the ocean–for example, swirling vortices versus streaming tidal currents–to assess its effectiveness in the wild. However, by incorporating their knowledge of ocean-flow physics within the reinforcement learning strategy, the researchers aim to overcome this limitation. The current research proves the potential effectiveness of RL networks in addressing this challenge–particularly because they can operate on such small devices. To try this in the field, the team is placing the Teensy on a custom-built drone dubbed the “CARL-Bot” (Caltech Autonomous Reinforcement Learning Robot). The CARL-Bot will be dropped into a newly constructed two-story-tall water tank on Caltech’s campus and taught to navigate the ocean’s currents.

“Not only will the robot be learning, but we’ll be learning about ocean currents and how to navigate through them,” says Peter Gunnarson, graduate student at Caltech and lead author of the Nature Communicationspaper.

The paper is titled “Learning efficient navigation in vortical flow fields.” Co-authors include Ioannis Mandralis, graduate student at Caltech, Guido Novati of ETH Zurich in Switzerland, and Petros Koumoutsakos (PhD ’92) of Harvard University. This research was funded by a National Science Foundation Graduate Fellowship to Gunnarson and by NSF Waterman Award funding to Dabiri.

Read More

Article: bioengineer.org

Biotech

Are Researchers One Step Closer to Developing the Theory of Impulse Circuits?

Published

on

Computers play an important role in many aspects of life today. Digital computers are the most widely used, while quantum computers are well known. However, the least known computers are the so-called Stochastic Pulse Computers. Their work is based on highly parallel logical operations between trains of electrical pulses, where the pulses occur at random times, as in neurons, the nerve cells in the brains of humans and mammals.

Computers play an important role in many aspects of life today. Digital computers are the most widely used, while quantum computers are well known. However, the least known computers are the so-called Stochastic Pulse Computers. Their work is based on highly parallel logical operations between trains of electrical pulses, where the pulses occur at random times, as in neurons, the nerve cells in the brains of humans and mammals.

The main motivation for the growing interest in research on RPC computers over the past decade is the hope that they could solve faster and with less energy consumption tasks that are normally easy for living beings, but difficult for digital computers, such as instantaneous responses to stimuli, pattern recognition, robustness to errors and damage in the system, learning, and autonomy.

In a recent study published in Scientific Reports, researchers from the Croatian Centre of Excellence for Advanced Materials and Sensors, Dr Mario Stipčević of the Ruđer Bošković Institute (RBI) and Mateja Batelić, a student at the Faculty of Science at the University of Zagreb (FS), Croatia, describe new or improved versions of RPC circuits that use quantum randomness for the first time, but also go a significant step further and lay the first foundation for RPC circuit theory.

Namely, while circuits for processing information in a digital computer can be assembled from logic circuits as building blocks based on the well-known Boolean theory, a similar theory for RPC circuits does not yet exist. Therefore, the synthesis of circuits for an RPC is limited to trial and error through experimentation or simulation.

‘’The central part of our paper is the formulation and proof of the so-called entropy budget theorem, which can be used to easily verify whether a given mathematical (or logical) operation can be performed or “calculated” by any physical circuit, and if so, how much excess entropy must be available to a circuit in order to perform the given operation.

In this paper, we demonstrate the theorem using several examples of mathematical operations. Perhaps the most interesting proof is the existence of a deterministic half-sum circuit (a + b) / 2. However, this circuit is not yet known, and finding it is a challenge for further research,” says Mario Stipčević, head of the Laboratory of Photonics and Quantum Optics at the Ruđer Bošković Institute.

Feedzy

Source: bioengineer.org

Continue Reading

Biotech

PSMA PET Validates EAU Classification System to Determine Risk of Prostate Cancer Recurrence

Published

on

Reston, VA (January 20, 2022)—New research has confirmed the accuracy of the novel European Association of Urology (EAU) risk classification system that groups prostate cancer patients based on their risk of recurrence. Prostate-specific membrane antigen (PSMA) PET imaging of men with prostate cancer validated the EAU groupings and provided insights that could further refine risk assessment for patients. This study was published in the January issue of The Journal of Nuclear Medicine.

Credit: Justin Ferdinandus, Wolfgang P. Fendler, Andrea Farolfi, et al.

Reston, VA (January 20, 2022)—New research has confirmed the accuracy of the novel European Association of Urology (EAU) risk classification system that groups prostate cancer patients based on their risk of recurrence. Prostate-specific membrane antigen (PSMA) PET imaging of men with prostate cancer validated the EAU groupings and provided insights that could further refine risk assessment for patients. This study was published in the January issue of The Journal of Nuclear Medicine.

The diagnostic workup of prostate cancer has changed rapidly over the past few years. Recently, the EAU introduced a clinical system separating patients with rising PSA values after first-line therapy (prostate surgery or radiation) into groups of those with high risk and those with low risk for development of metastases. Shortly after this, the U.S. Food and Drug Administration approved 68Ga-PSMA-11 as the first PET drug to target the PSMA for men with prostate cancer.

“Given the growing availability of PSMA-directed PET imaging, our study sought to assess disease in patients based on the EAU classifications while using PSMA PET to identify subgroups of patients, such as those with undetectable, locoregional or distant metastatic disease,” said Justin Ferdinandus, MD, nuclear medicine physician at University Hospital in Essen, Germany.

The multicenter, international study analyzed PSMA PET scans of nearly 2,000 patients with prostate cancer and rising PSA levels. Patterns of disease spread on PSMA PET imaging were used to classify prostate cancer patients into both low- and high-risk groups. High-risk groups were found to have higher rates of metastatic disease on PSMA PET compared to low-risk groups. However, PSMA PET also found metastatic disease in low-risk and no disease in high-risk patients.

“Our study underscores the utility of the EAU risk groups to determine risk of metastasis in biochemically recurrent prostate cancer. But not every high-risk patient has metastases and not every low-risk patient has locoregional or no disease,” said Wolfgang Fendler, MD, nuclear medicine physician at University Hospital in Essen.

He continued, “The ultimate aim of imaging is to provide the right treatment for each patient. As evidenced in this research, the accuracy of PSMA PET is essential to improve stratification and potentially outcomes both in low-risk and high-risk settings.” 

The authors of “PSMA PET validates higher rates of metastatic disease for European Association of Urology Biochemical Recurrence Risk Groups: an international multicenter study” include Justin Ferdinandus, Department of Nuclear Medicine, University of Duisburg-Essen and German Cancer Consortium (DKTK)-University Hospital Essen, Essen, Germany; Wolfgang P. Fendler and Ken Hermann, Department of Nuclear Medicine, University of Duisburg-Essen and German Cancer Consortium (DKTK)-University Hospital Essen, Essen, Germany, and Ahmanson Translational Imaging Division, Department of Molecular and Medical Pharmacology, University of California Los Angeles, Los Angeles, California; Andrea Farolfi, Department of Nuclear Medicine, University of Duisburg-Essen and German Cancer Consortium (DKTK)-University Hospital Essen, Essen, Germany, and Division of Nuclear Medicine, IRCCS Azienda Ospedaliero-Universitaria di Bologna, Bologna, Italy; Samuel Washington, Department of Urology, University of California San Francisco, San Francisco, California, and Department of Epidemiology and Statistics, University of California San Francisco, San Francisco, California; Osama Mohamad, Department of Radiation Oncology, University of California San Francisco, San Francisco, California; Miguel H. Pampaloni and Thomas A. Hope, Department of Radiology and Biomedical Imaging, University of California San Francisco, San Francisco, California; Peter J.H. Scott, Melissa Rodnick, Benjamin L. Viglianti and Morand Piert, Department of Radiology, University of Michigan, Ann Arbor, Michigan; Matthias Eiber, Department of Nuclear Medicine, Klinikum rechts der Isar, Technical University Munich, Munich, Germany; and Johannes Czernin, Wesley R. Armstrong and Jeremie Calais, Ahmanson Translational Imaging Division, Department of Molecular and Medical Pharmacology, University of California Los Angeles, Los Angeles, California.

Visit JNM’s new website for the latest research, and follow our new Twitter and Facebook pages @JournalofNucMed.

###

Please visit the SNMMI Media Center for more information about molecular imaging and precision imaging. To schedule an interview with the researchers, please contact Rebecca Maxey at (703) 652-6772 or [email protected].
 

About JNM and the Society of Nuclear Medicine and Molecular Imaging
The Journal of Nuclear Medicine (JNM) is the world’s leading nuclear medicine, molecular imaging and theranostics journal, accessed more than 13 million times each year by practitioners around the globe, providing them with the information they need to advance this rapidly expanding field. Current and past issues of The Journal of Nuclear Medicine can be found online at http://jnm.snmjournals.org.

JNM is published by the Society of Nuclear Medicine and Molecular Imaging (SNMMI), an international scientific and medical organization dedicated to advancing nuclear medicine and molecular imaging—precision medicine that allows diagnosis and treatment to be tailored to individual patients in order to achieve the best possible outcomes. For more information, visit www.snmmi.org.

Feedzy

Article: bioengineer.org

Continue Reading

Biotech

UCLA Researchers Develop Novel Microscopic Picoshell Particles

Published

on

Production of high-energy fats by microalgae may provide a sustainable, renewable energy source that can help tackle climate change. However, microalgae engineered to produce lipids rapidly usually grow slowly themselves, making it difficult to increase overall yields. 

Production of high-energy fats by microalgae may provide a sustainable, renewable energy source that can help tackle climate change. However, microalgae engineered to produce lipids rapidly usually grow slowly themselves, making it difficult to increase overall yields. 

UCLA bioengineers have created a new type of petri dish in the form of microscopic, permeable particles that can dramatically speed up research and development (R&D) timelines of biological products, such as fatty acids for biofuels. Dubbed PicoShells, the picoliter (trillionth of a liter), porous, hydrogel particles can enable more than one million individual cells to be compartmentalized, cultured in production-relevant environments, and selected based on growth and biomass accumulation traits using standard cell-processing equipment. 

Proceedings of the National Academy of Sciences recently published a study detailing how PicoShells work and their potential applications.

PicoShells consist of a hollow inner cavity where cells are encapsulated and a porous outer shell that allows for continuous solution exchange with the external environment so that nutrients, cell-communication molecules and cytotoxic cellular byproducts can transport freely in and out of the inner cavity. The shell also keeps the small groups of growing cells penned in, allowing researchers to study and compare their behaviors — what they do, how fast they grow, what they produce — to those of other groups inside various PicoShells. 

This new class of lab tool allows researchers to grow living, single-cell microorganisms — including algae, fungi and bacteria — under the same industrial-production conditions, such as in a bioreactor filled with wastewater or an outdoor cultivation pond. 

“PicoShells are like very tiny mesh balloons. The growing cells inside them are effectively fenced in but not sealed off,” said study leader Dino Di Carlo, UCLA’s Armond and Elena Hairapetian Professor in Engineering and Medicine at the UCLA Samueli School of Engineering. “With this new tool, we can now study the individual behaviors of millions of living cells in the relevant environment. This could shorten R&D-to-commercial production timelines for bioproducts from a few years to a few months. PicoShells could also be a valuable tool for fundamental biology studies.” 

PicoShells’ permeability can bring the lab to the industrial environment, allowing testing at a sectioned-off area of a working facility. Growth can occur more quickly and cell strains that perform well can be identified and selected for further screening. 

According to the researchers, another advantage of this new tool is that the analysis of millions of PicoShells is automated since they are also compatible with standard lab equipment used for high-volume cell processing.

Massive groups of cells, up to 10 million in one day, can be sorted and organized by certain characteristics. Continuous analysis could result in ideal sets of cells — ones that are already performing well in the environment with suitable temperature, nutrient composition and other properties that could be used in mass production — in just a few days rather than the several months it would take using current technologies.

The shells can be engineered to burst when the cells inside have divided and grown beyond their peak volume. Those free cells are still viable and can be recaptured for continued research or further selection. The researchers can also create shells with chemical groups that break down when exposed to biocompatible reagent, enabling a multifaceted approach to release selected cells.

“If we want to zero in on algae that are the best at producing biofuels, we can use PicoShells to organize, grow and process millions of single algal cells,” said lead author Mark van Zee, a bioengineering graduate student at UCLA Samueli. “And we can do that in machines that sort them using fluorescent tags that light up to signify fuel levels.”

Currently, cultivating and comparing such microorganisms are done mostly using traditional lab tools, such as microwell plates — cartons that hold several dozen small test tube-like volumes. However, these methods are slow and it is difficult to quantify their effectiveness because it can take weeks or months to grow large colonies for study. Other approaches, such as water-in-oil droplet emulsions, can be used to analyze cells in smaller volumes, but surrounding oils prevent the free exchange of medium into the water drops. Even cells or microorganisms that perform well in lab conditions may not do as well once they are placed in industrial environments, such as bioreactors or outdoor cultivation farms. As a result, cell strains that are developed in the lab often do not exhibit the same beneficial characteristic behavior when transferred to industrial production. 

Microwell plates also are limited in the number of experiments that can be performed, resulting in a great deal of trial and error in finding cell strains that work sufficiently well for mass production.

The researchers demonstrated the new tool by growing colonies of algae and yeast, comparing their growth and viability against other colonies grown in water-in-oil emulsions. For the algae, the team found that PicoShell colonies accumulated biomass rapidly while algae did not grow at all in water-in-oil emulsions. Similar results were found in their yeast experiments. By selecting the top growing algae in PicoShells, the researchers could increase the production of chlorophyll biomass by 8% after only a single cycle. 

The authors said PicoShells could offer a faster alternative to develop new algae and yeast strains, leading to improved biofuels, plastics, carbon-capture materials and even food products and alcoholic beverages. Further refinements to the technology, such as coating the shells with antibodies, could also lead to developing new types of protein-based medicines.

Di Carlo, van Zee and study co-author Joseph de Rutte Ph.D. ’20, a former member of Di Carlo’s research group, are named inventors on a patent application filed by the UCLA Technology Development Group.

The other UCLA authors on the paper are Rose Rumyan, Cayden Williamson, Trevor Burnes, Andrew Sonico Eugenio, Sara Badih, Dong-Hyun Lee and Maani Archang. Randor Radakovits from Synthetic Genomics of San Diego is also an author.

The study was supported by the Presidential Early Career Award for Scientists and Engineers and a planning award from the California NanoSystems Institute (CNSI) at UCLA.

Di Carlo holds faculty appointments in bioengineering, and mechanical and aerospace engineering at UCLA Samueli. He is a member of CNSI and the Jonsson Comprehensive Cancer Center at UCLA.

Read More

Source Here: bioengineer.org

Continue Reading

Trending

DXDD.com