For centuries humans have been fascinated by how easily birds glide through the air. Their instinctive ability to use wind to their advantage is what inspired Morpho, the new inspection drone developed by Elythor. Part winged drone, part quadcopter, Morpho is a hybrid unmanned aerial vehicle (UAV) that can change shape according to the task at hand. Its adaptive wings extend the drone&rsquo;s flight time and give it greater maneuverability. Combined with embedded sensors and cameras, these features let it fly just as well in enclosed spaces as in the open air, making it ideal for inspecting power plants and other infrastructure like high-voltage power lines, wind turbines, gas pipelines and offshore oil platforms.<iframe width="640" height="360" src="https://www.youtube.com/embed/tOUkn7YmYV4?&wmode=opaque&rel=0&enablejsapi=1&origin=https://www.wevolver.com" frameborder="0" allowfullscreen="" class="fr-draggable" data-lf-form-tracking-inspected-kn9eq4roawkarlvp="true" data-lf-yt-playback-inspected-kn9eq4roawkarlvp="true" data-lf-vimeo-playback-inspected-kn9eq4roawkarlvp="true"></iframe> <h2>Cutting costs by a third and reducing inspection time</h2>When sitting on its four legs, Morpho has a streamlined shape that resembles a tiny rocket. But that quickly changes once its rotors start whirring. The drone&rsquo;s vertical take-off capability means it can be deployed in confined spaces, fly to within just a few centimeters of a piece of equipment, and inspect the equipment without bumping into it. Once the inspection is completed, Morpho can shift its motors and rotate into a horizontal, more aerodynamic position and swiftly fly off to the next inspection site. If the drone encounters strong or rapidly changing winds along the way, it will expand or contract its wings as necessary to maintain its trajectory. It can also glide under the right wind conditions. All that saves time during inspections, but also means the drone can fly for longer. &ldquo;We calculated that using Morpho can cut the time and cost of a drone inspection by an average of 35%,&rdquo; says Harry Vourtsis, Elythor co-founder and CEO. Once Morpho reaches the next inspection site, it can fold up its wings and return to a vertical position in order to fly right up to the equipment.<blockquote>Our design can reduce power use by up to 85% when the drone is flying in the vertical position&nbsp; - Nathan M&uuml;ller, co-founder of Elythor&nbsp;</blockquote><img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1683465765223-003ee723.jpg" style="width: 766px;" class="fr-fic fr-dib">Harry Vourtsis, CEO and co-founder, and Nathan M&uuml;ller, co-founder of the start-up. EPFL/Alain Herzog&nbsp;<h2 id="isPasted">Asymmetric wing adjustment provides stability in strong winds</h2>&ldquo;Winged drones have the advantage of longer flight times, while quadcopters have better manoeuvrability,&rdquo; says Vourtsis. &ldquo;We combined the two and added an adaptative wing system that reduces the drone&rsquo;s power requirement even further.&rdquo; Most fixed-wing vertical take-off and landing (VTOL) drones represent a compromise, with wings that are small enough to reduce friction during vertical flight yet large enough to provide sufficient lift during horizontal flight. VTOL drones also have trouble taking off and landing in strong winds due to their large surface area perpendicular to the wind.Morpho&rsquo;s wing control system is the result of several years of research at EPFL&rsquo;s Laboratory of Intelligent Systems and has been described in a number of scientific articles. The system includes sensors linked to a software program for monitoring wind direction and speed. Nathan M&uuml;ller, also an Elythor co-founder, explains: &ldquo;The controller automatically selects whether to hold the wings in place or let them move freely with the wind, based on the drone&rsquo;s trajectory and effective speed, along with any changes in wind direction. The wings&rsquo; surface area can also be adjusted either symmetrically or asymmetrically depending on wind direction.&rdquo;The algorithms powering the control system seek not just to optimize the trade-off between air friction and lift, but also to minimize power use. This entails taking advantage of wind currents to let the drone glide or adjusting the wings asymmetrically to regulate its yaw &ndash; or the rotation around its vertical axis. This provides greater stability in heavy winds.Quantitative studies carried out at the EPFL lab show that Morpho delivers considerably better maneuverability and power efficiency. &ldquo;Our design can reduce power use by up to 85% when the drone is fyling in the vertical position,&rdquo; says M&uuml;ller. &ldquo;It also significantly improves the drone&rsquo;s attitude and stability.&rdquo; These benefits mean Morpho can perform inspections under a broader range of weather conditions.<h2>&ldquo;We want to revolutionize inspection platforms&rdquo;</h2>Because Elythor aims to provide customers with a turnkey solution, it&rsquo;s also developing software for compiling and analyzing the data collected by Morpho. &ldquo;We want to revolutionize inspection platforms,&rdquo; says Vourtsis. Today, power plant operators use different kinds of drones depending on where the equipment to inspect is located. And the facilities sometimes stretch over several kilometers and include complicated structures that must be entered, like wind turbines and transmission towersElythor has raised a substantial amount of startup funding and plans to get its drone ready for industrial production in the coming months, with a market launch slated for the end of this year.<hr>References<a href="https://onlinelibrary.wiley.com/doi/full/10.1002/aisy.202200297" rel="noopener" target="_blank">Wind Defiant Morphing Drones in Advanced Intelligent Systems</a> (January 2023)<a href="https://infoscience.epfl.ch/record/301328" target="_blank">Resilient drones with morphing wings&nbsp;</a>-Harry Vourtsis Thesis (March 2023)<a href="https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=9591382" rel="noopener" target="_blank">Robotic Elytra: Insect-Inspired Protective Wings for Resilient and Multi-Modal Drones</a> - IEEE Robotics and Automation Letters (January 2022)<a href="https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=9479684" rel="noopener" target="_blank">Insect Inspired Self-Righting for Fixed-Wing Drones</a> - IEEE Robotics and Automation Letters (October 2021)

Elythor, an EPFL spin-off, has developed a new drone whose wing shape can adapt to wind conditions and flight position in real time, reducing the drone’s energy consumption. What’s more, the position of the wings can change, allowing the drone to fly vertically or horizontally. These features make it a perfect candidate for inspecting power plants.

New inspection drone uses wind to lengthen flight times

In this article, we will dive deep into the world of simultaneous localization and mapping using Lidar technology. Lidar SLAM has been gaining popularity in recent years, thanks to its versatility and applications across various domains, including autonomous vehicles, mobile robotics, and indoor mapping.

Lidar SLAM: The Ultimate Guide to Simultaneous Localization and Mapping

<h2 dir="ltr" id="isPasted">Fast track into the cloud with CGConnect</h2>&quot;Enterprises who rely on drones and robotics for business operation often own a diverse range of uncrewed vehicles that may not be compatible with one another. <a href="https://youtu.be/lumuuYy2fHU" target="_blank">CGConnect</a> is designed to solve this pain point by linking them to the Cloud Ground Control platform, regardless of manufacturer or model, turning them into a holistic, connected fleet,&quot; said Michal Weiss, Head of Product at Cloud Ground Control.<iframe width="640" height="360" src="https://www.youtube.com/embed/fogtQQ3N4rQ?&wmode=opaque&rel=0&enablejsapi=1&origin=https://www.wevolver.com" frameborder="0" allowfullscreen="" class="fr-draggable" data-lf-form-tracking-inspected-kn9eq4roawkarlvp="true" data-lf-yt-playback-inspected-kn9eq4roawkarlvp="true" data-lf-vimeo-playback-inspected-kn9eq4roawkarlvp="true"></iframe> <h2 dir="ltr">Plug-and-play connectivity</h2>Using CGConnect, remote users gain instant access to Cloud Ground Control&#39;s features, including real-time telemetry, cloud storage, video and payload data, all from a web browser simultaneously. Weighing only 55 grams with similar sizing to a credit card, CGConnect can be integrated into any product design, offering the following benefits:<ul id="isPasted"><li dir="ltr">Open platform: The customisable open platform operates on the MAVLink standard. This multiplies potential product applications and enables diverse autonomous vehicles and payloads to operate as a coordinated fleet.</li><li dir="ltr">Robotic agnostic: Works flexibly with open-sourced libraries and is agnostic to the type of technology and vehicle enterprises may wish to use.&nbsp;</li></ul><ul><li dir="ltr">White labelled:&nbsp;Available as a white label product, allowing users to rebrand the user interface to complement business branding and coding requirements.</li><li dir="ltr">High-grade security:&nbsp;Utilises military-grade encryption and authentication to safeguard data and IP from vulnerabilities and security breaches, helping users meet compliance obligations.</li><li dir="ltr">Simple and accessible: Revolutionises multi-drone operation by making it simple, cost-effective and accessible to users of every skill grade.</li><li dir="ltr">AI modelling:&nbsp;The platform runs AI algorithms in the cloud, relaying real-time camera feed data to the end user to support versatile missions, such as object detection, tracking and thermal imaging.</li><li dir="ltr">Edge AI: CGConnect supports edge AI to perform intensive object identification and classification directly on the vehicle for dynamic missions.</li></ul><iframe width="640" height="360" src="https://www.youtube.com/embed/lumuuYy2fHU?&wmode=opaque&rel=0&enablejsapi=1&origin=https://www.wevolver.com" frameborder="0" allowfullscreen="" class="fr-draggable" data-lf-form-tracking-inspected-kn9eq4roawkarlvp="true" data-lf-yt-playback-inspected-kn9eq4roawkarlvp="true" data-lf-vimeo-playback-inspected-kn9eq4roawkarlvp="true"></iframe> <h2 dir="ltr" id="isPasted">Unlock new markets</h2>Integrating CGConnect into a vehicle&#39;s design expands the product&#39;s functionality and applications while saving development time. This allows manufacturers to fine-tune the product&#39;s competitive advantage, expanding the robot&#39;s potential and ultimately unlocking access to new markets.<iframe width="640" height="360" src="https://www.youtube.com/embed/-jVsC77f5bo?&wmode=opaque&rel=0&enablejsapi=1&origin=https://www.wevolver.com" frameborder="0" allowfullscreen="" class="fr-draggable" data-lf-form-tracking-inspected-kn9eq4roawkarlvp="true" data-lf-yt-playback-inspected-kn9eq4roawkarlvp="true" data-lf-vimeo-playback-inspected-kn9eq4roawkarlvp="true"></iframe> Cloud Ground Control helped Clean Earth Rovers transform its Rover AVPros into an autonomous, connected fleet. By integrating CGConnect, Clean Earth Rovers were able to provide customers with real-time situational awareness and greatly accelerate recovery efforts in ocean health.&nbsp;<a href="https://www.youtube.com/watch?v=-jVsC77f5bo&t=22s" target="_blank">Discover the integration here</a>.For more information, visit&nbsp;<a href="http://www.cloudgroundcontrol.com/manufacturers" target="_blank">www.cloudgroundcontrol.com/manufacturers</a>.&nbsp;

CGConnect, its cellular micro-modem that uses 4G/5G networks to link any drone or robotic vehicle to Cloud Ground Control’s cloud-based drone fleet management platform, enabling live-streaming, command and control from a web browser.

Transform multiple unmanned vehicles into connected fleet with new  micro-cellular modem

Their laser-based sensing approach, detailed in a paper <a href="https://www.sciencedirect.com/science/article/pii/S0034425723000640?dgcid=author" target="_blank">published</a> Mar. 1 in Remote Sensing of Environment, can accurately detect and quantify both large greenhouse gas leaks and leaks up to 25 times smaller than those typically detected at natural gas facilities using other methods, localizing the emissions source to within a meter. Because it takes advantage of the remote-sensing capabilities of lasers combined with the agility of drones, the new technology can also be used to quickly spot otherwise unseen leaks in hard-to-access areas, an innovation the researchers said unlocks game-changing potential for atmospheric sensing.&ldquo;Current approaches for detecting leaks often rely on handheld infrared cameras that are labor-intensive to operate and insensitive to small leaks, or they use methods that require setting up extensive measurement infrastructure ahead of time,&rdquo; said <a href="https://ece.princeton.edu/people/gerard-wysocki" target="_blank">Gerard Wysocki</a>, associate professor of <a href="https://ece.princeton.edu/" target="_blank">electrical and computer engineering</a> and associated faculty at the <a href="https://acee.princeton.edu/" target="_blank">Andlinger Center for Energy and the Environment</a>. &ldquo;But with a drone, you are completely free in how you are able to set up your sensing area.&rdquo;<img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1682346135815-20230328_methane_drone_BD_03_CROPPED_v3-300x225+%281%29.jpg" style="width: 766px;" class="fr-fic fr-dib">The retroflector mounted onto the drone. Photo by Bumper DeJesus&nbsp;The researchers&rsquo; approach consists of a small drone outfitted with only a retroreflector, a type of mirror that reflects incoming light directly back to the source, and a base station of gas sensing equipment with the capability to track the drone&rsquo;s movement during flight. Bouncing a laser beam off the drone as it flies to set points around a suspected leak allows an operator to pinpoint the source of the leak and measure its intensity.&ldquo;That&rsquo;s really the holy grail of leak detection,&rdquo; said <a href="https://cee.princeton.edu/people/mark-zondlo" target="_blank">Mark Zondlo</a>, study co-author, professor of <a href="https://cee.princeton.edu/" target="_blank">civil and environmental engineering</a>, and associated faculty at the Andlinger Center for Energy and the Environment.While drone-based techniques for atmospheric sensing exist, they usually require mounting a gas sensor directly onto a drone, a practice that Zondlo said quickly comes up against some major roadblocks related to how much weight a drone can carry and how risky it is to fly a drone overloaded with costly sensors in hazardous environments.&ldquo;You really can&rsquo;t fit more than one gas sensor on a drone at a time, otherwise it just becomes too big and bulky to fly. And you probably don&rsquo;t want to risk flying an expensive sensor over the lagoon of a wastewater treatment plant or the flaring of a compressor station,&rdquo; Zondlo said.Instead of overburdening the drone with sensors, the Princeton group offloaded the expensive gas sensing components to the base station, which can fit on a mobile platform such as a van, meaning that the drone only needed to carry a small mirror. That change allowed the researchers to use smaller, less expensive drones with longer flight times to collect highly detailed emissions data across large areas, potentially unlocking the ability to monitor entire natural gas transmission and distribution facilities in a single drone flight. &ldquo;Our approach allows us to bypass the major constraints of using drones and enables us to make full use of their potential,&rdquo; Zondlo said. &ldquo;It&rsquo;s really a paradigm shift in how we can use drones for atmospheric sensing.&rdquo;The sensing method could also enable simultaneous measurements of multiple gases, a feat that is exceedingly difficult with other drone-based approaches due to size, weight, and power considerations. Michael Soskind, the first author of the study and a graduate student in electrical and computer engineering, said adding the ability to measure other gases like carbon dioxide and ammonia alongside methane would be as simple as adding other lasers of different wavelengths to the base system. &ldquo;All you&rsquo;d need to do is add a secondary laser to the system,&rdquo; he explained. &ldquo;The rest of the system is already built out to do the work.&rdquo;Because it offers users a great amount of flexibility, Wysocki said he sees the team&rsquo;s approach as a technology platform that could spur future innovation and applications beyond methane leak detection.&ldquo;The most exciting thing is not simply the methane sensing abilities of the technology we developed,&rdquo; he said. &ldquo;It&rsquo;s really about unlocking the capability for researchers and practitioners to use drones and other remote sensing techniques to take detailed measurements of small leaks and reconstruct emissions plumes. It&rsquo;s a technology that opens the doors for efficient leak detection and repair, which can help producers mitigate the safety and environmental hazards of those leaks, while also saving them time and money.&rdquo;The article, &ldquo;<a href="https://www.sciencedirect.com/science/article/pii/S0034425723000640?dgcid=author" target="_blank">Stationary and drone-assisted methane plume localization with dispersion spectroscopy</a>,&rdquo; was published Mar. 1 in Remote Sensing of Environment. In addition to Wysocki, Zondlo, and Soskind, authors include Nathan Li, Daniel Moore, Lars Wendt, and James McSpiritt of Princeton University, as well as Yifeng Chen, a former graduate student at Princeton University who now works at NEO Monitors. The work was supported by the U.S. Department of Energy&rsquo;s National Energy Technology Laboratory, the National Defense Science and Engineering Graduate Fellowship, and the High Meadows Environmental Institute through the Walbridge Fund Graduate Award for Environmental Research.<img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1682346155567-IMG_3129_toned.jpg" style="width: 766px;" class="fr-fic fr-dib">Lars Wendt, study co-author, operates the drone during field experiments. (Photo courtesy of the researchers)&nbsp;

Michael Soskind, first author and graduate student in electrical and computer engineering, stands with the drone. Photo by Bumper DeJesus

As evidence mounts that gas drilling and sewer systems leak far more greenhouse gases than previously believed, a team of Princeton researchers has developed a method to pinpoint leaks both big and small for speedy repair.

Using drones and lasers, researchers pinpoint greenhouse gas leaks

This article was first published on&nbsp;<a href="https://news.mit.edu/2023/drones-navigate-unseen-environments-liquid-neural-networks-0419" id="isPasted" rel="noopener noreferrer" target="_blank">MIT News</a>. &nbsp;<hr>In the vast, expansive skies where birds once ruled supreme, a new crop of aviators is taking flight. These pioneers of the air are not living creatures, but rather a product of deliberate innovation: drones. But these aren&rsquo;t your typical flying bots, humming around like mechanical bees. Rather, they&rsquo;re avian-inspired marvels that soar through the sky, guided by liquid neural networks to navigate ever-changing and unseen environments with precision and ease.Inspired by the adaptable nature of organic brains, researchers from MIT&rsquo;s Computer Science and Artificial Intelligence Laboratory (CSAIL) have introduced a method for robust flight navigation agents to master vision-based fly-to-target tasks in intricate, unfamiliar environments. The liquid neural networks, which can continuously adapt to new data inputs, showed prowess in making reliable decisions in unknown domains like forests, urban landscapes, and environments with added noise, rotation, and occlusion. These adaptable models, which outperformed many state-of-the-art counterparts in navigation tasks, could enable potential real-world drone applications like search and rescue, delivery, and wildlife monitoring.The researchers&#39; recent study, <a href="https://www.science.org/doi/10.1126/scirobotics.adc8892" target="_blank">published today in Science Robotics</a>, details how this new breed of agents can adapt to significant distribution shifts, a long-standing challenge in the field. The team&rsquo;s new class of machine-learning algorithms, however, captures the causal structure of tasks from high-dimensional, unstructured data, such as pixel inputs from a drone-mounted camera. These networks can then extract crucial aspects of a task (i.e., understand the task at hand) and ignore irrelevant features, allowing acquired navigation skills to transfer targets seamlessly to new environments.<iframe width="640" height="360" src="https://www.youtube.com/embed/RPFbk6D_dNw?&wmode=opaque&rel=0&enablejsapi=1&origin=https://www.wevolver.com" frameborder="0" allowfullscreen="" class="fr-draggable" data-lf-form-tracking-inspected-kn9eq4roawkarlvp="true" data-lf-yt-playback-inspected-kn9eq4roawkarlvp="true" data-lf-vimeo-playback-inspected-kn9eq4roawkarlvp="true"></iframe>Drones navigate unseen environments with liquid neural networks.&nbsp;&ldquo;We are thrilled by the immense potential of our learning-based control approach for robots, as it lays the groundwork for solving problems that arise when training in one environment and deploying in a completely distinct environment without additional training,&rdquo; says Daniela Rus, CSAIL director and the Andrew (1956) and Erna Viterbi Professor of Electrical Engineering and Computer Science at MIT. &ldquo;Our experiments demonstrate that we can effectively teach a drone to locate an object in a forest during summer, and then deploy the model in winter, with vastly different surroundings, or even in urban settings, with varied tasks such as seeking and following. This adaptability is made possible by the causal underpinnings of our solutions. These flexible algorithms could one day aid in decision-making based on data streams that change over time, such as medical diagnosis and autonomous driving applications.&rdquo;A daunting challenge was at the forefront: Do machine-learning systems understand the task they are given from data when flying drones to an unlabeled object? And, would they be able to transfer their learned skill and task to new environments with drastic changes in scenery, such as flying from a forest to an urban landscape? What&rsquo;s more, unlike the remarkable abilities of our biological brains, deep learning systems struggle with capturing causality, frequently over-fitting their training data and failing to adapt to new environments or changing conditions. This is especially troubling for resource-limited embedded systems, like aerial drones, that need to traverse varied environments and respond to obstacles instantaneously.&nbsp;The liquid networks, in contrast, offer promising preliminary indications of their capacity to address this crucial weakness in deep learning systems. The team&rsquo;s system was first trained on data collected by a human pilot, to see how they transferred learned navigation skills to new environments under drastic changes in scenery and conditions. Unlike traditional neural networks that only learn during the training phase, the liquid neural net&rsquo;s parameters can change over time, making them not only interpretable, but more resilient to unexpected or noisy data.&nbsp;In a series of quadrotor closed-loop control experiments, the drones underwent range tests, stress tests, target rotation and occlusion, hiking with adversaries, triangular loops between objects, and dynamic target tracking. They tracked moving targets, and executed multi-step loops between objects in never-before-seen environments, surpassing performance of other cutting-edge counterparts.&nbsp;The team believes that the ability to learn from limited expert data and understand a given task while generalizing to new environments could make autonomous drone deployment more efficient, cost-effective, and reliable. Liquid neural networks, they noted, could enable autonomous air mobility drones to be used for environmental monitoring, package delivery, autonomous vehicles, and robotic assistants.&nbsp;&ldquo;The experimental setup presented in our work tests the reasoning capabilities of various deep learning systems in controlled and straightforward scenarios,&rdquo; says MIT CSAIL Research Affiliate Ramin Hasani. &ldquo;There is still so much room left for future research and development on more complex reasoning challenges for AI systems in autonomous navigation applications, which has to be tested before we can safely deploy them in our society.&rdquo;&ldquo;Robust learning and performance in out-of-distribution tasks and scenarios are some of the key problems that machine learning and autonomous robotic systems have to conquer to make further inroads in society-critical applications,&rdquo; says Alessio Lomuscio, professor of AI safety in the Department of Computing at Imperial College London. &ldquo;In this context, the performance of liquid neural networks, a novel brain-inspired paradigm developed by the authors at MIT, reported in this study is remarkable. If these results are confirmed in other experiments, the paradigm here developed will contribute to making AI and robotic systems more reliable, robust, and efficient.&quot;Clearly, the sky is no longer the limit, but rather a vast playground for the boundless possibilities of these airborne marvels.&nbsp;Hasani and PhD student Makram Chahine; Patrick Kao &#39;22, MEng &#39;22; and PhD student Aaron Ray SM &#39;21 wrote the paper with Ryan Shubert &#39;20, MEng &#39;22; MIT postdocs Mathias Lechner and Alexander Amini; and Rus.This research was supported, in part, by Schmidt Futures, the U.S. Air Force Research Laboratory, the U.S. Air Force Artificial Intelligence Accelerator, and the Boeing Co.<hr>&quot;Reprinted with permission of&nbsp;<a href="https://news.mit.edu/2023/drones-navigate-unseen-environments-liquid-neural-networks-0419" id="isPasted" rel="noopener noreferrer" target="_blank">MIT News</a>&quot; &nbsp;

Makram Chahine, a PhD student in electrical engineering and computer science and an MIT CSAIL affiliate, leads a drone used to test liquid neural networks. Photo: Mike Grimmett/MIT CSAIL

MIT researchers exhibit a new advancement in autonomous drone navigation, using brain-inspired liquid neural networks that excel in out-of-distribution scenarios.

Drones navigate unseen environments with liquid neural networks

This flying robot attaches to various assets to perform contact based Non-Destructive Testing

Intelligent flying robots can perform Non-Destructive Testing inspections on various assets across various industries, including tanks, towers, pipes, flare stacks, ships, and more.

FLYING ROBOT'S RANGE OF ASSETS

A&nbsp;<a href="https://www.ae.utexas.edu/facultysites/topcu/wiki/index.php/Verifiable,_Control-Oriented_Learning_On_The_Fly" target="_blank">research team</a> from Princeton, the University of Texas, and Northeastern University is working under a $7.5 million grant from the U.S. Air Force Office of Scientific Research to pave the way for creating such a system. The basic research the team is doing could someday extend to aircraft controls and many other applications, including controlling disease epidemics or making more accurate predictions about climate change or species survival, said&nbsp;<a href="https://engineering.princeton.edu/faculty/amir-ali-ahmadi" target="_blank">Amir Ali Ahmadi</a>, a professor of&nbsp;<a href="https://orfe.princeton.edu/" target="_blank">operations research and financial engineering</a> at Princeton and a member of the research team.The goal is to exert measures of control over a &ldquo;dynamical system&rdquo; &mdash; one that changes as it moves. Most dynamical systems are notoriously difficult to predict and manage. Ahmadi, along with colleagues&nbsp;<a href="https://www.math.princeton.edu/people/charles-fefferman" target="_blank">Charles Fefferman</a>, the Herbert E. Jones, Jr. &rsquo;43 University Professor of Mathematics, and&nbsp;<a href="https://engineering.princeton.edu/faculty/clarence-rowley" target="_blank">Clarence Rowley</a>, the Sin-I Cheng Professor in Engineering Science, are trying to design algorithms that can learn the behavior of dynamical systems from data.&ldquo;A dynamical system is any entity in some space that evolves through time,&rdquo; Ahmadi said. &ldquo;So, an airplane is a dynamical system; a robot is a dynamical system; the spread of a virus is a dynamical system.&rdquo;Gaining control is particularly tough when data is limited, said Ahmadi. In the case of a damaged aircraft, &ldquo;the plane has changed, and you have less than a minute to come up with a new model of control,&rdquo; he said.Predicting future performance based on extremely sparse data is a common problem. It is hard to recommend the best response to a disease outbreak, for example, when very little is known about the spread of illness.In a recent&nbsp;<a href="https://epubs.siam.org/doi/abs/10.1137/20M1388644" target="_blank">article</a> in the SIAM Review, Ahmadi&rsquo;s research team presented an approach that uses additional information to rapidly respond to changing conditions in which little data is available for decision-making. This additional information, which mathematicians call side information, acts in the same way that experience or professional expertise does for a human. For example, a doctor might never have seen a particular disease before, but years of experience will help her make a good judgment on how to treat the patient.&ldquo;That is what this entire project is about,&rdquo; Ahmadi said. &ldquo;It is about learning a system from very little data and eventually controlling it in a way that we desire.&rdquo;<h2>Starting simple</h2>Long-term goals, such as aircraft controls, are beyond the scope of the immediate project. Rather, the work under the Air Force grant is focusing on much simpler examples in the hopes of learning more about controlling a system riddled with unknowns.&ldquo;In standard control theory, you understand what the controls do. We&rsquo;re trying to make a more powerful version of that theory in which you don&rsquo;t know what the controls do, but you learn by applying them,&rdquo; Fefferman said. He is working with Rowley on relatively simple sub-problems of dynamical systems &mdash; for instance, trying to temporarily halt an object as it moves along a straight line at a constant speed.&nbsp;In addition, the researchers want to use as little energy as possible to exert control &mdash; just as a pilot would want to do in a plane with limited fuel.Another problem they may tackle is an advanced version of a problem commonly assigned to undergraduate mechanical engineering majors: controlling an inverted pendulum &mdash; similar to trying to balance a broomstick in the palm of your hand. The controller would learn the behaviors of the system almost instantly and without knowing where its mass is centered. To do that, they would create equations for controls based on a few seconds of observation, then modify the controls after recording what they do. The model would be designed to rapidly go through several learn-and-control iterations.<h2>Knowledge versus control</h2>The problems the team explores involve tradeoffs between exploring functionality and exploiting the knowledge gained, Rowley said. &ldquo;If you exploit your knowledge too soon, the model may not be good enough to land a plane. But if you spend too much time learning its behavior, the plane may crash.&rdquo;There is no single technique for controlling a system with unknown dynamics, said&nbsp;<a href="https://www.ae.utexas.edu/people/faculty/faculty-directory/topcu" target="_blank">Ufuk Topcu</a>, a team member and associate professor at the University of Texas. But one of the keys is to select the most valuable data to work on. &ldquo;You have to tackle it from multiple angles and chop the big problem into more manageable pieces to identify what&rsquo;s worth learning,&rdquo; he said.When the grant ends, the researchers expect to have algorithms for controlling at least some aspects of a dynamical system. Though their model may not be fast enough to operate in real time, it should be able to show which controls are possible in a changing system and with what degree of certainty they can succeed, Ahmadi said.The&nbsp;<a href="https://epubs.siam.org/doi/abs/10.1137/20M1388644" target="_blank">article</a>&nbsp;&ldquo;Learning Dynamical Systems with Side Information&rdquo; was published in the February issue of the Society for Industrial and Applied Mathematics Review. Besides Ahmadi, authors include Bachir El Khadir, a former graduate student at Princeton who is now at the investment firm Two Sigma. Support for the project was provided in part by the Air Force Office of Scientific Research, the National Science Foundation, DARPA, the Sloan Research Fellowship, the Google Faculty Award, and an Innovation Award of the Princeton School of Engineering and Applied Science.

The researchers are working to improve controls for flight systems and other applications that demand quick responses. Photo by Bumper DeJesus

Commercial airplanes can be controlled by autopilot. But what happens if a wing gets damaged or an engine malfunctions? Is it possible to design a software system with a feedback loop — a system that quickly tests how controls operate on the damaged vessel and makes adjustments on the fly to give it the best chance of landing safely?

Teaching robots to think on the fly

This article was first published on&nbsp;<a href="https://news.mit.edu/2023/new-algorithm-keeps-drones-colliding-midair-0329" id="isPasted" rel="noopener noreferrer" target="_blank">MIT News</a>. &nbsp;<hr>When multiple drones are working together in the same airspace, perhaps spraying pesticide over a field of corn, there&rsquo;s a risk they might crash into each other.To help avoid these costly crashes, MIT researchers presented a system called MADER in 2020. This multiagent trajectory-planner enables a group of drones to formulate optimal, collision-free trajectories. Each agent broadcasts its trajectory so fellow drones know where it is planning to go. Agents then consider each other&rsquo;s trajectories when optimizing their own to ensure they don&rsquo;t collide.But when the team tested the system on real drones, they found that if a drone doesn&rsquo;t have up-to-date information on the trajectories of its partners, it might inadvertently select a path that results in a collision. The researchers revamped their system and are now rolling out Robust MADER, a multiagent trajectory planner that generates collision-free trajectories even when communications between agents are delayed.<iframe width="640" height="360" src="https://www.youtube.com/embed/i1d8di2Nrbs?&wmode=opaque&rel=0&enablejsapi=1&origin=https://www.wevolver.com" frameborder="0" allowfullscreen="" class="fr-draggable" data-lf-form-tracking-inspected-kn9eq4roawkarlvp="true" data-lf-yt-playback-inspected-kn9eq4roawkarlvp="true" data-lf-vimeo-playback-inspected-kn9eq4roawkarlvp="true"></iframe>&ldquo;MADER worked great in simulations, but it hadn&rsquo;t been tested in hardware. So, we built a bunch of drones and started flying them. The drones need to talk to each other to share trajectories, but once you start flying, you realize pretty quickly that there are always communication delays that introduce some failures,&rdquo; says Kota Kondo, an aeronautics and astronautics graduate student.The algorithm incorporates a delay-check step during which a drone waits a specific amount of time before it commits to a new, optimized trajectory. If it receives additional trajectory information from fellow drones during the delay period, it might abandon its new trajectory and start the optimization process over again.When Kondo and his collaborators tested Robust MADER, both in simulations and flight experiments with real drones, it achieved a 100 percent success rate at generating collision-free trajectories. While the drones&rsquo; travel time was a bit slower than it would be with some other approaches, no other baselines could guarantee safety.&ldquo;If you want to fly safer, you have to be careful, so it is reasonable that if you don&rsquo;t want to collide with an obstacle, it will take you more time to get to your destination. If you collide with something, no matter how fast you go, it doesn&rsquo;t really matter because you won&rsquo;t reach your destination,&rdquo; Kondo says.&nbsp;&nbsp;Kondo wrote the&nbsp;<a href="https://arxiv.org/abs/2303.06222" target="_blank">paper</a> with Jesus Tordesillas, a postdoc; Parker C. Lusk, a graduate student; Reinaldo Figueroa, Juan Rached, and Joseph Merkel, MIT undergraduates; and senior author Jonathan P. How, the Richard C. Maclaurin Professor of Aeronautics and Astronautics, a principal investigator in the Laboratory for Information and Decision Systems (LIDS), and a member of the MIT-IBM Watson AI Lab. The research will be presented at the International Conference on Robots and Automation.<h2>Planning trajectories</h2>MADER is an asynchronous, decentralized, multiagent trajectory-planner. This means that each drone formulates its own trajectory and that, while all agents must agree on each new trajectory, they don&rsquo;t need to agree at the same time. This makes MADER more scalable than other approaches, since it would be very difficult for thousands of drones to agree on a trajectory simultaneously. Due to its decentralized nature, the system would also work better in real-world environments where drones may fly far from a central computer.With MADER, each drone optimizes a new trajectory using an algorithm that incorporates the trajectories it has received from other agents. By continually optimizing and broadcasting their new trajectories, the drones avoid collisions.But perhaps one agent shared its new trajectory several seconds ago, but a fellow agent didn&rsquo;t receive it right away because the communication was delayed. In real-world environments, signals are often delayed by interference from other devices or environmental factors like stormy weather. Due to this unavoidable delay, a drone might inadvertently commit to a new trajectory that sets it on a collision course.Robust MADER prevents such collisions because each agent has two trajectories available. It keeps one trajectory that it knows is safe, which it has already checked for potential collisions. While following that original trajectory, the drone optimizes a new trajectory but does not commit to the new trajectory until it completes a delay-check step.During the delay-check period, the drone spends a fixed amount of time repeatedly checking for communications from other agents to see if its new trajectory is safe. If it detects a potential collision, it abandons the new trajectory and starts the optimization process over again.The length of the delay-check period depends on the distance between agents and environmental factors that could hamper communications, Kondo says. If the agents are many miles apart, for instance, then the delay-check period would need to be longer.<h2>Completely collision-free</h2>The researchers tested their new approach by running hundreds of simulations in which they artificially introduced communication delays. In each simulation, Robust MADER was 100 percent successful at generating collision-free trajectories, while all the baselines caused crashes.The researchers also built six drones and two aerial obstacles and tested Robust MADER in a multiagent flight environment. They found that, while using the original version of MADER in this environment would have resulted in seven collisions, Robust MADER did not cause a single crash in any of the hardware experiments.&ldquo;Until you actually fly the hardware, you don&rsquo;t know what might cause a problem. Because we know that there is a difference between simulations and hardware, we made the algorithm robust, so it worked in the actual drones, and seeing that in practice was very rewarding,&rdquo; Kondo says.Drones were able to fly 3.4 meters per second with Robust MADER, although they had a slightly longer average travel time than some baselines. But no other method was perfectly collision-free in every experiment.In the future, Kondo and his collaborators want to put Robust MADER to the test outdoors, where many obstacles and types of noise can affect communications. They also want to outfit drones with visual sensors so they can detect other agents or obstacles, predict their movements, and include that information in trajectory optimizations.This work was supported by Boeing Research and Technology.<hr>&quot;Reprinted with permission of&nbsp;<a href="https://news.mit.edu/2023/new-algorithm-keeps-drones-colliding-midair-0329" id="isPasted" rel="noopener noreferrer" target="_blank">MIT News</a>&quot; &nbsp;

When multiple drones are working together in the same airspace, there’s a risk they might collide. But now AeroAstro researchers have created a trajectory-planning system that enables drones in the same airspace to always choose a safe path forward. Courtesy of the researchers

Researchers create a trajectory-planning system that enables drones working together in the same airspace to always choose a safe path forward.

New algorithm keeps drones from colliding in midair

Case Study: SKYRON Inspects Three Assets in a Refinery&nbsp;Refineries are known for operating under harsh conditions of high pressure and temperature, which can cause damage to equipment over time, leading to defects such as internal corrosion, cracking, and erosion. Non-Destructive Testing (NDT) inspections are necessary for refineries to ensure safety and reliability. These inspections can detect any defects in pipes, tanks, flare stacks, and other equipment that may affect their structural integrity and functionality. Regular&nbsp;Ultrasonic&nbsp;Thickness (UT)&nbsp;inspections help refineries maintain safe&nbsp;and&nbsp;reliable&nbsp;equipment, which prevents incidents, reduces downtime, and increases productivity.&nbsp;Conventional inspection methods in refineries are known to be costly, time-consuming, and pose risks to inspectors. However, technological advancements have led to the development of intelligent flying robots that can perform inspections faster, cost-effectively, and safely. These&nbsp;robots can avoid downtime as inspections can be done while the equipment is in service, resulting in increased productivity and cost savings.&nbsp;<img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNjgwMTI0MTA2MTc1LTE2ODAxMjQxMDYxNzUuanBlZyIsImVkaXRzIjp7InJlc2l6ZSI6eyJ3aWR0aCI6OTUwLCJmaXQiOiJjb3ZlciJ9fX0=" class="fr-fic fr-dii">&nbsp;&nbsp;<iframe id="6433ae788de910000801b259" width="250" class="specs-iframe" height="420" src="https://wevolver.com/iframe/specs/skyron?iframe=true" frameborder="0" scrolling="true" data-lf-form-tracking-inspected-kn9eq4roawkarlvp="true" data-lf-yt-playback-inspected-kn9eq4roawkarlvp="true" data-lf-vimeo-playback-inspected-kn9eq4roawkarlvp="true"></iframe>Recently, the intelligent flying robot&nbsp;SKYRON&nbsp;took&nbsp;UT measurements on three assets:&nbsp;a tank, a complex network of pipes, and an in-service flare stack. The inspection presented several challenges due to the&nbsp;height&nbsp;of the tank shell and roof,&nbsp;the&nbsp;requirement&nbsp;to inspect&nbsp;pipes&nbsp;from&nbsp;various angles, and the&nbsp;high temperatures&nbsp;of&nbsp;the flare stack.&nbsp;The inspection process in the refinery began with an examination of a tank, which is challenging for inspectors due to the risks associated with working at elevated heights and encountering obstructions or limited access points. Fortunately, an intelligent flying robot has simplified this process and allowed inspectors to operate securely on the ground. With its advanced HD camera system, SKYRON enables operators to conduct inspections beyond the line of sight and navigate to limited access points, such as the tank roof, even when the robot is out of sight due to distance. Furthermore, SKYRON&#39;s wind handling capabilities and&nbsp;100-meter&nbsp;tether cable allow operators to take UT measurements from the tank shell and roof&nbsp;with ease and precision, reducing the risks of accidents associated with high winds and obstructions. SKYRON enabled inspectors to take 155 readings within just 45 minutes of flight time, eliminating the need for high-altitude climbing and significantly reducing the risks involved for inspectors.&nbsp;<img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNjgwMTI0MTA2NTMyLTE2ODAxMjQxMDY1MzIucG5nIiwiZWRpdHMiOnsicmVzaXplIjp7IndpZHRoIjo5NTAsImZpdCI6ImNvdmVyIn19fQ==" class="fr-fic fr-dii">&nbsp;The second asset on the inspection list was a complex network of&nbsp;pipes. Inspecting these structures comprehensively requires examining them from multiple angles, which can be challenging and may require disassembling the pipes and elbows, causing disruptions and delays in refinery operations. Additionally, these structures may be located in hard-to-reach areas or obscured by other equipment, further complicating the inspection process.&nbsp;Intelligent flying robots have revolutionized the inspection process by allowing easy access to difficult-to-reach areas and providing multiple inspection angles, even while the pipes are still in service. This has greatly improved efficiency and reduced downtime during inspections. The intelligent flying robot SKYRON has a Rotary Arm that can adapt to any surface slope, providing measurements within a 2&quot; x 2&quot; square. The arm pivots 180 degrees to accommodate any gradient, regardless of orientation, ensuring that even the most challenging pipe surfaces can be inspected thoroughly.&nbsp;During the inspection, thanks to its advanced technology, SKYRON took 94 UT readings in 35 minutes of flight.&nbsp;<img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNjgwMTI0MTA2NjM5LTE2ODAxMjQxMDY2MzgucG5nIiwiZWRpdHMiOnsicmVzaXplIjp7IndpZHRoIjo5NTAsImZpdCI6ImNvdmVyIn19fQ==" class="fr-fic fr-dii">&nbsp;The third asset in question was an in-service flare stack, which posed a significant challenge to inspect due to its towering height and extreme temperature conditions.&nbsp;Flare stacks require shutting down&nbsp;operations&nbsp;to reduce the heat&nbsp;and&nbsp;make the conditions safe for inspectors,&nbsp;resulting in reduced productivity.&nbsp;By utilizing intelligent flying robots for inspections, the need for shutdowns can be eliminated. The robot remains in contact with the surface being inspected, while inspectors can maintain a safe distance, ensuring maximum efficiency and safety.&nbsp;The advanced technology of&nbsp;SKYRON&nbsp;allowed it to navigate the high-temperature&nbsp;surface and took 27 readings in 8 minutes of flight time.&nbsp;Using flying robots for NDT inspections proves that technological advancements&nbsp;can improve efficiency and safety in the&nbsp;oil&nbsp;and gas&nbsp;industry. With the ability to perform inspections faster, cost-effectively, and safely, intelligent flying robots like SKYRON can reduce downtime and increase productivity, saving refineries time and money. As the industry continues to evolve, it is clear that the use of advanced technology will play a vital role in ensuring the safety and reliability of equipment.&nbsp;

Refineries are known for operating under harsh conditions of high pressure and temperature, which can cause damage to equipment over time, leading to defects such as internal corrosion, cracking, and erosion. 

unmanned aerial vehicles (UAVs)