• A New Mathematical Mannequin to Enhance AI and Machine Studying – USC Viterbi

    A New Mathematical Mannequin to Enhance AI and Machine Studying – USC Viterbi

    A New Mathematical Mannequin to Enhance AI and Machine Studying – USC Viterbi

    Paul Bogdan (left) and PhD scholar Jayson Sia (PHOTO CREDIT: USC Viterbi)

    Arabidopsis is a small, principally forgettable weed. However this humble plant is definitely probably the most essential species, weed or in any other case, wherever on the planet. It’s what is named a “mannequin organism” – species which are extremely studied by scientists to higher perceive nature, biology, and even people. In actual fact, Arabidopsis is among the most studied species on earth. Now, information collected from Arabidopsis is the idea for brand spanking new analysis from Paul Bogdan, affiliate professor {of electrical} and laptop engineering at USC Viterbi, his PhD scholar, Jayson Sia revealed in Nature, Scientific Reviews.

    Bogdan and his analysis group, amongst different issues, concentrate on extremely advanced mathematical fashions to higher perceive information represented in graph kind. And visualizing advanced information in graph kind is massively essential. If carried out appropriately, which is not any small feat, researchers can analyze these graphs to higher perceive every little thing from drug interactions, to radicalization on-line, to details about genetically engineered vegetation (extra on that final instance later).

    “One technique to make sense of information is to symbolize it in a graph kind. Then, even when we have no idea the patterns and guidelines behind this information, we are able to attempt to decipher them by understanding how networks, communities, and different topological varieties may change over time,” Bogdan says.

    At this time now we have extra information than ever earlier than. These information units are the cornerstone for applied sciences like Machine Studying and AI that make the trendy world run. With out having the ability to shortly entry and analyze enormous quantities of knowledge, the world we all know – and the long run engineers are serving to to construct – couldn’t exist. In different phrases, with no quicker technique to make sense of all the knowledge we acquire, that brilliant, shiny, technological future filled with self-driving automobiles and digital actuality and customized healthcare won’t ever come to fruition. Consider the mathematical fashions that Bogdan and Sia work on because the engine that powers our future.

    So, what does that every one must do with one janky little weed, you could be asking.

    What Bogdan and Sia did was take the Arabidopsis protein-protein interplay community and use it as the info set for his or her mathematical fashions and graphs. “Arabidopsis is so nicely studied, and we have already got its total genome sequenced. The scientific group additionally has an enormous quantity of information on this plant, which makes it an excellent mannequin for our analysis,” stated Sia.

    They usually did this to assist remedy an enormous downside within the graph-making world known as “group detection.” Particular person information factors, or nodes, could be misrepresented on graphs. In actual fact, they’re usually misrepresented. Let’s say you’ve put information from a social community right into a graph. Every particular person person can be one node on the graph. Because the customers work together with one another you would, presumably, be taught extra about how the social community was working. You would even perceive the way it was evolving and higher monitor issues like radicalization on-line. However in the event you’re undecided the nodes in your graph are correctly represented, you’ll be able to’t do any of that.

    Bogdan and Sia might have chosen any variety of graph fashions to check their principle on. However given the immense affect local weather change is already having on the world, they selected to give attention to a plant genome so we’d higher perceive the way to deal with meals manufacturing and sustainability in a altering atmosphere.

    “Basically, we designed a novel mathematical mannequin utilizing the Arabidopsis protein interplay as our map,” stated Sia. “Our mannequin not solely bypasses the extraordinarily sluggish course of of information evaluation and experimental validation, but it surely additionally set us on a course to higher understanding plant robustness.”

    And a greater understanding of what makes sure vegetation stronger goes to be a vital piece of data as local weather change continues to wreak havoc throughout the globe.

    Meals manufacturing is already underneath menace from local weather change in a number of methods. Not solely do altering temperatures make many areas unable to supply meals, however lethal plant pathogens and parasites are additionally shifting into new areas quicker than ever earlier than. Bogdan and Sia are actually utilizing the mannequin they based mostly on Arabidopsis and making use of it to different plant species which are immune to sure pathogens. “We might in the future be capable to use our mannequin to establish what makes sure species stronger than others. And that might assist us engineer new crops that may higher survive in a quickly altering world,” Bogdan stated.

    This analysis was in collaboration with Edmond A Jonckheere, professor {of electrical} and laptop engineering at USC Viterbi, David Prepare dinner, affiliate professor of plant pathology at Kansas State College, and Wei Zhang, educational coordinator for the Genomics Core Institute for Integrative Genome Biology on the division of botany and plant sciences at UC Riverside. 

    Revealed on November third, 2022

    Final up to date on November third, 2022

  • Can Robots Create Fashion? – USC Viterbi
    Coding and Robotics

    Can Robots Create Fashion? – USC Viterbi

    USC students in tie-dye T-shirts on USC University Park Campus.

    USC Viterbi Students from the Department of Aerospace and Mechanical Engineering pose in their new tie-dye T-shirts, created by a Baxter robot at the Center for Advanced Manufacturing at USC Viterbi. From left to right: Meghana Sagare, Danielle Redmond and Olivia Vigano. PHOTO/MEGHANA SAGARE.

    Not only is tie-dye back in fashion, but it’s one of the most beloved summer pastimes by kids and adults alike. The steps are easy: grab a T-shirt, twist it up and segment it and then dye away to create a pattern. When you’re done, you have something fun, unique and yes, fashionable once more — at least, according to Louis Vuitton and Jennifer Lopez.

    The garment industry utilizes a lot of human labor for tedious, repetitive tasks like tie-dying, because while robots thrive working with hard, rigid materials, they struggle with softer, more flexible materials. This is the challenge that Professor S.K. Gupta, director of the Center for Advanced Manufacturing at the USC Viterbi School of Engineering and holder of the Smith International Professorship in Mechanical Engineering, presented three undergraduates when they came to him seeking a hands-on robotics project.

    The students — mechanical engineering seniors Danielle Redmond and Olivia Vigano and junior Meghana Sagare — deliberated on ways to explore this challenge, finally landing on tie-dye as a fun example of how robots can support the garment industry, while emphasizing human creativity.

    “There is so little being done with fabrics right now. It’s a need in the industry, especially in manufacturing widescale clothing production. What’s cool about this project is the human gets to design and the robot gets to execute,” Redmond said.

    She added: “It cuts down on a repetitive task that a human would have to do over and over again every single day to mass produce tie-dye. So you’re taking a lot of time consuming aspects of the process and eliminating them, but still fostering the creativity of making something yourself.”

    Meet Baxter, a Robot Made to Work Near Humans

    Once the challenge was set, in came Baxter, an industrial robot created specifically for human-robot collaborations. Baxter’s arms are more flexible than typical industry robots and its joints have springs, meaning that its arms are more compliant and resistant to injuring its human counterparts. This allowed the team to experiment with a few different specifications without compromising Baxter’s performance.

    To figure out how to create a tie-dye T-shirt using Baxter, the team first researched how a person tie-dyes a shirt in the first place. Watching numerous videos, the team decided on key milestones in the tie-dye process that it would have to hit using their robot.

    “We would try to accomplish the step or task ourselves first and then ask, ‘How can a robot do this?’” Sagare said. “A lot of people used forks for the twisting of the shirt, which worked really well for us.”

    “Once we found what worked, we could really refine the process,” Redmond said. “For example, 3-D printing parts that work better than what we have on hand.”

    The team also needed to enable the robot to infuse fabric with dye. This was accomplished using tubes attached to the robot’s arms and the cups full of fabric dye. A motor was used for each cup, to help drive that specific color of liquid through the tube and onto the T-shirt.

    Once all the hardware was optimized, the students had to create code to make sure the robot could calculate the correct waypoints — spots on the fabric — in order to add dye in the appropriate pattern.

    To begin, Baxter uses a camera located on one of its arms to identify an initial red spot added to the middle of the shirt. This allows Baxter to center its workspace prior to dyeing.

    At first, the students envisioned Baxter would create the same design over and over. But as their work progressed, they decided to allow for a more creative collaboration. One key feature the team created: a drawing mechanism so that users can design the tie-dyed graphic themselves.

    A Steep Learning Curve

    The students specifically took on a project that would expose them to new arenas and challenge them. As such, they hit a few roadblocks — from troubleshooting hardware issues to learning how to program Baxter from scratch.

    “I’m a mechanical engineer, and I didn’t have a lot of experience in coding prior to working on Baxter,” Sagare said. “I did the majority of the coding in Python, but I’ve never taken a class in Python before. At the same time, Baxter runs on the robot operating system (ROS), which was like gibberish to me when I first started.”

    Meanwhile, Redmond said getting the fabric to twist properly was a formidable task. “We tried so many things,” she said.

    Vigano added: “We tried to use Baxter’s original grippers at first, but they would get stuck to the shirt when it tried to pull its hand up. Nailing down a great design that worked when it tried to twist the shirt and having it release was really important.”

    Ultimately, while Baxter is moving on to new research projects, the team is interested in continuing to explore a “human designs, robot creates” configuration.

    Said Redmond: “Being able to design a special piece of art or clothing and have the robot create it and you don’t have to be there, works really well, especially in a COVID world. This project motivated me to pursue working in a creative industry. I like the mix of art and engineering.”

    Added Sagare: “This influenced my future plans to go further into robotics engineering. It’s very interdisciplinary, incorporating mechanical engineering and design and a lot of computer science.”

    Vigano said this project opened her mind to the fact that she like working groups. “I’m motivated by working with people who have the same goals and want to work hard to get it done. I also think a project like this has the potential to excite a younger generation who like to tie dye. This can get them interested in robotics in general.”

    She added, “To have a group of three women in engineering here is pretty cool.”