REU Experiences Uncategorized

Week Ten

Becoming a pilot may be one of the coolest things someone can do. I recently learned that some of our lab directors are also pilots and fly aircraft locally in Lincoln fairly often. While I already thought they were cool, the overall cool-factor certainly spike a bit. While I can’t fly planes (yet), this week I passed the FAA Part 107 Drone Pilot test! Early in the semester, the lab offered to fund our test and a group of us can now call ourselves certified drone pilots. This means that we can legally operate small UAVs in an official capacity. Studying for the test was actually super interesting because I learned a lot about airspace regulations, something I knew nothing about prior to entering the program. While this was a great way to start our week, there were certainly many bittersweet moments toward the end. 

I learned so much this summer. I worked with Robot Operating System (ROS), learned Python, explored neural networks, and dove into a huge data set. I also learned so many important pieces of the research process, from writing effective papers to understanding how research is funded. But for me, I think the relationships I developed will be the thing I value most. Our cohort was eleven people from all over the country (and the world!). While the majority of us studied computer science, we also had people from multiple engineering disciplines and even psychology. We worked really well together and I think our diverse backgrounds helped us to that end. I also appreciate the relationships I developed with my graduate student mentors and faculty advisors. From navigating graduate school to identifying impactful research topics, my mentors have helped me imagine my own future in research. 

Thursday was our final day in the SRP program (I’m currently writing from Philadelphia!) and it was certainly an exciting day. We began in the morning with our culminating research symposium. During the first hour, I spent time judging my assigned posters (it was a competition) and the second hour presenting my own work. The energy was so great in the room and I was glad for the opportunity to learn about some of the great research being done around campus! We ended the evening with a formal banquet, which was a great way to finish out the program. All of the faculty advisors got up to say great things about their summer researchers and awards were extended to those who did superior work. I am proud of one of our own cohort members who was awarded second place out of the entire SRP program for his poster presentation! While the banquet was great, afterward we all had to do the hard work of saying goodbye to one another. After lots of great pictures, I went home to pack up my thing and get ready to leave. 

This summer I made amazing relationships and I can’t believe how much I learned in such a short time! My experience in the NIMBUS lab was absolutely amazing and has certainly helped drive my goal for a future career in research. 

REU Experiences Uncategorized

Week Nine

Someone once told me that with writing, you only ever have a good draft; a final draft doesn’t truly exist. She was describing the feeling that a paper could always be better, but at some point, it needs to be done enough. The first time I ever experienced this was with art. For a painter, the piece can always be better, the details adjusted or the colors enhanced. I think this dilemma exists with any creative work and is best approached with a good iterative review process. It’s like minimizing a cost function; there may be peaks and valleys, but ultimately each iteration brings the work closer to where it should be. 

This week the first full draft of my project report was due. Early on in the semester, I spent significant time working on the literature review and was surprised to find that there isn’t a significant body of work that exists on UAV operator mode classification. So, I decided to spend time with research related to the general classification of UAV behavior. Similar to an iterative writing process, immersion in a body of research feels like it starts off with a broad net and a large margin of error. Much of the work I read was only tangentially related to my specific project but ultimately helped me position the work in a larger context. As I continue to iterate over my project draft and continue with the research, I imagine my net may grow much smaller, but also more precise. 

I was also reminded this week of the importance of adaptability in research. Last Friday, I realized my neural network was overfitting, something I hadn’t noticed until running a very large number of epochs. I was able to diagnose the problem and fix it (there was a class imbalance in the dataset, which I was able to offset with a weighted loss), but that also meant that I had to throw out my previous results and re-run all of my tests. I worked through the weekend and was able to get everything done in time for the final poster and rough paper draft! That said, I certainly plan to continue iterating over my paper.

In the lab this week, our lab administrators organized a potluck for the REU students – it was so great! We got to eat some home-cooked food with the lab directors and graduate students. We also had the opportunity to review our posters with the lab directors and give a two-minute overview. The lab directors gave us some great feedback, which was certainly helpful and confirmed that I need to practice. Outside of the lab, we attended a helpful presentation on giving effective research presentations. Again, we were given the opportunity to give one, three and five-minute synopses of our work. Next week, I’m excited to take the Part 107 Drone Pilot test and present at the final research symposium!

REU Experiences Uncategorized

Week Eight

The historic Haymarket district in downtown Lincoln is a charming spot where old meets new. Beautiful brick buildings and old railroad yards are intertwined with great restaurants and hip cafes. I visited a few times this week to either get work done at a cafe or take a break and enjoy the lively atmosphere. I especially enjoyed a local spot called Kinkaider Brewery – the beer was great and so was the company! I may have also found my favorite cafe. It’s called the Mill and it has great tea, tons of space and the kind of wooden floors that creak when you walk on them. While at the Mill, I met a local Nebraskan who certainly embodied the concept of ‘Nebraska Nice’. We must have talked for thirty minutes about things to do in Nebraska, he was filled with great recommendations and advice! 

I’ve always found that the best advice comes from people who are directly engaged in the topic of inquiry. Early this summer, I reached out to one of the graduate students in our lab to ask questions about graduate school and a research career. John-Paul was nice enough to grab lunch with me and imparted some of his sage advice and our meeting was certainly encouraging. That must have been six weeks ago and the time has flown by! This week, we were lucky to have been able to sit in on John-Paul’s Ph.D. dissertation defense! During the presentation, I was not only struck by how knowledgable John-Paul was in his research area, but also his facility in communicating complex concepts in digestible ways. It was really interesting to see both a Masters thesis and Ph.D. dissertations defenses almost back-to-back this week and last! 

Our own poster presentations are rapidly approaching! Luckily, Dr. Duncan led a discussion this week on giving effective one, three and five-minute presentations. As a group, we had the opportunity to discuss our research within a given timeframe and it was certainly an eye-opening experience. Before our symposium, I plan to practice a lot to make sure everything runs smoothly on presentation day. Our final posters are due on Tuesday with the symposium the following week. While I’m confident in my poster and my research in general, I still plan to spend the majority of the weekend making sure everything is in order. In the context of the research, I noted a small problem with the neural network performance, which I believe may be related to variance in the size of the data set for each separate class. Other than that, I spent some time reading additional material this week, writing some small scripts and training neural networks (sometimes overnight!). It’s been a great week and I look forward to getting my final poster completed! 

REU Experiences Uncategorized

Week Seven

In my opinion, winter isn’t fun. Every year in Boston, it feels like winter lasts forever while summer tends to fly by. This week I regularly reminded myself of this as the temperatures rose well above 100 degrees Farenheight. A heatwave has been spreading across the country and it has certainly been toasty here in Lincoln. While people have been finding creative ways to keep cool (i.e. wading through the campus fountain), all I’ve had to do is think about the larger context and appreciate the distinct lack of snow. 

This week I also found myself thinking about my project in a larger context, specifically regarding data processing. Starting with UAV position, I had originally thought to calculate average velocities with respect to varying slices of time. While the neural network was able to classify the data with a decent level of accuracy, I ran into problems that may suggest overfitting. I took some time to think about the data again and decided to change the way I was processing. Instead of average velocities over varying time slices, I will be looking at progressions of instantaneous velocity and acceleration. I hope this will allow the network to better draw patterns in UAV behavior. So, I spent time this week writing scripts that can process the data toward those ends. In addition, I revised my literature review and spent some time reviewing additional papers focused on behavioral classification research. I also completed the first draft of my research poster. In addition to my own work, I was thrilled to sit-in on a thesis defense from Ashraful, one of the graduate students in our lab. His research focused on the development of a mechanism for atmospheric profiling. It was interesting to hear about his work, but also to see what a masters thesis defense entailed. 

As a group this week, our cohort got to hear Dr. Bradley give a talk on how to prepare conference and journal paper manuscripts. We saw some great high-quality examples while learning strategies and applications that make it easier to create professional manuscripts. Outside of the lab, we also participated in a graduate student panel. Four graduate students from different departments discussed their experiences coming to the UNL graduate school and answered audience questions. They had some really great advice! Next week, I plan to conduct additional experimentation focused on the progression of velocity and acceleration, finalize my poster and integrate some additional sources to help round out my literature review.

REU Experiences Uncategorized

Week Six

A few years ago, a good friend of mine convinced me to attend one of her yoga classes. I had never liked the idea of yoga, in large part because my body has never been very flexible, but I decided to give it a try. Fast-forward to today and yoga has become an integral part of my life. I find that the practice helps me feel both grounded and centered, especially sequences that focus on balancing postures. That’s why I was especially curious when I discovered stand-up paddleboard yoga was being offered through the campus outdoor adventure center. Wednesday this week we arrived at the lake in the early evening. I had never used a paddle board before, but after a few wobbly minutes, I got the hang of it. Our instructor then led us through a sequence that culminated in some challenging balancing postures. As could be imagined, this resulted in more than a few people plopping into the lake! Back on shore, we ate smores and finished out the evening watching a perfect sunset over the lake. It was a great night!

Balance has been my mantra this week. In general, it seems like one research question leads to many more and it can be easy to get lost along multiple investigative paths. That’s why I continue to remind myself of the need to balance scientific curiosity with focused completion of specific project deliverables. This week I created a simple feed-forward dense layer neural network and ran my data through it for the first time. I was pleasantly surprised to find that the network was classifying the data with an initial accuracy over seventy percent, with lots of room for improvement. Since then, I’ve adjusted the data input slice size (allowing for incrementally larger representations of change in flight path) and the number of epochs (the number of times the network iterates over the data) and have increased the network’s accuracy to more than eighty-six percent. 

While I plan to continue making adjustments to improve accuracy (enhanced network architecture, data input-methods, etc.), I also made sure to spend time on specific project deliverables such as revising my literature review, drafting my methods and results, and outlining the sections of my poster. My graduate student mentor Mike gave me some great feedback on my literature review which I have been implementing. In the lab this week, Dr. Detweiler led a discussion that focused on thinking critically about the research process. We were each asked to give a two-minute description of our projects. This was helpful for me because I had to think about explaining my project in a succinct but comprehensible manner. We also learned more about how the research funding process functions. Overall it’s been a great week and the next will be very busy as well as our first poster draft is due!

REU Experiences Uncategorized

Week Five

Yesterday was the fourth of July and I must admit, I’ve never seen as many fireworks as I did yesterday. There were so many displays that downtown Lincoln was actually filled with smoke! The night prior, some friends and I attended a local celebration called Uncle Sam Jam which culminated in one the best-choreographed fireworks displays I’ve ever seen. While I’ve seen many fireworks shows, this one stood specifically for it’s consistent build-up to a truly awesome finale. 

We are ending week five of the research experience and our own finale is looming ever closer. As we are halfway through the program, it’s truly time to put the pedal to the metal. I started off this week working on a Python script to iterate over and parse what I need from the project data set. After spending the prior week engrossed in the literature review, it felt good to be programming again. Ultimately, I was able to write a script that parsed out the data points I needed and also classified operation mode (autonomous, semi-autonomous or manual) for all of the recorded flights. I also received some helpful feedback on my literature review from Dr. Duncan, so I spent time synthesizing the work and clarifying the research gap I am looking to help fill. 

Within the lab, we participated in a great discussion led by Dr. Sharif on designing effective experiments. Dr. Sharif highlighted many useful research design components and positioned them within different frameworks such as positivism and interpretivism. After this meeting, we gathered with the PIs and graduate students to view the world cup soccer match! We were fed snacks and got to watch the U.S. beat England (pretty cool!). Outside of the lab, we attended a workshop on writing personal statements for graduate school applications. We were given some great tips and also had the opportunity to review actual essays from graduate students who were admitted to UNL’s graduate programs.

Next week I intend to spend the majority of my time developing a simple neural network with Tensorflow. I will also be thinking about to slice up the data for best results!

REU Experiences Uncategorized

Week Four

This week marked the first Democratic debates for the upcoming presidential election. As an avid consumer of all things political, I was thrilled to go out and view the spectacle with some friends in downtown Lincoln. This election cycle is especially interesting, if only for the fact that there are more than twenty candidates vying for the Democratic nomination. As I watched impassioned speeches and twisting polemics, I found myself wondering how on Earth were they going to narrow the field. 

That same question echoed in my mind as I finished my first pass on approximately twenty-five academic papers for my literature review. An effective literature review will explore overlapping concepts in any relevant work while exposing any gaps in knowledge that may exist. Ultimately, a clear conceptual synthesis should arise as the review progresses. In the context of my own work, I spent time reading papers that focused on behavioral analysis and classification in a robotics context. I was hoping to find previous research that examined machine learning approaches to identify UAV mode of operation (autonomous or piloted). Instead, I found a variety of studies that classified UAV behavior in the context of sense and avoid system development, flight path prediction, and human-robot interaction. Under the guidance of Mike Turner, my graduate student mentor, I developed an initial synthesis of the work and completed a draft I’m proud of. In addition to all of the reading and writing this week, I spent time learning how to use LaTex, a document preparation system that makes it very easy to create well-formatted academic papers. 

Within the lab this week we attended two great meetings. First was a general lab meeting including all of the REU students, graduate students, and directors. We were fed pizza and salad while we all gave an update on the work we were currently doing. This was a great way for me to get to know all of the interesting work being done by the graduate students in the lab. After that meeting, Dr. Bradley led a useful discussion for the REU students on data management. We talked about the requirements generally associated with data management in the context of a research study, as well as specific requirements related to human-subjects research. Dr. Bradley outlined a variety of strategies that he employs to organize, secure and back-up his data. 

Outside of the lab, we had the opportunity to participate in a mock research symposium which involved multiple poster presentations from current graduate students. I head from three graduate students from three different disciplines: mechanical engineering, molecular biology, and physics. At the end of each presentation, we were tasked with evaluating their performance using the same metrics that will be used to evaluate us at the end of the summer. Overall, it was a very useful exercise! First thing next week I will be writing a script to iterate over my data set to parse the information I need and will hopefully have time to begin creating the neural network. I’m looking forward to writing code again next week!

REU Experiences Uncategorized

Week Three

I’ve started to discover a love for rock-climbing. The UNL Outdoor Adventure Center has a great rock-climbing gym that’s easily accessible for students and affiliates. After taking a two-hour belaying class, I took and passed a test that allows me to rock-climb for free for the rest of the summer! While climbing requires a variety of skills, including agility, stamina, and strength, the most interesting quality of the sport is how analytical it can be, specifically when bouldering. It turns out, each climbing path is like a puzzle, with only a few different possible ways to successfully make the climb. It’s an exercise for both the brain and body!

Sometimes starting a literature review can feel like a puzzle, thinking about what papers, concepts, and tools fit best in an overview of relevant work. This week, Dr. Duncan led a really helpful discussion about reading and organizing academic papers. Reading papers can sometimes seem like a daunting task, especially during the initial stages of a project. But Dr. Duncan laid out a process involving multiple stages and passes that makes the process much more manageable. I also liked her suggestion to use an organizational tool that outlined a thematic coding of the various papers which ultimately helps add structure to a literature review. As we have the first draft of our literature review due next week, this discussion has proven both useful and timely.

Early on in the week, Dr. Duncan and I sat down to formalize my project for the summer. I will be leveraging a data set from another project to build and train a neural network that can identify if a drone is flying autonomously or if it is piloted. I’m interested in machine learning and robotic systems, so I am thrilled to participate in this project. Identifying drone behavior and intent can be challenging in many contexts, so I hope to explore those challenges while thinking about how the results of my own project could apply. I imagine that understanding if a drone if piloted or autonomous could have applications to research in human-robot interaction, development of “sense and avoid” technologies, and enhancements to drone security.

Much of my week has involved wrapping my head about the literature that already exists and I completed a first-pass reading of approximately twenty papers in the field. This upcoming week, I will narrow the field and write a rough draft literature review. In addition to reading, I have spent time reviewing Tensorflow (Google’s machine learning platform), completing an overview of the data set, and outlining how I will iterate over the data set to parse the information I will need. It’s been a busy week! I’ve also sat down with Mike and Siya, two graduate students who have been helpful in guiding my project.

Overall, it’s been a great week! Early next week, I intend to spend the majority of my time on my literature review and will begin writing a script to parse my data toward the end of the week. I’m looking forward meetings including a discussion that Dr. Bradley will be leading on data discussion and analysis and a mock symposium for the whole program!

REU Experiences Uncategorized

Week Two

I really enjoy how organized the streets are in Lincoln, perfectly straight lines intersecting in neat orderly grids. I grew up in Boston and am accustomed to the opposite. My hometown has a lot of charm but is also known for it’s narrow and chaotic roads. For me, walking in downtown Lincoln is a breath of fresh air.

“Organization” has been my keyword this week. We only have eight weeks left in the program and that time is going to fly by. So, I’ve taken time this week to make sure I’m organized and ready to maximize the experience over the coming weeks. In general, my approach is to break a project up into parts, each one leading to the ultimate goal. Then I position each part on a timeline and try to overestimate how long each task will take me (underpromise, overdeliver). The most pressing issue at the moment and throughout the week has been training on ROS (Robot Operating System) and I have to say, it’s been pretty awesome.

ROS is an open-source set of libraries and tools that help developers build robotics applications. In general, ROS applications are built in C++ or Python. Initially, I worked through the tutorials using C++ as I haven’t worked in Python before. However, after the initial tutorials, I tried working in Python and have found it to be pretty straightforward (and I’m really enjoying it!). A large piece of the ROS tutorials involves working with a turtlesim node (a turtle visualization that functions similarly to a robot). This week, I spent time creating custom paths for the turtlesim node in both C++ and Python. We also wrote scripts that leveraged two XBee radio modules to send and receive control signals that move turtlesim nodes on different computers!

We also had some great discussions this week. Dr. Detweiler led a meeting on how to identify research questions that are worth asking. We talked about the value in communicating with people across industries and disciplines as well as keeping up with journals and papers within our field. The discussion was insightful and made me think about the value of interdisciplinary communication. Outside of the lab, we attended an interesting presentation on research writing. The presentation was thoughtful and interactive and encouraged us to consider the story arc of our research in our writing. Included in the presentation were research findings that suggested that scientific papers were cited more often if they followed a story-like structure.

Overall, week two has been great! I’m totally thrilled to be here!

REU Experiences Uncategorized

Week One

I arrived at the University of Nebraska, Lincoln a week ago to start my summer research internship. The National Science Foundation provides funding for students to work in research labs across the country and I was thrilled to be accepted as a student researcher. I am one of eleven students working in the NIMBUS Unmanned Systems Lab this summer. The lab is certainly one of the most interesting places I’ve ever worked, it’s an indoor drone flying facility. While it’s only week one, I’ve already done a lot, including training on ROS (Robot Operating System), studying for a drone certification exam, exploring a cube satellite and even flying drones in the lab.

We have also been very busy outside of the lab. UNL has a very well organized summer research program with students coming from all over the country to work in a variety of disciplines such as bioenergy systems or minority health disparities. Over the course of the first week, we attended multiple trainings intended to help prepare us for graduate school, understand authorship practices in research, evaluate our finances and create impactful mentor relationships. I found each one of them to be insightful, especially for those who wish to pursue a graduate degree, something I certainly intend to do.

I study computer science and mechanical engineering and hope to work with aerospace systems. At the NIMBUS Lab this summer, I will be supporting a project looking at drone system coregulation. We will be using consensus algorithms to get a multi-agent system of drones to communicate with each other and interact without any centralized control. I’m excited to contribute to the project and explore applications of the work. I’m also thrilled to work with and learn from Dr. Bradley and Chandima Fernando on this project!