Felipe Felix Arias

I am a Ph.D. student and NSF Graduate Research Fellow at the University of Illinois at Urbana-Champaign (UIUC), where I work on mobile robot cognition and navigation in Dr. Nancy M. Amato's group in Parasol Lab. I received my Bachelor of Science in Computer Science with Honors from UIUC in 2019 and was fortunate to work on great projects at UIUC, Stanford, and the University of California Berkeley during my undergraduate career. On my free time I enjoy making music, traveling, and taking pictures.

felipea2 AT illinois.edu  /  Google Scholar  /  CV  /  Github  /  LinkedIn

profile photo

7/21 - The CS department at UIUC wrote an article on this year's NSF GRFP winners
6/21 - My collaboration with Google Brain was published in ICRA 2021
4/21 - I was awarded an NSF graduate research fellowship!


I am interested in navigation in human/constrained environments and relevant topics in computer vision and machine learning. More precisely, I work on algorithms that enable agents to reason about their surroundings, experience, and the agents they work with in the context of navigation. Problems/topics I am interested in include: life-long learning, spatial reasoning for obstacle avoidance, and motion planning robust to dynamic obstacles.

Avoidance Critical Probabilistic Roadmaps for Motion Planning in Dynamic Environments
[Paper] [Slides]
Felipe Felix Arias, Brian Ichter, Aleksandra Faust, Nancy M. Amato
IEEE International Conference on Robotics and Automation, 2021

We propose a self-supervised methodology for learning to identify regions critical to obstacle avoidance and leverage it to construct probabilistic roadmaps well-suited for multi-agent motion planning.

Multi-Perspective Navigation using RL
Felipe Felix Arias, Victor Gonzalez

Mutiple viewpoints should provide additional information for reinforcement learning-based navigation, this work is the first step towards determining the best way to leverage the cameras of all the agents in the scene.

Video as a Sensor for Urban Risk
Felipe Felix Arias, Daniel Carmody, Richard Sowers, Jayati Singh, Kevin So

Risky events are subjective and hard to detect due to the lack of real-world training data. We use weak-supervision and heuristics to track human behavior in the roadways, assess risk, and detect areas that may need more law enforcement or traffic signs.

Weakly-Supervised Multi-Sentence Relation Extraction
Felipe Felix Arias, Alex Ratner, Christopher RĂ©

Snorkel combines heuristics to train a neural network that can classify semantic relationships from a single sentence as well as one trained with hand-labeled data. We develop a metholodogy and novel heuristics for extracting relationships across multiple sentences.

A big thank you to Jon Barron for the website template.