About
Hi! My name is Frank, I am currently a Computer Science MS student at Northwestern University at the Design Automation of Intelligent Systems Lab advised by Prof. Qi Zhu. I also work closely with Prof. Chao Huang at University of Southampton. Presently, I am also working as a Robotics Learning Intern at the Stanford Vision and Learning Lab advised by Prof. Fei-Fei Li.
I received my Bachelor’s Degree with summa cum laude in Computer Science and Mathematics in 2024, also from Northwestern University. I had the privilege of collaborating with Professor Florian Willomitzer at the 3DIM Lab.
I have a broad interest in robotic learning. I’m fascinated by the challenge of building autonomous robots that can navigate complex environments and perform long-horizon tasks efficiently and safely. This interest extends from humanoid robotics to more common applications like self-driving vehicles. Looking ahead, I aim to design state-of-the-art learning methods that effectively assist humans in complex tasks, while prioritizing safety alongside performance.
Please see my CV for a full list of work, teaching, and other experiences.
News
- [October 2024] - Invited talk to RV 2024
- [August 2024] - Submitted POLAR-express to Embedded Systems Week 2024 tool presentation
- [July 2024] - One paper accepted to RV 2024
- [June 2024] - Started as a Robotic Learning Intern at Stanford Vision and Learning Lab
- [May 2023] - One paper accepted to MMLS 2023
Publications
F. Yang, S. Zhan, Y. Wang, et al. Case Study: Runtime Safety Verification of Neural Network Controlled System. Runtime Verification, 2024. Paper
F. Yang*, Y. Wen*. Efficient Encoding of Graphics Primitives with Simplex-based Structures. Midwest Machine Learning Symposium, 2023. Paper
Research Projects
Behavior 1K: A Human-Centered, Embodied AI Benchmark with 1,000 Everyday Activities and Realistic Simulation
Advised by: Fei-Fei Li
Sources: Project / Paper
Development on BEHAVIOR-1K: 1000 embodied-AI robotic learning simulation benchmark built upon NVIDIA Omniverse engine; Created an efficient and distributed RL approach for action primitives (pick, place, navigate); Integrated path planning and mesh prim collision detection via curobo
POLAR-Express: Efficient and Precise Formal Reachability Analysis of Neural-Network Controlled Systems
Advised by: Qi Zhu
Sources: Paper / Tool
Performed runtime safety verification case study with POLAR-express on Turtlebot3 in ROS2 simulation; Proposed online controller switch strategy for safety-critical control systems with neural networks
Efficient Encoding of Graphics Primitives with Simplex-based Structures
Advised by: Ying Wu
Source: Paper
Surveyed the encoding of graphics primitives proposed by “Instant NGP”; established theoretical foundations for simplex-based structure encodings and accelerated sample and interpolation speed on NeRF and SDF rendering with C++/CUDA kernels
Research Interests
My projects have been contributing to a common goal: creating more intelligent and adaptable robotic systems through learning-based approaches. Within robotic learning, I am particularly interested in:
Data-Driven Control: Designing safe model predictive control strategies capable of handling uncertainties and delays in dynamic environments.
Runtime Verification: Equipping robots with real-time decision-making capabilities that continuously assess the safety of learned systems, especially in the presence of neural network controlled systems.
Skill-Based Learning: Developing long-horizon skill acquisition from expert demonstration. This includes creating benchmark metrics and high-fidelity action primitive sim2real transfer.
Teaching
Graduate TA for CS340: Computer Networking taught by Alexandar Kuzmanovic, Winter 2023
Graduate TA for CS310: Scalable Software Architectures taught by Joe Hummel, Fall 2023
Undergraduate TA for CS396: Intro to Web Development taught by Sarah Van Wart, Spring 2022
Project Manager for Institute of Electrical and Electronics Engineers, 2022
Other Projects
Quadrotor Design and Control
(Mar 2024 - Jun 2024)
Source: Code
Developed a WiFi-enabled quadrotor using Raspberry Pi and IMU; implemented PID control, safety measures, and joystick interfacing in C that allows stable manual flight control; integrated Vive Lighthouse with IR sensors to enable autonomous flight control with precise 3D positioning
Reminiscia
(Dec 2022 - May 2023)
Source: Code
Implemented a multimodal text-to-image search application using pretrained vision-language models; employed Vision and CoreML to allow calculations of cosine similarity between text and image embeddings; distilled original 224MB CLIP model into an 85MB, 6-layer image encoder to improve inference speed
Transformer-based Lie Detection
(Feb 2022 - Jan 2023)
Advised by: Zach Wood Woughty
Source: Code
Conceptualized a vision-based transformer that detects lies from multimodal inputs with PyTorch; trained a transformer encoder from fine-tuning Inceptionv3; pinpointed 20 micro-gestures and facial AUs that most contribute to lying; resulted an out-of-sample lying classification of 76%
Work Experiences
Software Engineer Intern @ Target (June 2023 - Aug 2023)
Developed a Golang application within a Vela CI/CD pipeline to enforce security and compliance standards
Integrated Postgres with RestAPI for build lifecycle and versioning information retrieval and storage
Lead Tech Engineer @ Skuy (Apr 2022 - Jun 2024)
Built a college community network app startup, amassed 1000+ users on both the App Store and Google Play
Led a 2-months database migration from Heroku to Firebase for service growth and stability
Configured CI/CD pipeline on Expo for IOS Pod and Android Gradle builds
Software Development Engineering Intern @ Amazon (Jun 2022 - Sep 2022)
Implemented a Sagemaker site that provides benchmarked health & architecture evaluations for ML models
Presented a demo to Sagemaker engineers; received candidacy to beta-launch model cards on AWS Re:Invent
Skills
Languages: Python, Go, TypeScript, SwiftUI, HTML/CSS/JavaScript, C++
Robotic Learning: ROS2, Torch, CUDA, TensorFlow, OpenCV
Web/Mobile Development: React, React Native, Flask, Redux, Node, ESLint, Cypress
DevOps: RestAPI, AWS, Firebase, Heroku, Elastic Beanstalk, Git, Vela, Docker, MySQL, PostgresSQL
Cimematography
Outside of school, I am a freelance photographer taking landscape, portrait, and graduation photos. In my creative endeavor, I am a cinematographer working on film projects such as Applause For A Cause and TEDx. I am committed to creating cinematic lighting and true-story shots that evokes emotion. Check out my portfolio for my fun side!
