WATonomous Mapping, Environment Modeling, and Decision Making
A Dynamic Relation Graph for unified prior-online environment modeling in urban autonomous driving
Portfolio
- Automated Driving 6
- Neural Networks 4
- Education 2
- Computer Vision 2
- Reinforcement Learning 1
- Trajectory Planning 1
- Control 1
- Mapping 1
- Software Architecture 1
Automated Driving
WATonomous Trajectory Planning and Control
A unified optimal control approach to trajectory planning and control via MPC.
WATonomous Action Classification
A two-stage approach to the ICCV 2021 ROAD Challenge, featuring optical flow based activity detection and a novel dynamic RoI-Alignment approach.
Reinforcement Learning for Intersection Traversal
Learning to negotiate unsignalized intersections safely via hierarchical planning and curriculum RL.
WATonomous Onboarding Workshops
Workshop series from Winter 2022 geared towards onboarding new WATonomous members.
WATonomous Tech Talk Series
WATonomous hosts biweekly AV tech talks at UW, of which I have presented at six.
Neural Networks
WATonomous Action Classification
A two-stage approach to the ICCV 2021 ROAD Challenge, featuring optical flow based activity detection and a novel dynamic RoI-Alignment approach.
Reinforcement Learning for Intersection Traversal
Learning to negotiate unsignalized intersections safely via hierarchical planning and curriculum RL.
Self-Supervised Disparity Estimation with Perceptual Loss
Building a disparity estimation model using self-supervised learning and perceptual loss to combat color-consistency assumption violations.
COVID Outcome Prediction
Building a classification model using Deep Neural Networks on the Ontario COVID-19 dataset.
Education
WATonomous Onboarding Workshops
Workshop series from Winter 2022 geared towards onboarding new WATonomous members.
WATonomous Tech Talk Series
WATonomous hosts biweekly AV tech talks at UW, of which I have presented at six.
Computer Vision
WATonomous Action Classification
A two-stage approach to the ICCV 2021 ROAD Challenge, featuring optical flow based activity detection and a novel dynamic RoI-Alignment approach.
Self-Supervised Disparity Estimation with Perceptual Loss
Building a disparity estimation model using self-supervised learning and perceptual loss to combat color-consistency assumption violations.
Reinforcement Learning
Reinforcement Learning for Intersection Traversal
Learning to negotiate unsignalized intersections safely via hierarchical planning and curriculum RL.
Trajectory Planning
WATonomous Trajectory Planning and Control
A unified optimal control approach to trajectory planning and control via MPC.
Control
WATonomous Trajectory Planning and Control
A unified optimal control approach to trajectory planning and control via MPC.
Mapping
WATonomous Mapping, Environment Modeling, and Decision Making
A Dynamic Relation Graph for unified prior-online environment modeling in urban autonomous driving
Software Architecture
WATonomous Software Architecture and Data Pipeline
A Dockerized automated driving stack, allowing for fast onboarding, prototyping, and deployment.