| CARVIEW |
Select Language
HTTP/2 200
server: GitHub.com
content-type: text/html; charset=utf-8
last-modified: Sat, 22 Nov 2025 07:52:46 GMT
access-control-allow-origin: *
strict-transport-security: max-age=31556952
etag: W/"69216bce-993c"
expires: Sun, 28 Dec 2025 16:07:51 GMT
cache-control: max-age=600
content-encoding: gzip
x-proxy-cache: MISS
x-github-request-id: 0F8B:21D6A4:7DBCD5:8CD1C9:6951537C
accept-ranges: bytes
age: 0
date: Sun, 28 Dec 2025 15:57:51 GMT
via: 1.1 varnish
x-served-by: cache-bom-vanm7210041-BOM
x-cache: MISS
x-cache-hits: 0
x-timer: S1766937471.347916,VS0,VE214
vary: Accept-Encoding
x-fastly-request-id: 07b05fb5e350c7cc6237b88a12ea11de6782d689
content-length: 8179
Tyler Lum
Tyler Lum
Building intelligent robots with learning-based perception and control
Hi, my name is Tyler Lum and I am a 4th year Computer Science PhD student at Stanford University studying artificial intelligence and robotics. I am advised by Professor C. Karen Liu in The Movement Lab (TML) and Professor Jeannette Bohg in the Interactive Perception and Robot Learning (IPRL) Lab, and I am supported by an NSERC Postgraduate Scholarship (PGS-D).
I am broadly interested in building robots that can move and act in dynamic real-world environments in an elegant and efficient manner. My research focuses on developing the most effective ways to integrate learning-based perception and control, and studying the inductive biases we can exploit to improve the efficiency and reliability of these systems. I hope to uncover the underlying principles that enable intelligent agents to reason through uncertainty, continuously learn from their environments, and adapt to new challenges. My long-term goal is to create robots that have some form of common sense reasoning, which will make them reliable enough to create tremendous value for people in their everyday lives.
Before my PhD, I studied Engineering Physics at the University of British Columbia (UBC), where I graduated as a Wesbrook scholar - one of UBC's most prestigious designations given to the top 20 overall senior students. I was advised by Professor Michiel van de Panne studying reinforcement learning and motion planning for quadruped robots. I also worked with Professor Purang Abolmaesumi on deep learning for medical image analysis.
Research Interests
Elegant Motion
Interactive Perception
Dexterous Bi-Manual Manipulation
Reasoning through Uncertainty
Long-Horizon Planning
Creative Problem Solving
Common Sense Reasoning
Exploration Strategies
Experience
Robotics Research Intern at RAI (Jun 2025 - Present)
I am working on training reinforcement learning policies for robust and reactive humanoid loco-manipulation.
Robotics R&D Intern at NVIDIA (Jun 2023 - Feb 2024)
I designed and trained DextrAH-G: a safe, continuously reacting pixels-to-action policy that achieves state-of-the-art dexterous grasping in the real world and was trained entirely in simulation using RL and teacher-student distillation with a geometric fabric controller
Robotics R&D Intern at NVIDIA (May 2022 - Aug 2022)
I built an obstacle-aware global navigation system for robotic arms using a combination of deep reinforcement learning and supervised learning to quickly move through complex obstacle distributions without getting stuck in local minima
Deep Reinforcement Learning Research at UBC Motion Control and Character Animation Lab (Sep 2021 - Jun 2022)
I developed a quadruped robot control policy for discontinuous terrain (gaps, stepping stones) using a combination of learning-based and model-based control for robust sim-to-real transfer.
Autopilot Software Engineering Intern at Tesla (May 2021 - Aug 2021)
I developed safety-critical C++ software for Tesla's Full Self-Driving Beta system for planning and trajectory optimization.
Medical AI Research at UBC Robotics and Control Lab (Sep 2020 - Apr 2021)
I spearheaded the development of an attention-based deep learning system that detects COVID-19 pneumonia from lung ultrasound videos. I published this work as the first author at the Advances in Simplifying Medical UltraSound 2021 conference and was awarded Best Presentation - Runner-up.
Embedded Systems Software Intern at Tesla (May 2020 - Aug 2020)
I developed firmware for Tesla seat motors, heaters, pumps, and sensors.
Software Engineer in Robotics at Kardium (Sep 2019 - Dec 2019)
I built robotic systems that construct and validate a medical device that treats cardiac disease.
Robotics Software Developer at Open Robotics and Google Summer of Code (May 2019 - Aug 2019)
I developed ROS software in C++ for an autonomous surface-vessel simulator called Virtual RobotX (VRX), and published this work as a coauthor at the OCEANS 2019 MTS/IEEE Seattle conference.
Robot Localization and Navigation Developer at A&K Robotics (Jan 2018 - Apr 2018)
I improved the quality assurance and development speed on an autonomous floor-cleaning robot.
Publications
Crossing the Human-Robot Embodiment Gap with Sim-to-Real RL using One Human Demonstration
Tyler Ga Wei Lum*, Olivia Y. Lee*, C. Karen Liu, Jeannette Bohg
Conference on Robot Learning (CoRL) 2025
Scaffolding Dexterous Manipulation with Vision Language Models
Vincent DeBakker, Joey Hejna, Tyler Ga Wei Lum, Onur Celik, Aleksandar Taranovic, Denis Blessing, Gerhard Neumann, Jeannette Bohg, Dorsa Sadigh
Neural Information Processing Systems (NeurIPS) 2025
DextrAH-G: Pixels-to-Action Dexterous Arm-Hand Grasping with Geometric Fabrics
Tyler Ga Wei Lum*, Martin Matak*, Viktor Makoviychuk, Ankur Handa, Arthur Allshire, Tucker Hermans, Nathan D. Ratliff, Karl Van Wyk.
Conference on Robot Learning (CoRL) 2024
Get a Grip: Multi-Finger Grasp Evaluation at Scale Enables Robust Sim-to-Real Transfer
Tyler Ga Wei Lum*, Albert H. Li*, Preston Culbertson, Krishnan Srinivasan, Aaron Ames, Mac Schwager, Jeannette Bohg.
Conference on Robot Learning (CoRL) 2024
Neural Attention Field: Emerging Point Relevance in 3D Scenes for One-Shot Dexterous Grasping
Qianxu Wang, Congyue Deng, Tyler Ga Wei Lum, Yuanpei Chen, Yaodong Yang, Jeannette Bohg, Yixin Zhu, Leonidas Guibas
Conference on Robot Learning (CoRL) 2024
Reinforcement Learning Enables Real-Time Planning and Control of Agile Maneuvers for Soft Robot Arms
Rianna Jitosho*, Tyler Ga Wei Lum*, Allison Okamura, C. Karen Liu.
Conference on Robot Learning (CoRL) 2023
What Makes Certain Pre-Trained Visual Representations Better for Robotic Learning?
Kyle Hsu, Tyler Ga Wei Lum, Ruohan Gao, Shixiang Shane Gu, Jiajun Wu, Chelsea Finn.
NeurIPS 2022 Foundation Models for Decision Making Workshop
Imaging Biomarker Knowledge Transfer for Attention-Based Diagnosis of COVID-19 in Lung Ultrasound Videos
Tyler Lum, Mobina Mahdavi, Oron Frenkel, Christopher Lee, Fatemeh Taheri Dezaki, Mohammad H. Jafari, Nathan Van Woudenberg, Ang Nan Gu, Purang Abolmaesumi, Teresa Tsang.
MICCAI Advances in Simplifying Medical UltraSound (ASMUS) 2021
Toward Maritime Robotic Simulation in Gazebo
Brian Bingham, Carlos Agüero, Michael McCarrin, Joseph Klamo, Joshua Malia, Kevin Allen, Tyler Lum, Marshall Rawson, Rumman Waqar.
OCEANS 2019 MTS/IEEE SEATTLE
Projects
Data-driven Mixed Martial Arts (MMA) Automated Scoring
When an MMA fight ends without a finish, judges render a decision based on a scoring system that can be quite subjective and prone to error. To address this issue, I built a dataset consisting of thousands of UFC fights, and used it to train an ML-based automated scoring system for MMA fights to produce unbiased fight decisions.
Virtual RobotX (VRX): Simulator for Autonomous Surface Vessels
VRX is an autonomous boat simulator that uses ROS and Gazebo for physics, rendering, and communication. I developed software to add features for customization and control, enhance the simulator's overall realism, and improve the infrastructure for competitions that use VRX.
UBC Sailbot: Fully Autonomous Sailboat
I led the development of pathfinding and perception algorithms for a fully autonomous sailboat named Raye. Raye will be the first fully autonomous sailboat to sail from Victoria to Maui (over 4000km) in 2022.
Vim Configuration for Fast Text Editing
Vim is a highly efficient and configurable text editor that I love. I shared the Vim configuration file I use for fast text editing, and I wrote articles about Vim that have over 60,000 views.
Contactless Vitals Monitoring System
Research from Press Ganey Associates found that the average ER wait time in 2009 was 6 hours, which is plenty of time for a patient's condition to worsen significantly before they are seen. To address this issue, we used a Kinect sensor to monitor the breathing of multiple people simultaneously, and send an alert if the breathing rates are outside of normal ranges.
Fully Autonomous, Multi-Robot System
We designed and built a fully autonomous, multi-robot system from scratch in 13 weeks. We developed mechanical, electrical, and software systems that allowed the robots to navigate through a complex obstacle course.
Volta Flex: Electromyography (EMG) Garment for Injury Recovery
The VOLTA Flex V1 is a garment with integrated EMG sensors that supports injury recovery. It allows users to visualize their muscle activation during strengthening exercises to support faster recovery.
Lumina: An Improved Toolkit for Central Venous Access
When a patient has critical liver deterioration and a rapid drop in blood pressure, physicians must set up a central venous access line to help normalize their blood pressure. If this is not done quickly enough, the patient will die. However, the current process is cumbersome and error-prone. We worked with an emergency room doctor to develop tools for faster setup of a central venous access line.
Get In Touch
If you're interested in my work, feel free to contact me. I would love to connect!
- tylerlum at stanford dot edu
- Vancouver, BC, Canada