Kexin Shi (石可心)
Email: kexins@andrew.cmu.edu
I am a visiting scholar at CMU Robotics Institute advised by Prof. Deepak Pathak. Previously, I was a master research student at Robotics and Perception Group (RPG) led by Prof. Davide Scaramuzza. My current interests focus on developing trustworthy embodied robotic agents using onboard sensing with real world validation.
During my master, I also have research experiences at Advanced Interactive Technologies Lab (AIT) led by Prof. Otmar Hilliges, Robotic Systems Lab (RSL) led by Prof. Macro Hutter and Computer Vision and Geometry Group (CVG) led by Prof. Marc Pollefeys. Prior to master, I obtained Bachelor of Engineering in Computer Science at Xi’an Jiaotong University in 2020.
I have spent many colorful summers in different places. In 2022, I attended ETHz Robotics Summer School. In 2019, I participated in NUS SOC Summer WorkShop. In 2018, I was an exchanged summer student at Shanghai Jiao Tong University.
News
| Nov 10, 2023 | It is super exciting to meet lots of new friends at CoRL 2023 in Atlanta and discuss about our extreme parkour work. We also show a live demo, please check! |
|---|---|
| Sep 16, 2023 | Our work Extreme Parkour with Legged Robots is released! |
| Jul 24, 2023 | I arrive in Pittsburgh and officially start my visiting at CMU Robotics Institute. |
| Jun 10, 2023 | I have spent a wonderful week at ICRA 2023 in London, United Kingdom. It is my pleasure to present my work as well as meet many new friends! |
| Jan 17, 2023 | Our work Learning Perception-Aware Agile Flight in Cluttered Environments is accepted by ICRA 2023. |
| Jan 15, 2023 | I start as a research assistant at Advanced Interactive Technologies (AIT) advised by Dr. Xi Wang, working on Human Object Interaction. |
| Sep 10, 2021 | I join in Robotics and Perception Group (RPG). |
Publications
 |
CoRL 2023 Generalist / Roboletics / Deployable Workshop (Oral)
An end-to-end data-driven approach that can scale to the challenging task of precise and extreme parkour with onboard sensing. |
 |
ICRA 2023
A learning-based method that achieves perception-aware, minimum-time, vision-based flight in cluttered environments. |
Projects
 |
An offline python SLAM using COLMAP and ORB-SLAM which is robust, accurate, and highly extensible. |
 |
An accurate toolbox for localization and pose estimation of ANYmal without external sources like depth cameras or QR codes. |
 |
A point-voxel-based model for efficient spatio-temporal processing of event data in classification task and optical flow regression task. |
Personal
I spend my spare time in reading, painting and traveling. More photographs can be found in my Instagram.
Particularly, I also enjoy some thrill sports to challenge myself.
Life is waiting us to explore more. Let's keep moving.