Projects

CAD2CAV: Computer Aided Design for Cooperative Autonomous Vehicles

CAD2CAV is a project focusing on multi-agent exploration in unknown environments. It attempts to build a complete system from perception to planning and control, exploring a designated unknown environments with multiple autonomous vehicles. It is built in xLab at the University of Pennsylvania, on multiple F1TENTH race cars. Useful documents: GitHub link Spring 2021 Final Report (not complete) Demo This is a sample scenario of a single F1TENTH race car exploring the 2nd floor of Levine Hall at the University of Pennsylvania, right outside xLab.

December 27, 2021 Read

Volume Rendered ReSTIR in Vulkan

This is a project of a Vulkan implementation of Fast Volume Rendering with Spatiotemporal Reservoir Resampling. It achieves the following: Vulkan ray tracing pipeline with nvpro and Vulkan Ray Tracing KHR extension. Volume assets loading and rendering through OpenVDB. ReSTIR algorithm rendering on GLTF scene and volume assets. Authors Zhihao Ruan (ruanzh@seas.upenn.edu), Shubham Sharma (sshubh@seas.upenn.edu), Raymond Yang (rayyang@seas.upenn.edu) Tested on: Windows 10 Home 21H1 Build 19043.1288, Ryzen 7 3700X @ 3.

December 27, 2021 Read

Vulkan Grass Rendering

Tested on: Windows 10 Home 21H1 Build 19043.1288, Ryzen 7 3700X @ 3.59GHz 48GB, RTX 2060 Super 8GB Highlights This project implements physically-based grass rendering & culling with Vulkan compute shaders: Physically-based real-time rendering of grass blades; 3 different culling tests: orientation culling, view-frustum culling, distance culling; Tessellating Bezier curves into grass blades with GLSL tessellation shader A detailed instruction of this project can be found here. Vulkan Vulkan is considered as the next-generation graphics API developed by Khronos group, in replacement for the old OpenGL.

December 27, 2021 Read

CUDA Path Tracer with À-Trous Denoiser

Tested on: Ubuntu 20.04 LTS, Ryzen 3700X @ 2.22GHz 48GB, RTX 2060 Super @ 7976MB CUDA Path Tracer Highlights Finished path tracing core features: diffuse shaders perfect specular reflection 1st-bounce ray intersection caching radix sort by material type path continuation/termination by Thrust stream compaction Finished Advanced Features: Refraction with Fresnel effects using Schlick’s approximation Stochastic sampled anti-aliasing Physically-based depth of field OBJ mesh loading with tinyobjloader Background: Ray Tracing Ray tracing is a technique commonly used in rendering.

December 27, 2021 Read

My F1TENTH Journey — Lab 6, Pure Pursuit (Waypoint Tracker)

This series of blogs marks the journey of my F1/10 Autonomous Racing Cars. All my source codes can be accessed here. Previous post: My F1TENTH Journey — Lab 1, Introduction to ROS My F1TENTH Journey — Lab 2, Automatic Emergency Braking My F1TENTH Journey — Lab 3, PID-Controlled Wall Follower My F1TENTH Journey — Lab 4, Reactive Planning Methods for Obstacle Avoidance Overview In autonomous driving development, recording waypoints for future testing is very important.

January 29, 2021 Read

My F1TENTH Journey — Lab 4, Reactive Planning Methods for Obstacle Avoidance

This series of blogs marks the journey of my F1/10 Autonomous Racing Cars. All my source codes can be accessed here. Previous post: My F1TENTH Journey — Lab 1, Introduction to ROS My F1TENTH Journey — Lab 2, Automatic Emergency Braking My F1TENTH Journey — Lab 3, PID-Controlled Wall Follower Overview Obstacle avoidance is one of the critical parts of autonomous driving. Crashing into obstacles could possibly bring serious damages to both the car and the object that it runs into.

January 27, 2021 Read

My F1TENTH Journey — Lab 3, PID-Controlled Wall Follower

This series of blogs marks the journey of my F1/10 Autonomous Racing Cars. All my source codes can be accessed here. Previous post: My F1TENTH Journey — Lab 1, Introduction to ROS My F1TENTH Journey — Lab 2, Automatic Emergency Braking Lab Materials The lab materials can be accessed here. A PDF version is also attached here. Previous Next Page: / The lab was built on the F1Tenth Simulator, which can be accessed here.

January 27, 2021 Read

My F1TENTH Journey — Lab 2, Automatic Emergency Braking

This series of blogs marks the journey of my F1/10 Autonomous Racing Cars. All my source codes can be accessed here. Previous post: My F1TENTH Journey — Lab 1, Introduction to ROS Overview This lab focuses on implementing an AEB (Automatic Emergency Braking) in ROS node for F1/10 racing cars. AEB is widely used in autonomous vehicles as a basic safety guarantee to avoid collisions with objects. The lab materials can be accessed here.

January 25, 2021 Read

My F1TENTH Journey — Lab 1, Introduction to ROS

This series of blogs marks the journey of my F1/10 Autonomous Racing Cars. All my source codes can be accessed here. Overview In Spring 2021 I finally got a chance to join the F1/10 Autonomous Racing Car Lab at Penn, supervised by Prof. Rahul Mangharam. To start my autonomous racing journey, I followed the official tutorials and studied the labs & projects there. ROS (Robotics Operating System), is a robotics project development platform widely-used all across the world.

January 24, 2021 Read

Minecraft Game Programming

In the fall of 2020, I finally had a chance to explore the field of computer graphics at University of Pennsylvania, and did a very interesting course project which builds a simplified Minecraft from scratch. This game is built on OpenGL 3.2 with Qt 5.15.0. Through building this game, I learned a lot about OpenGL rendering pipeline, game engine & texture mapping. Demo Video Overview Minecraft is originally well-known for its block-like structure.

December 9, 2020 Read

Real-Time On-Device Flow Statistics Detection and Prediction

This project was served as the Undergraduate Major Design Experience at UM-SJTU Joint Institute, Shanghai Jiao Tong University. Background: The Impact of COVID-19 The outbreak of novel coronavirus COVID-19 in 2020 has brought huge attention to the administration of human society. “Social distancing” has been advocated as a very effective method to fight the virus. In order to facilitate the implementation of social distancing, we propose an idea of building a portable device that is capable of detect and predict flow statistics in real time, in order to provide insights and suggestions for traffic management.

August 29, 2020 Read

SLAM — Simultaneous Localization And Mapping

Simultaneous Localization And Mapping, also known as SLAM, is a set of very important techniques in robotics, and it’s still in active research. It aims at developing different algorithms for figuring out the position of the robot in the real world by the robot itself.

January 13, 2020 Read