Project 1. [Fault-Tolerant Robot System Design Based on Fault Models]: (Performance, Software Infrastructure Design): • Developing fault-tolerant robot systems, including designing fault models to capture relationships among faults, systems, algorithms, and robot tasks. • Designing a strategy to utilize the fault models as a guiding framework for robot developers, focusing on enhancing fault-tolerant capabilities and ensuring effective fault detection and handling during development and runtime phases.Project 2. [Robot Behavior Coordination Systems]: (End-to-End System, Autonomous vehicle robot) • Collaborating within a cross-functional team to design and develop end-to-end robot systems in C++, involving high-level coordination of robot behaviors to low-level implementation of motion controllers and navigation.

Project: [Data Model for Neo4j Graph Database]: (Database, Sensor Data, Human-robot Interaction) • This volunteer position is at UT Austin's Department of Aerospace Engineering and Engineering Mechanics. • Design and create graph data models in the Neo4j database to model human-robot interaction data and relations. • Design Cypher queries w/ Python language for human-robot interaction and the sensor data.

My Ph.D. aimed to invent an end-to-end sensing-aware robot programming recommendation system that quantified the system/robot timing performance and camera sensing capability as quality measures to provide user programming recommendations.Project 1. [Sensing Quality-aware Programming System]: (Sensing, Performance, Distributed, Network, Camera, Human-robot)• Designed, architected, and implemented an end-to-end robot programming system that provided user programming recommendations to meet robot tasks' timing and camera sensing quality requirements. • Developed a scalable graph-based method to quantify the timing and camera quality metrics, automatically computed from camera geometry, system performance profiles, and robot tasks.• Implemented the framework on distributed Linux systems communicated through TCP/IP network and achieved real-time camera-based monitoring by modeling the monitoring modules as a distributed sensor network for parallel processing.• Demo video for programming & system recommendation & robot execution & fault handling: https://www.youtube.com/watch?v=CBDI0vgJHCA=> Selected Achievements: * Recommended users in selecting parameters, adding cameras, or reconfiguring workspace to meet users’ requirements and improved the system performance by at least 13 seconds for an original 58-sec robot task. * Conducted a user study resulting in reduced programming time and improved programming correctness.* Accepted in IEEE ICRA 2021 (top robot conference) and IEEE RTSS 2020 (demo, top real-time system conference)Project 2. [Programming System for Perception Attention]: (3D, Tradeoff, Cameras)• Developed a Python-based programming system in both the real-world and simulator for users to specify the location of a target object in a multi-camera environment and 3D reconstructed the object location for a 6DOF robot to pick up.• Quantified the quality of 3D reconstruction and provided tuning for tradeoffs between the quality and performance.

• "CS378: Introduction to Cyber-physical Systems" is a CS undergraduate course introducing cyber-physical systems w/ homework related to "1/10th-scale" autonomous vehicles/robots.• I assisted in assignment design and software development for the sensing/perception of an F1TENTH autonomous vehicle (NVIDIA Jetson embedded system) and the simulated Husky car in the Gazebo robot simulator.• The F1TENTH includes various sensors, such as cameras and LIDAR sensors.

1. [Dependency Detection System for Preserving Web Functionality]: (Efficiency, Back-end, Database, Web) • The internship was focused on the software development of a system that could automatically detect web dependencies to ease the efforts of non-savvy users to preserve web functionality. • Developed a BFS-based algorithm to detect web dependencies, such as external files, from 100 PHP files of "The Speech Presentation in Homeric Epic" website, achieving detection precision of 78% and recall of 82%. • Efficiently preserved the web codes and the MySQL database based on detected dependencies, migrated the system to another platform and evaluated the ease of using the system by a non-savvy user.===========================

My CS M.S. aimed to develop Ubiquitous computing (IoT) system applications related to smartphones, machine learning, computer vision, embedded systems, and sensors. Project 1. [Smartphone System for Drivers' Safety]: (Parallel, Smartphone, Imaging, Computer Vision, Machine Learning)• Developed an Android mobile phone-based camera system that adopted computer vision with machine learning, AdaBoost classifier, on images to detect real-world road rage events around drivers.• Classified the road rage events, such as tailgating and overtaking, from images based on the detected road white lines, the boundary of cars, and the estimated distance to the cars.• Conducted parallel computation on one smartphone for event detection to achieve real-time performance, on average 175 ms per image frame, resulting in high precision and recall, 85% and 84%, respectively, in real-world experiments.Project 2. [Embedded System for Alcohol-dependant Patients] : (Embedded System, Bluetooth, IoT) • Developed software using C language to analyze sensor inputs from a Bluetooth breath analyzer prototyped by an Arduino to monitor alcohol usage for alcohol-dependent patients. • Detected human breath based on data from a pressure sensor and adopted a mathematical interpolation method to estimate alcohol consumption based on the reading from an alcohol sensor.• Implemented communication protocol between the breath analyzer and a smartphone through Bluetooth so that the alcohol data could be visualized from both the smartphone and the cloud. • Iterated on system design on a cross-functional team of hardware designers, mobile phone developers, web developers, and medical doctors.• Decreased patients' total alcohol consumption and the number of drinking days by 96.5% and 82.3%, respectively.Project 3. [Wearable Embedded System]• Developed an algorithm using C on a wearable embedded system and encoded/decoded a thermal tag to emit a unique temperature signature for user identification.