Robotics Engineer
Autonomous systems · Perception · Simulation · Industrial robotics
Hi, I'm Nixon. I build robotics software for navigation, perception, simulation, and industrial automation. My work spans ROS/ROS 2, LiDAR calibration, computer vision, path planning, and robot simulation.
I enjoy the parts of robotics where software meets imperfect sensors and real system constraints: transforming point clouds, debugging robot behavior, and making autonomy easier to inspect.
- Autonomous navigation: global and local path planning, frontier exploration, Nav2 goal handling, and recovery behavior.
- LiDAR and point clouds: multi-sensor calibration with NDT and GICP, frame transforms, and aligned point-cloud pipelines.
- Perception: camera detection, ground-plane projection, semantic landmark tracking, and language-guided navigation.
- Simulation and tools: URDF robots, Gazebo/RViz environments, ROS GUIs, system monitoring, and repeatable deployment.
Core
Robotics and perception
Simulation and interfaces
Engineering tools
Problem: A factory robot needs to carry a part between stations while proving navigation, station identity, and PLC transfer completion.
My work: Built the ROS 2 coordination state machine, camera marker gates, MQTT replay-safe mission interface, Modbus handshakes, and correlated protocol traces with deterministic failure tests.
Stack: ROS 2 Humble, C++, Python, Nav2, Gazebo, RViz, MQTT, DDS, Modbus TCP, OpenCV, Docker
Result: A locally verified simulation mission completed in 72.6 simulated seconds. All 12/12 deterministic scenarios matched: four real Gazebo cases, seven protocol cases with a navigation driver, and one DDS experiment. Version 1 supports one part and one fixed route.
Problem: A mobile robot needs to explore an unknown environment, build a persistent semantic map, and navigate to an object named by a person.
My work: Built ROS 2 packages for frontier detection and selection, YOLO object detection, ground-plane projection, semantic landmark tracking, and language-command handling through Nav2.
Stack: ROS 2, Python, Nav2, YOLO, OpenCV, Gazebo, RViz
Result: A simulation workspace that explores frontiers, tracks supported objects, and converts commands such as go to the chair into navigation goals.
Problem: Comparing global and local planning behavior is difficult when the map and algorithm state are hidden.
My work: Implemented grid-based A* and a teaching-oriented Dynamic Window Approach, then connected both planners to an interactive map interface.
Stack: Python, PyQt5, NumPy, A*, DWA
Result: A 50 × 50 visualizer that animates paths and replans when the user adds obstacles.
Robotics Software Engineer Intern · AIDRIVERS
- Developed ROS tools for multi-LiDAR calibration with NDT and GICP, point-cloud alignment, and sensor configuration.
- Connected industrial PLC hardware to ROS through Modbus and built Qt tools for configuration and status monitoring.
Robotics Simulation Developer · Foodlink-System
- Built a URDF-based restaurant robot simulation with differential drive, LiDAR, IMU, depth cameras, and realistic operating scenarios.
- Developed modern C++ and Qt tooling, packaged the system with Docker, and tested robot behavior around docking, low battery, and crowded environments.
Professional projects described here are experience summaries; their source code is private.
My public work currently centers on robotics, path planning, computer vision, and machine-learning experiments.
|
|
|
I am currently focused on making robot navigation and perception systems easier to test, inspect, and improve.