TL;DR

This paper compresses my robotics coursework and public repos into one technical collection: visual odometry with camera and IMU, FC-RRT* path planning for UAVs, human obstacle detection and tracking with YOLO/Kalman-style perception, and supporting classical planning/perception exercises. The page exists to strengthen the robotics story without turning the homepage into a repo dump.

A robot is not one algorithm. It is perception, planning, estimation, and the humility to know which layer is uncertain.

System note, Robotics collection, 2026

Problem

Robotics systems need perception, state estimation, planning, tracking, and operator-facing explanation. Many small projects are weak as standalone project pages but strong as a single technical arc.

Visual Odometry

The Real-Time Visual Odometry repo combines camera frames and IMU data to estimate 3D pose using Python, NumPy, OpenCV, Matplotlib, SciPy, and FilterPy. It records IMU and pose output for later refinement.

Path Planning

The FC-RRT* repo extends RRT* with flight-cost-aware sampling for UAV path planning in complex environments. It uses Python with Pygame, NumPy, Open3D, Trimesh, and Matplotlib.

Interactive figure

Path planning playground

Pick an algorithm, set rough terrain density, and click grid cells to toggle obstacles.

Explored: 0 cells
Path cost: 20

Blue cell = start, orange = goal/path, dark = obstacle, striped = rough terrain, gradient = explored accumulated cost.

Grid search with weighted rough cells. BFS expands by step count and ignores weight; Dijkstra expands by accumulated cost and routes around rough cells; A* adds a Manhattan heuristic and funnels toward the goal. Color gradient on visited cells shows accumulated cost.

Human Tracking

The Acme Robotics perception module focuses on human obstacle detection and tracking with YOLO, monocular camera input, OpenCV, Eigen, C++14, CMake, Google Test, static analysis, and documentation artifacts.

Limitations

This collection is intentionally not framed as production robotics infrastructure. It demonstrates foundations that support the stronger systems work: sensing, planning, testing, and explainable pipeline design.

Sources