Robotics Systems I Built
Custom path planners, full autonomous navigation stacks, and competition-grade mechanical hardware. Every system deployed to real hardware and tested under competition conditions.
Path Planner for Differential Drive Robots
A from-scratch Hybrid A* path planning algorithm implemented as a drop-in Nav2 global planner plugin in C++. Generates smooth, curvature-continuous trajectories using Dubins primitives with non-holonomic curvature bounds, adaptive lookahead, and dynamic velocity profiling.
Standard A* violates differential drive kinematics — zig-zag paths that require stop-and-turn mechanics.
Smooth, dynamically feasible trajectories in highly cluttered environments without stopping.
Path Planner — 3-Wheel Omnidirectional Swerve
Extension to the kinematically richer 3-wheel omnidirectional swerve configuration. Custom kinematic modelling of 3WD geometry, holonomic velocity decomposition into per-wheel speed and heading commands, and fully decoupled heading control.
Differential drive cannot translate independently of heading — limiting competition task precision.
Robot translates freely in any direction while reorienting to face targets simultaneously.
Robocon Autonomous Navigation Stack
Complete ROS2 autonomous stack built for competition. SLAM Toolbox builds a live occupancy grid. AMCL provides probabilistic Monte Carlo localisation. EKF from robot_localization fuses wheel odometry and IMU. Nav2 BehaviorTree handles high-level mission sequencing.
Competition robots need reliable autonomous navigation on unknown fields under real-time pressure.
Tuned, tested, and deployed on hardware. Contributed to back-to-back national podium finishes.
Competition Robot Mechanical Design
End-to-end mechanical design of Robocon competition robots. Full parametric CAD assemblies — chassis layout, custom drivetrain geometry (differential and swerve), bearing and shaft sizing, fastener selection, and manufacturing tolerances built in from the start.
Footprint limits, weight budgets, and field interaction requirements with competition-grade tolerances.
Fabricated directly and fielded. 2× national podium with these designs on the floor.
Self-Balancing Inverted Pendulum Robot
Two-wheeled self-balancing robot engineered for the Robofest 4.0 competition. Implements cascaded PID control with sensor fusion (IMU + wheel encoders) to maintain continuous inverted pendulum balance under disturbances.
Classic inverted pendulum — millisecond response to gravitational pull and external disturbances.
Passed both design report and proof-of-concept stages. Stable self-balancing under competition conditions.