Autonomous Parking: Parallel & Reverse

A ROS 2 project for autonomous parallel and reverse parking on a real car and in Gazebo simulation. The system combines SLAM-based mapping, EKF localization, custom path planners, and parking space detection to fully automate the parking maneuver.

Autonomous parking demo

My Contributions

Extended Kalman Filter (EKF) Integration: Integrated robot_localization to fuse wheel odometry with IMU data, correcting odometry drift and providing a stable map → odom → base_footprint transform chain.
SLAM + Localization Launch: Created the main launch file combining slam_toolbox and ekf_node, configuring process noise, sensor timeout, and the odometry/IMU fusion matrix.
New TAS Car Hardware Integration: Configured hardware drivers, merged dual LiDAR scans, and added URDF models and Gazebo world assets for the new car platform.
Costmap & Obstacle Avoidance Tuning: Tuned inflation radius and cost scaling factors in the Nav2 costmap configuration for reliable obstacle avoidance in tight parking spaces.
Path Following Controller Tuning: Configured RegulatedPurePursuitController parameters (lookahead distance, max angular acceleration, desired linear velocity) for smooth and precise parking execution.
Parking Manager Launch: Consolidated the planner, controller, and parking manager nodes into a single unified launch file for both simulation and hardware deployments.