Re-architected the motion control interface for autonomous/self-driving box trucks on city roads.

• Transitioned the earlier framework to ROS2.

• Reduced steering oscillations by updating interpolation functions used for selecting weights of Linear Quadratic Regulator controller used for lateral control (Python, C++) by ~80%

• Improved longitudinal tracking by retuning position and velocity PID controllers based on data visualization (Python, MATLAB), and introducing a first-order filtered linear acceleration controller (C++). Reduced final tracking error by ~60%

• Analyzed data for a multivariable function (throttle/brake = f(vel, accel)) using various polynomial fits to inform a redesign of vehicle calibration process (MATLAB)

Motion Planning and Controls Software for Self-Driving Sedans

• Designed a geometric path stitching algorithm (Python, C++) to blend pre-mapped paths using desired parameters.

• Analyzed field data to propose algorithms for ramp/slope detection and compensation in motion controller (Python).

Mapping and Localization Software for Autonomous Driving

• Developed an embedded application for high-fidelity localization by fusing redundant sensor data using advanced sensor fusion techniques.

• Designed and implemented graph SLAM algorithms for map generation, with features like loop closure to ensure accuracy and drift correction.

• Optimized localization and mapping pipelines for resource-constrained embedded platforms, ensuring real-time performance and reliability.

• Integrated the localization system with ROS (Robot Operating System) for modular development and seamless communication between system components.

• Worked on drive-by-wire systems, enabling precise vehicle control for autonomous operation.

• Engineered real-time data acquisition and processing frameworks to handle inputs from multiple sensors, including LiDAR, IMU, GPS, and cameras.

• Enhanced map representation with graph-based optimizations for efficient querying and scalability across diverse environments.

• Developed robust testing frameworks to validate performance under various scenarios, including urban, off-road, and indoor environments.

• Deployed software on embedded platforms with tight hardware constraints, balancing computational efficiency and power consumption.

• Collaborated with cross-functional teams to align localization outputs with navigation, path planning, and perception systems for seamless autonomy.