Adith Sundram

Autonomous Factories

• Leveraging pre-trained transformers for End to End control in single agent racing • Game theory inspired RL approaches for partially observable multi-agent adversarial racing • Online adaptation of dynamics parameters for high performance racing

• Lead development, testing, and tuning of primary Model Predictive Controller for L4 Autonomous IndyLights racecar at 160+ mph, from simulation to full scale hardware • Integration of controller into real time C++ and ROS2 software stack with Casadi • Development on offline nonlinear trajectory planner for generating optimal racing lines • Directing on-site testing and development, leading sub-team of 4 Masters and PHD students • 1st Place - Passing Competition Indy Autonomous Challenge @ Las Vegas Motor Speedway, CES 2025

Software Development for Autonomous Vehicles • Developed a dead reckoning solution for temporary GPS dropout using a Kalman Filter • Designed architecture for simulating and testing against adversarial agents

• Nuero-symbolic Intelligence for hyper-manueverable UAVs for DARPA Interpretable AI program • Built simulation/testing infrastructure with PX4 Gazebo SITL, Mujoco, and Microsoft AirSim. • Implemented baseline motion planning and control algorithms in ROS2 software stack. • Developed Gymnasium interfaces for training RL approaches to UAV motion planning.

Developing swarm of hybrid robots with jumping as form of locomotion to explore sparsely mapped extraterrestrial surfaces, and collect soil samples • Co-lead of team with 20+ students, to submit design proposal for NASA Rasc-Al Challenge

Designing a capsule with a parafoil to be dropped from 100,000 feet and autonomously descend to target location for NASA Floating Dragon Challenge. -Custom milled and machined parts, incl. motorized multi stage deployment subsystem and steering. -Design of Fuselage based on structural and thermal constraints

Design and Integration team -> developing a 21U Satellite Analysis driven design on core structural components and integration of hardware onto core satellite BUS - Designed parts for manufacturing and assembly -> tolerance stackup analysis and machinability - Modified core structural components to mount satellite hardware (boards, reaction wheels, etc) -> consideration of overall spacing, mass, and usage requirements - Selection of fasteners (screws, helicoils, etc) for all hardware mounting and structural interfaces -> revision of interface design per machine design principles and analysis (screw engagement, tap clearance, etc)

Designed and built components / fixtures that were part of development and production of medical devices. Extensively used CAD (Solidworks and PDM), and 3D printers and mechanical tools to turn CAD designs into functional hardware.

Student led team — designed, built, and programmed partially autonomous robots utilizing Android Studio, Fusion 360, power tools, and custom components ( 3D printed). Developed robots with omni and mecanum drivetrains, linear slide end effectors, and sensors including IMU, camera, and ultrasonic. Worked with image recognition libraries, including Vuforia to implement localization and self-navigation. Ranked top 10 in New Jersey; won Control Award for intelligent software.

Worked with Graduate students on projects researching different materials’ piezoelectric properties and electric circuits involving piezoelectric crystals. Worked with soft robotics materials, developed dynamic prototype for a search and rescue robot made of paper and other soft materials, including a H-Bridge drivetrain, temperature, humidity, light sensors, and conductive paint.