Ameya Phadke with a Unitree humanoid robot
ROBOTICS ENGINEER  /  NYU TANDON

Ameya
Phadke.

MECHATRONICS HARDWARE AUTONOMY

I design and build robots that move, sense, and think — from full-body humanoid teleoperation to autonomous navigation and custom hardware.

VIEW WORK  ↓ GET IN TOUCH

Selected Work

04 PROJECTS
▼  SELECT A PROJECT
▶ PHOTOS + VIDEO
01 — TELEOPERATION

Humanoid Teleoperation on Unitree

  • A low-latency, full-body retargeting pipeline that maps an operator's motion onto a Unitree G1 — enabling intuitive remote manipulation and locomotion.
  • Built support for Apple Vision Pro and Meta Quest 3, and developed a wireless solution that allowed for headless control.
ROS2 Motion Retargeting OpenCV
JOINT ANGLE / TORQUE — TELEOP CYCLE LIVE TRACE
SHOULDER ELBOW TORQUE
  • Sub-30 ms operator-to-robot motion latency across the retargeting loop.
  • Full-body IK retargeting mapping 20+ DoF onto the Unitree platform.
  • Vision-based operator tracking fused with joint-encoder feedback.
02 — HARDWARE

Unified Hardware Platform

  • A single modular electronics and compute stack designed to power multiple robots — custom PCBs, power distribution, and sensor breakouts engineered end-to-end.
  • Designed with modularity and simplicity in mind, this thermal-safe enclosure plugs into a robot's battery outlet and is mounted on its back. A standardized hardware stack ready-to-go!
  • Custom PCB support for 20-28V inputs and 5/10/24V outputs, with overvoltage protection, inductance shielding and discrete power distribution rails for additional safety.
Altium KiCad PCB Design
Internals of the unified hardware platform — custom PCBs and wiring PHOTOS
SYSTEM BLOCK DIAGRAM — POWER & DATA v3
BATTERY PDB COMPUTE SENSOR HUB MOTOR DRV 22.2V 5/12/24V
  • 6-layer control board integrating regulation, CAN bus, and sensor breakouts.
  • Hot-swappable 22.2V pack with per-rail protection (5V / 12V / 24V).
  • One stack reused across three distinct robot platforms.
▶ PHOTOS + VIDEO
03 — AUTONOMY

Asynchronous AMR with SLAM

An autonomous mobile robot that navigates unstructured environments using real-time SLAM and an asynchronous planning stack for robust, reactive obstacle avoidance.

ROS2 SLAM Nav / Planning
SLAM OCCUPANCY GRID — LIVE MAP ● SCANNING
OCCUPIED PATH ROBOT + LIDAR
  • 10 Hz graph-based SLAM with loop-closure over 2D LiDAR.
  • Asynchronous planner decouples localization from control at 50 Hz.
  • Under 8 cm localization drift over 40 m indoor loops.
04 — MANUFACTURING

Autonomous 3D Extruder on UR10

A custom 3D-printing extruder mounted to a UR10 arm for non-planar, large-format additive manufacturing — driven by toolpath-aware motion planning.

UR10 Additive Mfg Toolpathing
▶ VIDEO

About

I'm a robotics engineer focused on the full stack of getting a machine to act in the physical world — mechatronic design, custom hardware, and the autonomy that ties it together.

My work spans humanoid teleoperation, autonomous navigation, and robotic manufacturing, with an emphasis on systems that are robust enough to leave the lab.

EDUCATION
M.S. · NYU Tandon School of Engineering
NYU Tandon School of Engineering
SKILLS & TOOLING
Software & Autonomy
ROS2 OpenCV PyTorch TensorFlow
Hardware & PCB
Altium KiCad
CAD & Design
Onshape Catia Blender

System Capability

DOMAIN OVERVIEW
82%
AUTONOMY
70%
PERCEPTION
80%
HARDWARE
76%
CONTROLS

Proficiency

SELF-ASSESSED
HARDWARE
Rapid Prototyping90
0-to-1 Development85
CAD80
PCB Design55
Hardware & Data Integration85
SOFTWARE
ROS285
Nav280
Sensor Fusion45
OpenCV70
Robot URDF Design85
LET'S BUILD SOMETHING

Get in touch.

Emailameya5phadke@gmail.com → LinkedIn/in/ameya-5-phadke → ResumePDF →
© 2026 AMEYA PHADKE ROBOTICS · HARDWARE · AUTONOMY