aa / Let’s connect

ANUSHRI ARORA / ROBOTICS & PHYSICAL AI

Pune, India

Learning to see.
Learning to act.
Building robots.

I work where perception meets action—bringing vision-language models, reinforcement learning, and robot control into the physical world.

Explore my workRead my résumé PDF

B.Tech CS (AI/ML), VIT Pune · Class of 2028

FOCUS AREASPhysical AIRobot manipulationImitation learningReinforcement learningVision-language-action

01 / SELECTED WORK

Ideas, tested in motion.

From learning policies on robot arms
to navigating with limited compute.

01
MANIPULATION / IIT BOMBAY

Vision, language,
and two robot arms.

Deploying VLA policies on a bi-manual LeRobot SO-101 setup during my e-Yantra summer internship.

LeRobotRGB-D perceptionImitation learning
HARDWARE + LEARNING2026

THE WORK

Connected perception, data collection, policy training, and hardware deployment into a hands-on robot learning workflow.

PlatformBi-manual SO-101
PerceptionIntel RealSense D405
MODELS EXPLORED

SmolVLA · pi0.5 · OpenVLA · GR00T

Watch SO-101 demo on YouTube
Inside the project

Integrated RGB-D cameras, robotic arms, and perception pipelines. Collected robot datasets, trained VLA and imitation learning policies, and deployed them on the physical setup.

The internship included comparative analysis and deployment of multiple VLA models, connecting model experimentation to real hardware.

e-Yantra, IIT Bombay · May–July 2026

02NAVIGATION / ROBOT LEARNING

Big-picture reasoning.
Real-time control.

A latency-aware navigation system pairing high-level VLA goals with a local reinforcement learning controller on low-cost hardware.

VLA + RLRaspberry Pi 4Perception
Watch demos DriveView code GitHubRead paper DOI
Inside the project

Designed and integrated a Raspberry Pi 4 robot with ultrasonic sensors and perception pipelines for obstacle detection and environmental awareness.

The VLA pipeline generates high-level navigation goals, while a local RL agent handles control under inference latency constraints.

Smart Navigation Robot · VIT Pune

03AERIAL ROBOTICS / AUTONOMY

A drone with
a mission.

An autonomous quadcopter for greenhouse pesticide spraying, with PID flight control, plant detection, and waypoint-based missions.

ROS 2GazeboPID controlOpenCV
Open drone demo in Google Drive
Inside the project

Designed a PID-based controller for stable hover and autonomous navigation, alongside an image-processing pipeline for detecting plants from overhead camera data.

Developed mission logic for takeoff, waypoint navigation, obstacle traversal, targeted spraying, and landing. Validated the system in Gazebo and worked with field-testing environments.

Krishi Drone · e-Yantra Robotics Competition

02 / EXPERIENCE

Learning by
building.

Research, hardware, and tools that help others learn robotics.

SEPT 2026 — PRESENT

Robotics Platform Engineer Intern

Innova8r · Remote

Designing robotics programming problems and reference solutions in Python and C++, with test cases and evaluation logic that cover correctness and edge cases.

MAY — JULY 2026

Summer Intern

e-Yantra, IIT Bombay

Implemented VLA policies on bi-manual robot arms. Worked across RGB-D integration, robot datasets, imitation learning, and deployment on hardware.

03 / RESEARCH & OPEN SOURCE

Beyond the build.

PUBLICATION / ROBOT LEARNING · 2026

Q-Learning with VLA Perception for Real-World Navigation Using Low-Cost Sensors

Research on a slow-fast hybrid architecture pairing cloud-based VLA perception with local reinforcement learning for responsive navigation on low-cost robots.

Atlantis Press · ICAIML-AE 2026 · Advances in Intelligent Systems Research

Read the publication DOI
PUBLICATION / MACHINE LEARNING · 2026

Adaptive TabNet–β-VAE Fusion Ensemble Closes Uncertainty Gap in Fraud Detection

Published work exploring machine learning beyond robotics, with a focus on uncertainty in fraud detection.

Springer Cham · ICAIN 2025 · Lecture Notes in Networks and Systems

Read the publication DOI
CODE & EXPERIMENTS

Work in the open.

Explore my repositories, ongoing experiments, and open-source activity on GitHub.

github.com/aroraanushri

04 / ABOUT ME

Curious about intelligence.
Grounded in robotics.

I’m Anushri, a Computer Science student specializing in AI and ML at Vishwakarma Institute of Technology, Pune.

I’m interested in how robots connect what they see with what they do. My work spans learned policies, perception pipelines, and control—with a growing focus on robotics and Physical AI research.

EDUCATIONB.Tech · Computer Science (AI/ML)Vishwakarma Institute of Technology, Pune2024–2028
MY WORKBENCH

Build & simulate

ROS 2 / Gazebo / Isaac Sim / Raspberry Pi / PID control / RealSense

Learn & perceive

PyTorch / Hugging Face / LeRobot / OpenCV / Imitation learning / Reinforcement learning

Write & experiment

Python / C++ / C / Java / Weights & Biases / Camera calibration / Pose estimation

WHAT’S NEXT?

Let’s build something
that moves.

Interested in robot learning, research collaborations, or a robotics project? Let’s talk.

GitHubLinkedInRésumé