Robot learning · Dexterous manipulation · Visual RL · Sim-to-real

Arhan Chavare

I’m an Electronics and Telecommunication Engineering student at VJTI Mumbai, with a minor in Data Science. I work on learning-based robotics, with a focus on dexterous manipulation, visual policies, and deploying ideas on real hardware.

Research Intern · Robotics Research Centre, IIIT Hyderabad

About

I’m interested in building robots that can learn robust skills from visual input and interact with the physical world. My recent work includes dexterous grasping, camera-conditioned manipulation, and reinforcement learning for robotic systems.

EducationVJTI MumbaiB.Tech, Electronics and Telecommunication · Data Science minor · CGPA 8.39/10
ResearchRRC, IIIT HyderabadRobotics Research Centre
InterestsRobot learningDexterous manipulation · Visual policies · Sim-to-real

Current work

Code

Long-horizon manipulation on the SO-101 with Q-chunking

Joint work with Prajwal Avhad

I’m extending Squint, a fast visual RL method for sim-to-real on the SO-101 arm, to multi-stage tasks like stacking, unstacking and packing several cubes, which plain RL from scratch doesn’t solve.

Unstack3 policy in simulation
Cube picking with QC-FQL

Read the full write-up →

23%Place33 cubes into a bin
13%TrayPack33 cubes into assigned tray slots
11%Tower3Size-ordered 3-cube tower
Q-ChunkingFlow MatchingOffline-to-Online RLSO-101

Selected work

All projects

Experience

Full experience

Research Intern

Robotics Research Centre, IIIT Hyderabad · Dexterous grasping and robot learning

2026–Present

Core Member

Society of Robotics and Automation, VJTI · Robotics workshops and project mentoring

2025–Present

Get in touch

I’m happy to connect about robotics, research, or collaboration.

arhanjain1606@gmail.com