Centre for Intelligent Robotics · IIIT Allahabad

The AI & Robotics lab

AIRXLAB

Where machines learn to see, think and move. Four wings — AI, ROS, hardware and design — building real robots together, from the first sketch to the last line of code.

  • Machine Learning
  • Computer Vision
  • Reinforcement Learning
  • ROS 2
  • Embedded
  • CAD
  • Drones
Explore the wings Meet the team
Four wings Cortex · Nexus · Forge · Blueprint AI · ROS · Hardware · Design
One lab 14 people, one robot at a time Mentors · leadership · wing coordinators

Projects & showcase

See what the lab builds

Two minutes inside AIRXLAB: the robots, the drones and the people behind them, from first prototype to the demo floor.

  • 2:17Runtime
  • 4Wings at work
  • 2026Edition
AIRXLAB Showcase 2026

Four wings, one lab

Every machine needs four kinds of builder

AIRXLAB is organised into four wings. Each one owns a layer of the robot, and every project crosses all four: it is designed, built, taught to think and taught to move.

Wing 01 · Cortex

AI & Machine Learning

Perception, learning and decision making: the models that let a robot see, predict and choose.

  • Vision
  • Deep learning
  • RL

CoordinatorsVilasAkhila

Wing 02 · Nexus

ROS & Autonomy

The Robot Operating System layer: sensors, navigation, simulation and the integration that ties every part together.

  • ROS 2
  • SLAM
  • Gazebo

CoordinatorsAryanKashish

Wing 03 · Forge

Hardware

Electronics, embedded systems and fabrication: the circuits, motors and boards the software runs on.

  • Embedded
  • PCB
  • Actuation

CoordinatorsSayakDiya

Wing 04 · Blueprint

CAD, 3D Modelling & Design

Mechanical design, 3D modelling and visualisation: every part is drawn here before it is printed, cut or built.

  • CAD
  • 3D print
  • Render

CoordinatorsKhavinaRoshni

The team

The people who run the lab

  1. 01 MentorGuides the lab and its direction.
  2. 02 PhD mentorsResearch depth behind every project.
  3. 03 Lab leadershipRuns the lab day to day and ties the hardware and ROS stack together.
  4. 04 Four wingsEach wing owns one layer of the robot.

14 people · 4 wings · 1 lab

  1. Dr. Surya PrakashMentor, AIRXLAB
  2. Rohit KumarPhD Mentor
  3. Ujair AlamPhD Mentor
  4. Manoj KumarPhD Mentor
  5. SmitOverall Coordinator
  6. AdityaChief Technical Officer · Hardware & ROS Integration
  7. Wing 01 · Cortex AI & Machine Learning VilasAkhila
  8. Wing 04 · Blueprint CAD, 3D Modelling & Design KhavinaRoshni
  9. Wing 02 · Nexus ROS & Autonomy AryanKashish
  10. Wing 03 · Forge Hardware SayakDiya

The four domains

Not slides. Working models

Each panel below is running the thing its track teaches, in your browser, right now — the fit, the training loop, the policy and the kinematics are all real. This is the level the programme works at.

ML

Machine learning

Supervised and unsupervised learning, from the mathematics up. Below: a least-squares fit solved in closed form every frame, with residuals drawn.

DL

Deep learning

Neural networks, CNNs, transfer learning, vision and NLP. Below: a small network descending real cross-entropy on two interleaved arcs.

RL

Reinforcement learning

From the Bellman equation to control on real hardware in week four. Below: tabular Q-learning, with arrows showing the policy as it forms.

RO

Robotics & drones

ROS 2, simulation, navigation, SLAM and the hardware underneath. Below: a two-link arm solving its own inverse kinematics — move your pointer.

pointer

Build with us

Got an idea that should move?

Find the lab at the Centre for Intelligent Robotics, IIIT Allahabad, or write to us. Every project here crosses all four wings.

Contact the lab See the wings