Robots

Dextrous Robot Apprentice (DORA) aims to help humans do dextrous tasks, like cooking, assembly and crafts. DORA too is an apprentice and can learn quickly from demonstrations.

Home Apprentice Learner (HAL) aims to help humans automate daily tasks around the house. Since HAL is an apprentice, it can learn a new task, such as clean up after dinner, only from a few human demonstrations.

Working with an amazing team at Aurora to develop and deliver self-driving, safely, broadly and quickly.

Led a team of students to design a fleet of low-cost autonomous RC cars. The hardware and software is completely open-sourced , and intended for education and research.

Developing the motion planning stack for a mobile manipulation platform RoMan for ARL RCTA project. The robot can navigate over difficult terrain, lift weights of up to 25 kg, and clear a pile of debris autonomously.

Developed the motion planning software pipeline for a full-scale autonomous helicopter for ONR AACUS project. The platform flies at 60m/s and can sense up to 1200m. The software was evaluated on multiple platforms: Boeing’s Unmanned Little Bird and Bell 206 Jetranger.

Developed the planning software for a hexarotor (DJI-M600) platform that executes missions for over 2 hours in an obstacle course flying at speeds of upto 15 m/s.

Developed a quadcopter (DJI-M100) that navigates over diverse terrains avoiding obstacles like hills, trees, electric wires at high speeds of upto 5 m/s.

Developed a quadcopter for autonomously exploring the inside of an unmapped ship relying on only a floor plan.

Developed an octocopter that autonomously explores and maps out river banks, avoiding obstacles at speeds of upto 3 m/s

Developed sensor suite mounted on a full-scale helicopter for actively tracking a landing site on a shipdeck.