Ishir Roongta

I'm a Masters student at Carnegie Mellon University studying Robotics and AI. I have prior experience in developing software stacks for perception, control, and autonomy (FSM, Behavior Trees) for mobile robots. Previously I completed my BTech in Electrical Engineering at the Indian Institute of Technology Kanpur. Reently I passed the FAA part 107 examination for small UAS pilots.

Email  /  CV  /  LinkedIn  /  Github  /  Scholar

Hire Me!

I'm currently looking for full time roles (starting summer 2025) as a Robotics Engineer/Software Engineer/Deployment Engineer/... any role that involves building software stack for autonomous robots, integrating with hardware, iterative testing and deployment in the wild. Why Hire Me?

Work Experience

I've been fortunate enough to work with some amazing people during my internships. I'm a generalist in my skillset and have demonstrated my breadth at the following places:

Reverse Engineered BlueTooth protocol of installed BMS to access battery status and radio mesh network of multiple UGVs using a module written in C++ based framework.
Implemented custom ROS2 wrapper for in-house camera module and integrating a 2D Lidar to create a global mapping pipeline using SlamTollbox in ROS2.

I was selected in the MITACS Globalink Research program as a research intern to work on the problem of indoor localization of a pair of drones using limited UWB ranging data.

Projects & Competitions

A list of key projects which helped me build foundational knowledge about robotic systems from simulating physcial attributes in Gazebo to deploying a full autonomy stack on a custom built aerial platform. I have worked on problem statements like autonomous obstacle avoidance, path planning, pure pursuit control, and exploration of unknown areas.

An aerial platform capable of navigating in complete GPS denied environment (simulating sub-canopy conditions of a forest), obstacle avoidance (simulating avoidance of tree trunks), and autonomous exploration of an area to map out fire hotspots and relay video feed in realtime.

Implementing and experimenting with autonomous driving algorithms involving reactive planners like gap following, graph-based planners like different flavours of RRT, predictive planners like Model Predictive Control (MPC), and learning-based end-to-end aproaches to generate control from LiDAR data. Task is simple: Go around a track as fast as possible!

I worked on generating waypoints for the FRANKA arm by sampling appropraite points on the segmented image taken from a RS D435i camera. Post preliminary results, I implemented the Travelling Salesman Problem on sampled waypoints to generate efficient and smooth trajectory.

More of an exploratory project, we attempted to generate more appropriate grasp points/locations/zones for on object by queueing a VQA to describe the scene, an LLM to understand context of the request and the object, and VLM to suggest the contextually appropraite grasp location.

Problem Statement: Map out rough waypoints of a mountainous road in summer conditions using a UAV, and then navigate a UGV with the help of an overhead UAV in winter conditions i.e road and mountain covered in snow.

Problem Statement: Navigate an unkown maze using a RGBD camera attached in front of the drone. At the end, land at a specific point using a Aruco marker.

Problem Statement: An aerial platform capable of long distance transport of heavy equipment. The drone also performs removal and replacement of tightly-integrated heavy objects on a moving platform (moving ship).

Publications

We propose a novel control law using Control Barrier Functions to force a unicycle modeled particle to convergence around a target while imposing restrictions to it's orbit. We presented the work at the AIAA SciTech conference in January 2024 at Orlando, Florida.

Resources

This will be a collection of resources for those after me. I hope you find some use in these, and If you want to correct/add/subtract something please hmu via email.

• Grad School Applications - Personal Pieces of Advice - here
• GRE Prep YouTube Channel I recommend - GregMat

Some Hobbies

Well if you read this far, here's a little non-work related stuff about myself. I love landscape photography and you can find my stuff here: Instagram and YouTube My current gear is a beat-up Fuji XT200 with kit lens. I ocassionaly play the guitar as well.

Last updated 18th January 2025. This template is taken from John Barron's website.