I have been working on a variety of research topics, ranging from stochastic control and information (belief) space planning, to simultaneous localization and mapping for mobile robots, robot vision and satellite based navigation. You can visit my research group’s page at

[Google Scholar Profile]

Autonomous Vision-based Navigation for Micro Aerial Vehicles: From May 2015 to December 2015 I was a co-op at Qualcomm Research Center in San Diego. I worked on full onboard vision-based autonomy for aerial robots. We demoed this system at CES 2016. This system implements full waypoint navigation, builds a map of an initially unknown world and also assists the pilot in flying by not letting him/her collide into obstacles! Here is a video of a demo in the Qualcomm lab.


Motion Planning in Non-Gaussian Belief Spaces: While recent work has dealt with planning for Gaussian beliefs, for many cases, a multi-modal belief is a more accurate representation of the underlying belief. This is particularly true in environments with information symmetry that cause uncertain data associations which naturally lead to a multi-modal hypothesis on the state. Thus, a planner cannot simply base actions on the most-likely state. We propose an algorithm that uses a Receding Horizon Planning approach to plan actions that sequentially disambiguate the multi-modal belief to a uni-modal Gaussian and achieve tight localization on the true state, called a Multi-Modal Motion Planner (M3P).

Robust Real-time Replanning in Belief Space in Changing Environments:  We have devised a principled way of real-time replanning in belief space via realization of the rollout policy based on the FIRM framework. This video demonstrates the performance of the method on physical mobile robots over a long (about 25 minutes) and complex run. In this run, we investigate how such a real-time replanning method can generate a feedback plan that is robust to discrepancies between real models and computational models as well as robust to changes in the environment, failures in the sensory system, and large deviations from the nominal plan. We believe these framework lay the ground work for further advancing the theoretical POMDP framework toward practical applications, and achieving longterm autonomy in robotic systems.

Development of an Aerial Platform for Vision Based Simultaneous Localization and Mapping: Aerial vehicles need an accurate estimate of position and orientation for navigation. In a GPS denied setting, one needs to access other sensor information. A vision based real time Extended Kalman Filter based simultaneous localisation and mapping application was successfully implemented in C++ based on Andrew Davison’s MonoSlam. I successfully designed, developed and flight tested a quadrotor micro aerial vehicle equipped with onboard computer, laser range scanner and camera for mapping of unknown indoor environments. Implemented a 6 Degree of Freedom quadrotor model in MATLAB and designed a PID based controller. 

Satellite Based Precision Navigation for Commercial Aircrafts: I implemented a fully functioning model of GPS constellation in MATLAB; Simulated Dual- Frequency GPS receiver and demonstrated a positioning accuracy of 1-12 m for an aircraft in-flight. Demonstrated a new algorithm for carrier phase tracking using smoothed pseudorange measurements. Simulated a precision autolanding to determine positioning errors using carrier phase measurements; was able to meet FAA specified CAT-III accuracy requirements.

Be First to Comment

    Leave a Reply

    Your email address will not be published.