Open Source

  • Designed an algorithm for determining the indoor position of an autonomous vehicle by populating the open-source ORB-SLAM with predefined Landmarks to get real world positioning using CUDA platform.
  • Ran RTAB Map using two D435 camera's with limited field of view and used the map for inventory control in the warehouse.
  • Designed and constructed a faster mode of Adaptive Monte Carlo Localization that can run the algorithm on rosbag data and reproduce results for the purpose of localization improvements, automation testing etc.
  • Integrated Cartographer in the codebase to use landmarks for auto stitching occupancy maps of the warehouse.

System Design

  • Designed a localization system that provided a modular codebase to have one source of truth about robot’s state and added flexibility to support multiple devices and configurations.
  • Developed a feature that adds ability for robot to switch floors in a multi level warehouse.
  • Handled responsibilities for design and development of robot’s behavioral framework for navigation particularly to integrate a new planner, move base flex, add faults, charging behavior etc.
  • Developed a feature for vacuum robot to navigate to a targeted zone in the map for cleaning within the zone boundaries.

Research and Development

  • Designed a feature for robot to retain localization after reboot including recovery behavior using Iterative Closest Point algorithm.
  • Designed an algorithm for creating an indoor floor map using PiCamera.