Currently, the construction of retail inventory distribution specifications (or "store planograms") is chiefly a manual, tedious and unreliable process, which may extend further than the period of time for which the planogram is actually valid. Moreover, there exist dire limitations in the flexibility, convenience and excitement experienced by customers within various retail environments, which could be otherwise instantiated through robotics and augmented reality.
The goal of this project is to implement and optimize an autonomous robotic system as a solution to these concerns -- by extending the Robotic Operatic System (ROS) Linux implementation and various open-source perception libraries -- with the following specific requirements: recognition/perception of environmental objects, dependable visual simultaneous localization & mapping (VSLAM), real-time data uplink communication of environmental parameters, distributed vision-based object training, and peer node collaboration.
My roles on this project will be: researcher & system architect, roboticist / embedded engineer, and fabrication / system-integration engineer
Affiliations: Intel Science & Technology Center (ISTC), Intel Corporation, CyLab, Department of Electrical & Computer Engineering at Carnegie Mellon University, Robotics Institute at Carnegie Mellon University