BluHaptics proposes Deft Control Software (DCS), which utilizes machine learning to enable intuitive and efficient control of robotic arms in remote operations with underlying structure. The human-centered control methodology utilizes 3d sensor fusion for workspace visualization, machine learning with on-the-fly training, and pilot assist features to garner operator trust, improve safety, mitigate training latency, and support rapid task switching. The integrated algorithms identify and track underlying structure to enable pilot assistance and other safety features such as collision avoidance. DCS utilizes a common interface across robotic platforms and supports variable levels of autonomy specific to each task and/or operator. DCS permits robotic execution of exceedingly complex tasks that require high-levels of cognition and precise motor control which, to date, have been intractable for purely manual or automated control schemes to accomplish. The overall Phase I and II objectives are to: (1) demonstrate the value of a DCS interface to support intuitive manual control for remote operations, (2) demonstrate 3d visual-feedback and operator assistance supported by machine learning for tasks with underlying structure and varying levels of complexity, and (3) demonstrate the DCS platform can be extended to support different classes of robots with varying levels of autonomy. The objectives specific to Phase I are to: (1) Demonstrate intuitive manual control of a simulated NASA robot, (2) mitigate program risk by demonstrating basic assistive features, and (3) refine Phase II technical objectives based on collected user feedback and specific scenario requirements.