Steven Gray Email & Phone Number

San Mateo, Ca, Us

San Francisco, California, Us

Niskayuna, New York, Us

• Planning lead - ARL Scalable, Adaptive, and Resilient Autonomy (SARA). Off-road autonomous navigation program, incorporating uncertainty in terrain segmentation and classification into planning• Planning lead - Telerobotics program. Created robot-side software infrastructure for high-latency teleoperation using VR. Responsible for planning, predictive display of robot actions.• Planning lead - UAV Trajectory Planning for Asset Inspection (led to Avitas Systems, a GE Venture). Enabled remote inspection of outdoor industrial assets as planning lead. Created framework to transform remote inspector designated points on asset to required vehicle locations and generate optimal visiting trajectory.• Software Lead - GE Power In-Situ Turbine Blade Blending. Enabled semi-autonomous operation of prototype multi-axis blending robot by creating robot model, inverse kinematics, trajectory-following controller and UI

Team lead for Team Walk Softly - a top 6 finisher in the qualifying round of the 2nd NASA Space Robotics Challenge, in which teams of robots mine for volatiles in a simulated lunar landscape. (https://www.nasa.gov/directorates/spacetech/centennial_challenges/space_robotics/22-teams-crack-code-qualify-for-final-stage-of-challenge.html). Exact rankings not given - but placed in the highest scoring group. I completed the qualifying round solo, but will be working again with Shiraj Sen for the competition round. Maximizing effectiveness by using as many open source components as possible: OrbSLAM2, modified ROS control 4WD controller, OpenCV image classification. Gazebo-based simulation, using ROS Noetic for all code infrastructure.

Team lead for Team Walk Softly, second place winner in NASA Space Robotics Challenge ( https://www.nasa.gov/directorates/spacetech/centennial_challenges/space_robotics/nasa-awards-citizen-inventors-top-prizes-in-space-robotics-challenge). The competition had over 400 registered participants, including multiple university teams, while we were a team of two working on this in addition to our day jobs.Created ROS-based software stack for humanoid robot teleoperation in high-latency, low bandwidth situations. Responsible for planning (whole-body IK using IK solver with constraint framework from Drake, motion planning, and footstep planning), user interface, and perception. Using experience teleoperating the Atlas robot, designed system to make efficient use of the human in the loop (e.g. perception detects valve, then presents interactive marker of the valve to operator in the estimated pose -- so the human can quickly refine)

Bethesda, Md, Us

DARPA Robotics Challenge (DRC) - Track B, LM ATL/UPenn/RPI Team (TROOPER)(Motion Planning and Control Design Lead)- Created whole-body motion planning implementation (based on Task Space Regions) in our ROS stack used in the DRC Trials and Finals, including whole-body optimization-based IK with constraints- Created behaviors and corresponding UI elements for multi-level controller framework- Hundreds of hours robot teleoperation experience, finding and improving pain points- Small team (~5 developers at Lockheed), so dealt with whole stack including behaviors, UI, and perception

Philadelphia, Pa, Us

DARPA Robotics Challenge (DRC) - Track B, LM ATL/UPenn/RPI Team (TROOPER)(Motion Planning and Control Design)- Created quasi-static balancing and walking controllers used in the Virtual Robotics Challenge (Gazebo simulated robots, ROS infrastructure)- Created behaviors for multi-level controller frameworkDoor opening using the Willow Garage PR2(Motion Planning and Control Design)- Wrote planner which creates and searches graph of motion primitives for motion of mobile base and end-effector contacts- Incorporated switching points of contact with the door and bracing to enable passing through spring-loaded doors- Implemented in ROS, planner class built atop Anytime Repairing A* search in the Search-Based Planning Library (SBPL)DARPA Autonomous Robotic Manipulation - Software Track (ARM-S)(Motion Planning and Control)- Responsible for planning framework, including collision detection- Integrated with end-to-end manipulation framework in ROS and run on ARM-S robot platformDARPA Programmable Matter(Simulation and Design)- Goal: design self-actuated reconfigurable robots resembling origami- Created robotic origami design tool and simulator in C++ using nVidia PhysX and Qt

Doctor Of Philosophy (Phd), Mechanical Engineering (Robotics)

Bachelor Of Science (Bs), Mechanical Engineering