Joshua Love Email and Phone Number

I help make Anduril's robots work even better... ;)

Black Arrow attempted to create a new approach to automatically post-process and extract labels from the real images/point clouds produced by safety critical robots that move around their environments. This labeled data would have helped robots be better trained to better recognize their human and mechanical neighbors. This would have improved the safety of everyone involved while reducing the costs per label.We looked at generating labeled sensor data for multiple heavy industries including: mining, construction, distribution yards, ground support equipment at airport terminals, ground operations at airport terminals, eVTOLs, and National Defense.In the end we faced a chicken-and-egg problem. We needed to raise funding from a Pre-Seed Round to get the hardware needed to build a hardware MVP, but we needed that hardware MVP to close Letters-of-Intent from enterprise customers, those Letters-of-Intent were necessary to get VCs to invest in our Pre-Seed Round.As a Co-Founder of a startup I performed every role ranging from janitor, to software engineer, to financial planner, to CEO.

Worked as the technical lead for Control, Planning, and Localization on Boxbot’s self-driving car. The entire ’Agile’ self-driving software team never exceeded 5 and typically averaged 3 employees. Primary responsibility was to adapt Baidu’s Apollo open-source project for self-driving cars. Adapted from the Apollo standard Lincoln MKZ base platform to Boxbot’s Polaris GEM eL XD base platform. Created cloud-based tools using Google BigQuery, Pandas, Python, and Jupiter Notebooks to perform the System Identification necessary to fit the lumped-parameter dynamics models used for Control and planning. Tuned the Control and Planning modules for maximal driving smoothness while ensuring stability. Corrected multiple C++ bugs in Apollo’s Control and Planning logic to improve the robustness and smoothness of trajectory tracking. Calibrated the self-driving sensors used for localization and object detection and tuned the Localization module for Boxbot’s specific sensor stack and available computing resources. As employee #7 at a early-stage startup, also did anything and everything from shooting videos to cleaning the office/warehouse to manually delivering packages when hourly drivers were sick.

Worked to establish a new Basic Research enterprise within Northrop Grumman Aerospace Systems. Developed research roadmaps, set up infrastructure, set up work processes and procedures, developed proposals for external funding, established partnerships with academia (U. Washington, MIT, UCLA), acted as an external spokesman, mentored interns, directed junior researchers, and executed research in the area of Cognitive Autonomy. Researched state-of-the-art algorithms for Geometric Computer Vision: Visual Odometry, Mapping, Localization, Simultaneous Localization And Mapping (SLAM), and Structure From Motion (SFM). Developed a multiple UAV/UGV swarm prototype, leveraging ROS1/ROS2/Gazebo/OpenCV/PCL and the NVIDIA TK1 and Pixhawk, which was utilized by Northrop Grumman Mission Systems for DARPA OFFSET. Additionally, applied Geometric Computer Vision and Multi-Target Tracking algorithms for high-consequence space applications.

Worked on a multi-institution research team developing a 29 DOF energy-efficient bipedal humanoid for responding to Fukushima-like events for DARPA’s Maximum Mobility and Manipulation program (M3). Created and integrated Matlab/SolidWorks/ROS/Gazebo models of Sandia’s humanoid design to enable rapid virtual prototyping. Shared, coordinated, and tested models, controllers, and hardware with our collaborators at MIT, IHMC, and OSRF. The final WANDERER humanoid walked for 4 hours and 2.8 kilometers. Additionally, created and evaluated alternatives for Multi-Target Tracking algorithms for utilizing space-based gigapixel cameras to monitor cars operating in urban environments. Developed concepts, proposals, and presentations for the newly formed Physical Security Center of Excellence while interacting with multiple levels of Sandia management.

Worked on a multi-disciplinary research team developing the Sandia Architecture for Heterogeneous Unmanned System Control (SAHUC). Developed and integrated multiple levels of control with COTS hardware and open source software (ROS) to allow one operator to control many unmanned assets in an intuitive fashion. Work resulted in U.S. Patent 10,168,674 “System and method for operator control of heterogeneous unmanned system teams“. Also worked on the 3D World Building System (SLAM) for rapidly exploring and mapping a disaster site with a manually driven UGV, which leveraged ROS & the UNITY game engine.

Worked for the Center for Collaborative Control of Unmanned Vehicles (C3UV) on a multi-disciplinary, multi-focused, research team developing tools for the collaborative control of unmanned air vehicles (we flew several MLB Bat IV UAVs with 12 foot wingspans and 50 lbs of payload). Helped to create and develop algorithms, controllers, graphical user interfaces, web services, video streaming solutions, and system architectures to provide autonomous surveillance solutions to the Navy SEALs (through the Office of Naval Research).

Student Intern at the Controls Conveyors Robotics and Welding (CCRW) group. Performed validation tests to confirm supplied robot performance. Developed and documented robotic signal specifications.

Student intern at the General Motors Systems (GMS) group. Performed job function analysis and applied lean manufacturing techniques to improve efficiency and reduce waste.

Structural Validation testing. Constructed testing apparatuses and developed testing procedures as well as performed tests on products including: mobile home floors, trailer hitches, Space Shuttle Ceramic Tiles, termite testing for decking, fences, 'tornado proof' rooms, and more.