48 | ACCOMPLISHMENTS | X DEPARTMENT
X20
UNMANNED SYSTEMS, AUTOMATION, AND PROCESSING DIVISION
Autonomy in a Box
The Autonomy in a Box (Autobox) project is an effort to dramatically reduce the timeline required to deploy novel autonomy algorithms, frameworks, and software on unmanned systems. Currently, integration of novel autonomy algorithms, behaviors, and high-level frameworks onto embedded systems after initial development requires considerable effort due in large part to the lack of a common integration environment between systems or even between time periods in a system’s use. Shared assets, such as Nation Unmanned Systems Shared Resource Center (NUSSRC) vehicles, often require rebaselining payload computing modules (PCMs) after each use by formatting hard drives to maintain the integrity of the assets. The rebaselining, FIGURE 1: (TOP) CONVENTIONAL SOFTWARE DESIGN FOR A MVC AND PCM. hardware simulation to verify functionality, (BOTTOM) NEW APPROACH WITH AUTOBOX USING DOCKER. and integration of new autonomy algorithms to test on the system often takes days to weeks. Often, system libraries must be manually installed, a lengthy installation process must be followed for the autonomy framework, and then communications with the main vehicle computer (MVC) must be configured. Only after the aforementioned steps can integration of novel autonomy behaviors, algorithms, or arbitration frameworks occur. Autobox provides a common, hardware-agnostic software deployment solution leveraging the industry standard Docker containerization environment to dramatically reduce the burden involved in vehicle reconfiguration and rebaselining. With Docker, all software is developed and integrated into a Docker “container.” The container file can then easily be transferred from one system to another. Once an experiment is performed, the vehicle can be rebaselined simply by removing the Docker container used during the test. Autobox reduces the time required for transitioning software from development to experimentation from days to less than an hour in most situations. Other benefits of our approach include configuration management capabilities, cross-platform compatibility, and the ability to share a common environment between developers. In FY16, we successfully developed a working prototype of Autobox for generic UUV autonomy. The Autobox container, using the MOOS-IvP autonomy framework, successfully connects and gains control of a REMUS 100 vehicle and runs a baseline mission under PCM control. A graphical user interface (GUI) was developed to facilitate user deployment, extraction, and configuration of Docker containers between an external computer and an unmanned asset. Containerized versions of the Robot Operating System, Interval Programming (ROS-IvP) and Autonomous Vehicle Architecture (AVA) frameworks were also established. Autobox has been sea tested and used under several related Office of Naval Research projects in FY16, including Unmanned Warrior 2016 and the ONR Advanced Sensors Enabling Capability Program.
ALMOND-U, CWP
The Department of Defense’s Coalition Warfare Program (CWP) leverages U.S. and foreign investments to conduct cooperative research and development projects with foreign partners. The Allied Munitions Detection Underwater Initiative (ALMOND-U) is a U.S. Navy CWP project that partners with the Bundeswehr Technical Center for Ships and Naval Weapons, Maritime Technology and Research (WTD71) in Germany. The goal is to evaluate, analyze, and advance technologies for the detection, classification and mapping of bottom and buried underwater munitions. The program also supports the development of an automated unmanned underwater survey system. Bringing together advanced mine countermeasures (MCM) technologies from both countries, provides an opportunity to develop and configure innovative technological underwater munitions survey concepts and then test and develop these concepts during sea trials conducted in both Germany and the United States.