Tuesday, July 7, 2015

2015-07-06: Autonomy Incubator Presents in Special Session atAIAAAviation 2015 Conference

Jim Neilan presents his paper on autonomous decision making

As those who follow the Autonomy Incubator (AI) Twitter account know, the entire permanent AI team presented at the AIAA's Aviation 2015 conference in Dallas, Texas.  Presenting to such a prestigious group of colleagues is, in itself, an honor, but the AIAA devoted an entire day's worth of programming to the Autonomy Incubator and its work.  In total, the PIs gave eight presentations in two sessions, spanning over three hours of solid talking time. The AI Head, Danette Allen, served as the Session Chair for these special sessions on autonomy showcasing the ongoing research in the AI portfolio.

The all-AI schedule

The multifarious papers from the Autonomy Incubator shared an overarching theme of Star Trek references; every single one had a quote from somewhere in the franchise as an epigraph. The paper that Danette presented about the Autonomy Incubator's goals for autonomous flight was titled "Who's Got the Bridge? — Towards Safe, Robust Autonomous Operations" and included an opening quote from Star Trek: The Next Generation. In the conclusion of the paper, she further alluds to Star Trek in a metaphor for autonomous flight. "Therefore, while the 'Captain,' who is responsible for defining the mission, may have stepped out (or is so remote that s/he cannot tactically command), an autonomous system must have the authority to redirect itself within some set of overarching goals," she wrote.

Danette introduces the session themes themes: autonomy and Star Trek

Danette was followed by Jim Neilan, who presented a paper on the AI's AEON network on behalf of Charles Cross. "Towards an Open, Distributed Software Architecture for UxS Operations" describes the AI's Autonomous Entity Operations Network and outlines its two purposes: to develop an "extensible, cross-platform onboard software system" and to provide a "test-bed for integration of of the proposed heterogeneous fleet [of ground, air and marine vehicles]." He also highlighted AEON's use of DDS, a DoD standard for middleware software.

Loc Tran gave an overview of his tree-dodging work in a paper titled "Reinforcement Learning with Autonomous Small Unmanned AerialVehicles in Cluttered Environments."  In it, he explains both the practical applications of small unmanned aerial vehicles (sUAVs) in research and commercial settings and the technical particulars of his sensor hardware, as well as his methods for training the machine learning algorithm through correction and example.

Loc brings tree-dodging to the masses

Rounding out the morning session,AI member Mark Motter presented his progress on the application of machine learning in critical phases of flight. His talk was entitled,"His actions indicate two dimensional thinking” - Deciding to Go Around via Machine Learning" and he spoke about UAV flight safety during landing.

Jim presented again to open up the afternoon session, this time with his own research.  The paper, "Using Multimodal Input for Autonomous Decision Making for Unmanned Systems," proposes a way of using the AEON framework to solve the "difficult systems integration problem" of creating autonomous hardware that can make decisions reliably in the face of uncertainty. Jim's goal is to bolster robustness in autonomous machines with object classification to keep them agile and able to adapt to unforeseen circumstances.

Anna Trujillo, the AI's human factors researcher, presented her paper, "Collaborating with Autonomous Agents," which proposes ways in which a natural language interface can be used to allow human operators of fleets of small unmanned aerial vehicles interact with their machines on a high level, verbally describing the mission to the sUAVs and then letting them figure out the details of how to execute it.  Possible real-world applications of such fleets include search and rescue missions, data sampling, and package delivery.

Anna shocks and amazes with her human factors research

Paul Rothhaar spoke next, introducing his paper "A Flexible Flight Control System for Rapid GNC and Distributed Control Deployment" with a brilliantly chosen quote by McCoy: "Did you see the love lights in Spock's eyes? The right computer finally came along." Paul focused on a "subset of the AEON system" called AEON-FCS, for Flight Control System. His research brings avionics into the Incubator through AEON-FCS, which facilitates the physical flight tests the AI does and advances the difficulty of the missions we can attempt.

Paul shows off some of the AI's hardware

Garry Qualls closed out the AI portion of the afternoon session with his research, presenting a paper entitled "Operating in 'Strange New Worlds' and Measuring Success– Test and Evaluation in Complex Environments." Garry spoke about the current and future environments for test flights made possible by AEON and LaRC facilities, which range from test flights inside the building to outdoor caged flights to tethered flights in the open air.

1 comment: