Foray: the Autonomous Agile Airship Armada
Abstract
1 OverviewFor long duration aerial missions, we turn to lighter-than-air vehicles (LTAVs): by using buoyancy forces to counteract gravity, we obviate the need for constant thrust power, and can thus maintain flight nearly indefinitely. Without active thrust, LTAVs necessarily have much stricter weight constraints, with their dimension prescribing a maximum lifting volume and therefore allowable weight of sensing, actuation, communications, and processing hardware for the entire mission, along with their power system and structural scaffolding. The design question then becomes highly integrated: no longer can each subsystem be designed independently and assembled into the final robot; the cross-subsystem impacts on weight, power, and configuration form critical constraints on design decisions.To regain exibility in system design, we can move to a multi-robot swarm, building an armada ofheterogeneous LTAVs each optimized for specific subtasks in the mission specification. This creates opportunities to enhance overall performance and robustness, while at the same time introducing other design challenges such as communications and coordination as operators cannot anymore directly drive the entire collection of robot actuators. Thus, control over this armada must incorporate autonomy, particularly exploiting the benefits while addressing the challenges of a distributed multi-agent system.Within this autonomous armada, the individual agents must be optimized for their mission tasks, which in this case are modeled by the 99++ Luftballoons World Cup, an aerial soccer game. The agents must find and direct balls through an opponent s goals, while guarding their own goals against opponent robots trying to do the same. A winning strategy therefore calls for agile robots that can out-maneuver the adversarial robots both on defense and offense, reacting faster to the changing game state to better block and score. In this proposal, we introduce Foray: the Autonomous Agile Airship Armada, that addresses these mission goals through the integrated co-design across robot subsystems. In designing our multi-agent LTAV system, we will implement sensing and control algorithms that effectively use the constrained resources across our team to accomplish autonomous behavior robust to the adversarial conditions of the game. Each individual agent will be co-designed across mechanical and electrical systems to support the necessary autonomous algorithms while realizing highly agile dynamics with minimal actuation hardware and power.2 Project Description2.1 Agile vehiclesAt the core of the Foray swarm are a number of potentially heterogeneous LTA robots. The success of the swarm in the game starts with the dynamic capabilities of each individual robot. With game subtasks involving blocking opponent scorers, scoring around opponent blockers, or beating opponent seekers to targets, the key design goal will be to create an agile LTAV that can outmaneuver other designs. To do so, we will explore two new control2.2 Distributed autonomyWith the Foray team comprising a multi-agent swarm of LTAVs across a large arena, we must design on-board autonomy to enact the behaviors necessary to accomplish the game. In this project we will use pre-programmed high level behaviors and focus our efforts on the low level behaviors supporting the autonomous tasks. In particular, we consider the problem of distributed state estimation using sensing and controls. The key design principle for the Foray autonomy becomes minimality through efficiency. We look to extract maximal value from each sensor reading, data transfer, and control action.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- May 08, 2020
- Source ID
- N000142012508
Entities
People
- Ankur Mehta
Organizations
- Office of Naval Research
- United States Navy
- University of California, Los Angeles