Autonomous Multi-Vehicle Cooperative System (AMCOS)
Abstract
The Virginia Tech National Security Institute and the Ted & Karyn Hume Center for National Security and Technology are proposing the development of the Autonomous Multi-vehicle Cooperative System (AMCOS) project as a research-focused program to develop algorithmsthat will enable cooperation within a heterogeneous team of agents with limited computational power and communication bandwidth with applications to Unmanned Underwater Vehicle (UUV)s. The team will develop multi-vehicle cooperation and formation movement to use in the investigation, comparison, and evaluation of the control authority improvement of distributed versus localized control as applied to the capture net.Given the fact that autonomous UUVs are very expensive and difficult to use for rapid prototyping due to thehigh resource requirements # material costs, programming, transportation, deployment, on-site staff (dive and command crew), etc. #a surrogate for UUVs will be used. For the AMCOS program students will use Lighter-Than-Air (LTA) vehicles and demonstrate the system capabilities in a semi-annual Defend the Republic (DtR) competition. These LTA vehicles have similar constraints to manportable UUVs, especially in size, power, and especially weight (payload capacity). The competitions will provide a rich environment for situational awareness, cooperative planning, targeting, and communications by deploying a #team# of autonomous vehicles against another adversarial team of autonomous agents. It also provides dramatic drive and excitement in the students to ensure active participation and creativity. The program will use simple navigation rules and range-only navigation that will allow AMCOS to perform rapid autonomous search and retrieval in a fixed area. Fundamentally, AMCOS will develop a comparison of distributed vs centralized, single-vehicle command and control.To capture a game ball, Virginia Tech (VT) will leverage centuries of human fishing knowledge to exploit a net system. Specifically, we will explore a trawling method, where a single net will be dragged behind a single vehicle or formed viaa swarm of vehicles flying in a defined formation. In order to isolate the control aspects and reduce complexity, locating and capturing of game balls by both types of vehicles will be done via search pattern alone # no object detection used. Exploration and comparison of search patterns and their ability to capture objects adrift will be investigated and the impacts and improvements between vehicle types will be characterized.This characterization will include the demonstration of devices and algorithms that use inter-vehicle, range-only localization to maintain a multi-vehicle formation for net formation and control, and characterize the capabilities of the swarm to perform complex maneuvers. The impact on search pattern efficacy and enabling of potential new, high turn rate search patterns will be demonstrated. Such patterns could dramatically improve capture in confined or congested areas.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Mar 03, 2023
- Source ID
- N000142312210
Entities
People
- Thomas Krauss
Organizations
- Office of Naval Research
- United States Navy
- Virginia Tech