|
Wednesday
Using Augmented Reality to Improve the Interface Between Humans and Unmanned Vehicles
Yohan Baillot, ITT
In the upcoming years the military will increasingly use unmanned vehicles (UV) during operations. This position paper describes how the current mode of operation of these UVs could be improved to serve better the personnel on the battlefield by using Augmented Reality (AR) as this technology becomes mature. On one hand, we propose to use AR to display the aggregated sensor data collected by UVs directly on the environment they spatially relate to and provide personnel on the battlefield with the data they need where they need it. The location and identification of the UVs themselves would be the first data that would be useful to display on the environment so that controlling personnel can known where they are related to what happens on the battlefield. On the other hand, the use of AR on the battlefield can improve the control of the UVs. Indeed, target locations gathered by intelligence personnel and displayed graphically through AR can then be used by UV controllers as control interface to direct a UV to go to corresponding location significantly improving the navigation of the UV over the current procedure involving controlling personnel driving the UV using a first-person video and a joystick.
Sensors for Unmanned Systems
Jeff Callen, Penn State University Electro-Optics Center
Autonomous off-road mobility remains a difficult automation challenge. Sensors are required that permit the unmanned system to build a reliable view of its environment in a timely manner. To date, there has been little available in sensors purpose-built for the challenges of autonomous navigation for ground robots. Robot developers must either use commercially available sensors that are often inadequate in environmental robustness or form factor (size, weight and power), or acquire sensors developed for other military programs that are extremely expensive.
The Pennsylvania State University Electro-Optics Center (EOC) is working to make more sensors suitable for robotic use. The EOC is a government sponsored national center for electro-optics development, which provides an independent entity for integration and evaluation of sensors. The EOC has on-going projects internally and through subcontractors to develop and enhance Ladar and mm wave radar systems, evaluate Flash Ladar and FLIR cameras, incorporate sensor health monitoring, and develop designs for flexible robotic sensor suites. The EOC also has projects pending on small Ladar, position estimation in GPS denied environments, plug and play ISR sensor packaging and low cost UAV payloads. This paper describes the progress on these efforts.
Design of a Wireless Instrumented Payload to Test Medium to Large Unmanned Vehicles
John Clemmensen Jr., Mike Webster, Cheryl Bauman, Mike Fleming, Al Wicks, Virginia Tech
To provide real-time position and orientation information as well as data logging for the testing of large unmanned vehicles. A self contained package with standard hard points for mounting allows for integration with all vehicles. To minimize cost and size the payload was designed around a PC104 stack.
The payload incorporates 802.11 wireless connectivity, Differential GPS, 3-Axis Laser Ring Inertial Measurement Unit, 16-channel 12-bit Data Acquisition, and 4-input analog video recording with MPEG4 compression for quick streaming. The data is collected, integrated and stored onboard on a standard IDE hard drive. The data can then be viewed in real time via a 802.llb wireless LAN. The user can also decide what being collected and the rate at which it is sampled, to more accurately compare with the data collected by the vehicle itself. The payload interfaces with other devices through Ethernet, USB or Serial connection. The payload size will be 10x4.5x5.5 inch and weigh about 12 pounds without external IMU.
Design and Experimental Results of the Autonomous Deployment and Recovery of the HELIUM RED UGV from the VaTAR UAV
Michael Fleming, Brett Gombar, Shane Barnett, Dr. Charlie Reinholtz, Dr. Al Wicks, Virginia Tech
The Virginia Tech JOUSTER (Joint Unmanned Systems Test, Experimentation & Research) Program has developed the autonomous helicopter VaTAR (Virginia Tech Autonomous Rotorcraft) based on the Yamaha RMAX platform. Virginia Tech researchers have developed an UGV Deployment and Recovery Payload (DARP). The lightweight autonomous ground vehicle HELIUM RED (Helicopter-Lifted Unmanned Reconnaissance and Exploration Drone) has also been developed for aerial deployment and recovery by the DARP mounted to VaTAR. This paper presents the design, development, and experimental results of autonomous deployment and recovery of the HELIUM RED UGV by the VaTAR DARP. In particular, this paper discusses the mechanical DARP s design, command and communication for coordinating UAV and UGV deployment and recovery, and HELIUM RED onboard sensors, navigation, and mapping. The ability to deploy an UGV from an UAV enables heterogeneous missions in urban and countermine operations. An UAV may deploy an UGV in urban environments to provide surveillance and reconnaissance inside a building in which UAVs are physically restricted from entering through doorways and windows. In addition, deploying an UGV from an UAV may assist in efficient IED missions involving an UAV scanning for IED from the air and then deploying an UGV for pinpoint marking of the detected IED.
Holistic Contingency Management for Autonomous Unmanned Systems
Jerry Franke, Adria Hughes, Stephen M. Jameson, Lockheed Martin Advanced Technology Laboratories
Contingency Management -- the problem of recognizing, assessing, and responding to unanticipated events or conditions that impact plan execution -- is a key enabler for unmanned systems to become autonomous systems able to carry out assigned tasks or missions without continuous human supervision. We present a concept and technology for holistic contingency management that addresses all levels of autonomous system mission execution: from real-time flight-critical failures to long-term issues that can affect teams of vehicles. This concept features several key elements:
* Multi-level assessment: Monitoring, assessment, and response occur at multiple levels.
* Plan-based assessment: Monitoring is triggered by an assessment of dependencies and constraints on plan execution.
* Capability-based assessment: Ongoing assessment of vehicle mission-related capabilities based on subsystem and environment status.
* Predictive assessment: Monitoring and assessment of anticipated future events or conditions.
* Team-based assessment: Assessment occurs not just of individual vehicles, but at the team level as well.
We have implemented a holistic contingency management technology based on this concept to manage detection and response to unexpected events in unmanned vehicle operations. We will describe this Mission Effectiveness and Safety Assessment (MENSA) technology and its application to a number of Department of Defense autonomous system programs.
Discover Vision: A Framework for Building, Evaluating, and Testing Performance Based Machine Vision
Dr. Fernando Gonzalez, Daniel Barber, Brian C. Becker, University of Central Florida
This paper presents Discover Vision, a framework for the fast creation, evaluation, and testing of machine vision applications used in real time systems such as autonomous vehicles. The framework utilizes user edited scripts describing what image processing and feature extraction techniques to employ. Users can easily and quickly build a vision system by altering these scripts without changing the underlying framework, thus saving time when testing new methods. A graphical user interface is used to display the real time performance in the form of visual displays, processed frame rates, and system accuracy based on validation sets. It is possible to evaluate the effectiveness of a script by loading in live or recorded video for visual or numerical analysis. Higher level capabilities such as context switching are used within the framework to swap between scripts during runtime. This allows for an unmanned system to dynamically adapt to a new environment or lighting condition by executing a different set of image processing techniques. Context switching can be done using predetermined rules, or machine learning techniques. This paper describes the Discover Vision framework, demonstrates its application in an autonomous ground vehicle, and analyzes the resulting performance.
Modular Open Systems Approach for UUV Rapid Capability Insertion
Mr. David Herman, The Boeing Company
This paper describes a new approach to UUV system design that supports evolving industry standards for mission reconfiguration and rapid COTS insertion. Utilizing the principles of the DoD s Modular Open Systems Approach for Acquisition (MOSA) this new approach includes two principle parts: development of a modular, network-centric architecture for Boeing s legacy autonomous embedded software and development of an advanced simulator.
The new autonomous embedded software architecture is a complete reorganization of software subsystem interaction including well-defined and open interfaces, while maintaining all the capabilities of the legacy software through reuse of proven algorithms and logic. The result is a modular, network-centric architecture that is flexible and reconfigurable, regardless of whether the reconfiguration is a payload or sensor or an upgrade to system capability.
The advanced simulator supports real-time simulation for hardware-in-the-loop integration testing and faster-than-real-time simulation for software testing and mission evaluation. Each vehicle system is a node in the environment, allowing for simulation of a single system or multiple systems interacting.
The synthesis of the new architecture and advanced simulator creates a means for upgrading UUV system components and software in an efficient and reliable manner, allowing new capabilities to get to the field faster and cheaper.
Implementation of JAUS on a 2004 Cadillac SRX using a Potential Fields Architecture
Grant Gothing, Virginia Tech - JOUSTER, Jesse Hurdus, Virginia Tech
The Joint Architecture for Unmanned Systems (JAUS) is a unifying message set designed to reduce development time and promote interoperability between unmanned systems. By communicating only through JAUS specified messages, components can be reused in multiple systems and can communicate with any other JAUS compliant component. Currently, JAUS is mostly implemented on smaller ground vehicles with very basic waypoint navigation and teleoperation capabilities. This article discusses the implementation of the JAUS architecture on a 2004 Cadillac SRX using a potential fields reactive paradigm. Several new JAUS components have been defined and implemented to make this possible. A Vector Summer intelligently decides a path for the vehicle to follow by accepting a global vector from several components (waypoint driver, obstacle avoidance driver, etc.) and sums them based on a desired weighting algorithm. A Control Arbiter monitors vehicle status and controls the balance of vehicle control between the computer and a remote-operator. This allows for multiple E-Stop levels and several dynamic modes of control. In total, this JAUS compliant system allows for the deeper exploration of behavior-based vehicle control than is currently defined by the JAUS architecture.
Method Development & Performance Evaluation - Shipboard Operations With The Boeing Unmanned Little B
Mark Hardesty, The Boeing Company
An immediate goal of the program is to demonstrate precision autonomous terminal operations on a moving elevated platform, such as a ship. NovAtel has offered a solution in the form of a Moving Baseline RTK GPS with a relative accuracy the same as the fixed reference station solution. This moving baseline solution has been further integrated into the NovAtel SPAN system, which integrates an RTK GPS receiver with a tactical grade Inertial Measurement Unit (IMU). The developmental test shall employ a SPAN system on the moving platform along with a system on the aircraft. The position solution provided by this system has been demonstrated to be accurate to approximately 15 centimeters. Using this position data, the aircraft should be able to repeatedly land and depart from a standard NATO Harpoon grid in a reliable and repeatable manner. Integration of a system that constantly analyzes ship motion and predicts periods of quiescence shall be concurrently completed, so that intelligent landing commands can be issued in heavy sea states.
Integration of a PEM Fuel Cell into a Slow Speed UAV
Christopher Herwerth, Alan Ko, Dr. Maj Mirmirani, Multidisciplinary Flight Dynamics and Control Laboratory, California State University, Los Angeles
An all composite, hydrogen fuel cell powered UAV is being developed at the Multidisciplinary Flight Dynamics and Control Laboratory (MFDCLab, www.calstatela.edu/centers/mfdclab) at California State University, Los Angeles. The fuel cell UAV is intended to demonstrate the application of fuel cell technology to unmanned aerial vehicle design. A need was identified for a slow speed, low altitude UAV that offered the advantages associated with fuel cell power, such as zero hydrocarbon emissions and low heat signature. A proton exchange membrane (PEM) fuel cell with a metal hydride storage tank provides up to 670 Watts of power to an electric motor. The low power density of the PEM fuel cell demanded the use of a high aspect ratio wing in the UAV s aerodynamic configuration. This paper details the integration of a PEM fuel cell into an unmanned aerial vehicle. Results of fuel cell testing, optimization of the fuel cell power plant configuration and the design and construction of the UAV is presented. Flight testing under fuel cell power is scheduled for May 2006.
Rotorcraft Unmanned Air Vehicle Navigation in Uncertain Urban Environments
Jason Howlett, Matt Whalley, US Army AFDD, Peter Tsenkov, Greg Schulein, San Jose State University Foundation
The ability for a small-scale Rotorcraft Unmanned Air Vehicle (RUAV) to navigate effectively in uncertain obstacle environments is an essential requirement for RUAV nap-of-the-earth (NOE) operations, particularly in urban settings. The ability for the vehicle to detect, update its terrain representation with, and react to, changing obstacle information is critical to vehicle survivability and mission success. Furthermore, other factors, such as exposure to threats, mission execution time, and vehicle performance capabilities, need be considered. A solution to the NOE Obstacle Field Navigation (OFN) problem for a small RUAV with an on-board obstacle detection sensor will be presented. The approach builds on previous work in Obstacle Field Route Planning and involves the periodic replanning of the desired flight route as obstacle information changes. A three-dimensional terrain representation is updated and validated with sensor information. Navigation plans are generated from the current terrain information, as well as both navigational and mission constraints. A simulation evaluation and demonstration of a small-scale RUAV successfully navigating through an uncertain urban environment will be given. Additionally, preliminary flight test results of the component technologies will also be presented.
PDA Technology for Small UAV Operations
Dr. Douglas Kliman, Peter A. Krawczak, TacGeo
Small ruggedized devices such as Personal Digital Assistants (PDAs) are rapidly becoming a viable means for managing tactical assets in the field, including the feeds from small UAVs. TacGeo recently demonstrated the use of PDAs as part of a military field experiment. In the experiment, PDAs were used to manage a tactical wireless sensor network deployed over a 10km area. Two Raven UAVs were continuously flying over the area, collecting and broadcasting real-time video imagery on the network. A self-replicating peer-to-peer data structure was used to share data between devices, eliminating critical nodes. Data from the sensor network and UAVs were integrated into a real-time geospatial display on Tacticomp PDAs carried by soldiers in the field. The Tacticomps displayed a map interface with moving icons representing the sensors and UAVs. The soldiers used a touch-screen to query the UAVs for status and view streaming video. In addition, the soldiers could communicate through the Tacticomps via VOIP and text messaging. The results of the experiment showed that the PDA equipped soldiers gained a high level of situational awareness and were able to conduct tactical operations with increased effectiveness.
Long-Term Mine Reconnaissance System (LMRS) Field Testing Overview
Damian Drozd, Fred Sheldon, David Flowers, Steve Nicinski, The Boeing Company
The LMRS is a UUV based system developed by Boeing Marine Systems in Anaheim, CA for the US Navy with the unique capability of being impulse launched from and recovered through a 688 class submarine s torpedo tubes. The LMRS is currently the US Navy s primary 21 UUV program, and is serving as the first generation of submarine based UUV systems, paving the way for follow on efforts such as the Mission Reconfigurable UUV System (MRUUVS). Over the past two years, LMRS hardware has undergone extensive field testing and has repeatedly demonstrated its performance capabilities on both surface craft and 688 class submarines. The system has completed hundreds of hours of in-water operations and has demonstrated such functionality as dynamic impulse launch from a 688 class submarine, GPS position fixing via surface operations, autonomous docking with a submarine, and its recovery capabilities. This paper will highlight the capabilities demonstrated during real world operations and will address numerous lessons learned throughout the testing of this highly complex system.
The authors are engineers with the Boeing Company and have been intimately involved in the LMRS system test program.
The Use of Multidisciplinary Optimization Methods in the Design of Unmanned Vehicles
Dr. Dominic Palumbo, AAI Corporation
The use of Multidisciplinary Optimization (MDO) software has been highly productive for design optimization of unmanned air vehicles of various types and for the development of aircraft mission performance models. Three different applications that highlight the use of MDO tools are presented in this paper; namely, the design of ducted fan propulsion systems, the sizing of UAVs to meet specified mission profile and payload requirements, and deriving performance models of propeller driven UAVs based upon flight test data. These applications required integrating third party computational fluid dynamics and structural analysis codes as well as MATLAB and Excel application codes written in-house. The specifics of the computational logic used in each application and some results are presented.
Modeling, Simulation, and Analysis of Crew Performance U.S. Army Unmanned Aerial Vehicles
Regina Pomranky, US Bruce Hunn , US Army Research Laboratory
This paper discusses the importance of modeling, simulation, demonstration, and the use of other analytical methods in order to understand and predict crew performance for unmanned aerial vehicle (UAV) military environments during system evaluation or as an evolutionary process. The problem with developing specialized UAVs for novel military environments is the difficulty of making informed risk assessments of design options early enough in the design process to be cost effective. Unfortunately, this does not always occur and systems are fielded before being fully evaluated. Modeling and analysis of UAV systems allows the designer to develop a set of expectancies about the performance implications of design options, crew sizes, and crew training requirements before formal testing or even prior to producing a fully developed system. Examples are provided from two dissimilar U.S. Army UAV programs, the Raven, a small unmanned aerial vehicle and the Shadow-200, a tactical unmanned aerial vehicle. Specific issues discussed include automation, crew workload, crew coordination, and their implications on effectiveness and safety. The discussion will involve a review of where strengths and weaknesses of the system were better identified by modeling, simulation, or demonstration, and allowed designers to make trade-offs often concurrent with operational deployment.
Enabling Tight Coordination for Complex Multirobot Missions
Nidhi Kalra, Anthony Stentz, Carnegie Mellon University Robotics Institute
In this paper we present a coordination framework capable of autonomously achieving very complex real-world multirobot tasks such as security sweeping and communication-constrained exploration. In security sweep, robots must work in concert to clear an area of mobile adversaries in a single pass of the environment. In communication-constrained exploration, we task a team with exploring an environment but require that, for safety and to monitor progress, the team maintain an connected an ad-hoc communication network. Achieving these tasks involves solving a tightly-coupled multirobot planning problem efficiently and robustly. Our system, called Hoplites, is a market-based coordination approach that does so by dynamically responding to the changing difficultly of the mission. That is, when problem scenarios are fairly simple, robots save time and energy by using a fast, inexpensive coordination mechanism. When problem scenarios are challenging and require significant coordination, robots employ a more capable, more expensive mechanism to produce better solutions. Hoplites is robust to communication failures, uncertain task information, and dynamic environments. We have demonstrated Hoplites in simulation, on a team of indoor planar robots, and on a team of large outdoor vehicles.
GHMD Modeling and Simulation for Experimentation
Keith Weisz, John Auborn, NAVAIR
NAVAIR s Integrated Battlespace Arena (IBAR) located at China Lake, California has executed tasking to develop a virtual simulation of the GHMD air vehicle, sensors and the Mission Control Element. The goals of the simulation include the ability to act as a distributed simulation with data to flow to multiple data consumers and to receive tasking from the battle commanders. The IBAR developed models for the new maritime sensors that are part of GHMD. To date, the IBAR s simulated GHMD has successfully participated in the Trident Warrior 2005 forcenet experiment and the Joint Expeditionary Forces Experiment 2006. The simulation was successfully used to provide a virtual GHMD system during the simulation days and allowed transition to GHMD livefly days in the TW05 MAINEX. The simulation data was successfully transited through the data processing chain to Home Land Security/Home Land Defense and defense organizations. Personnel using the simulation data exercised their tasks using the simulation data as if it came from the real GHMD platform. This has been a tool which exercises all of the ground based processes and personnel and is poised to support future HALE UAVs.
Global Hawk Automatic Contingency Generator Flight Test
Steven Yamaguchi, 452 FLTS, Bill Norton, JT3, Lejui Brand, USAF
This paper describes the Global Hawk (RQ-4A) Automatic Contingency Generator (ACG) software integration flight test. The RQ-4A employs contingency (unexpected event) planning as part of its autonomous mission plan. With the present system, an mission plan must contain many manually-planned contingency routes and undergo a 6-degree of freedom (6-DOF) verification. Average mission planning has required weeks of effort when the operational requirement to plan a new mission is within 12 hours. The ACG software sought to reduce the time and effort to plan a mission by automatically generating a contingency routing aboard the vehicle from anywhere on the primary route with no impact to flight safety. The test program objective was to verify proper functionality of ACG during mission planning and flight operations. Testing consisted principally of triggering each contingency scenario, typically in an emulated fashion, under multiple system and flight conditions and autonomously flying-out the ACG-generated route. Air vehicle and ACG behavior was observed and compared against simulation predictions. The suitability of pilot displays and controls was evaluated. This paper describes the RQ-4A system, legacy and ACG contingency planning, ACG features, and the test/safety planning, test conduct, and results, plus conclusions and lessons learned.
Modeling for UAS Collision Avoidance
Dr. Andrew Zeitlin, MITRE Corp.
If UAS are to be granted full access to civil airspace, their safety case must address collision avoidance, including the lack of an onboard pilot who could see-and-avoid other traffic, as on conventional aircraft.
Today, many manned aircraft are equipped with the Traffic Alert and Collision Avoidance System (TCAS II), the world standard system for collision avoidance. However, simply installing that system aboard UAS is problematic for a number of reasons. These include new mission types; the limited aerodynamic climb and descend performance of many today's unmanned vehicles, rendering them incapable of appropriately responding to TCAS Resolution Advisories; and latencies involving the downlinked traffic information and advisories to a remote pilot and the subsequent initiation of an avoidance maneuver. Each of these factors affects the safety calculation.
The MITRE Corporation developed the Monte Carlo simulation capability that has been a key part of the approval process of every version of TCAS. MITRE's internally-directed research program is using this capability to investigate the unique aspects of UAS collision avoidance. Modeling of some air vehicle flight profiles, new missions, and encounter geometries are beginning to produce results that could greatly affect integration of UAS in civil airspace.
Mixed-Initiative Adjustable Autonomy in Multi-Vehicle Operations
Vera Zaychik Moffitt, Jerry L. Franke and Stephen M. Jameson, Lockheed Martin ATL
Unmanned systems must provide robust operations despite dynamic levels of human workload, trust, and mission requirements. One of the promising approaches to this problem is adjustable autonomy, which allows responsibility for decision making to be shifted between vehicle and operator. This enables the overall system to adapt to changes during mission execution, but presents several challenges including: controlling the shift of decision-making responsibility; maintaining consistent and stable operation during the shift; maintaining user trust at different levels of system autonomy; and maintaining user situational awareness.
5 hp Class Heavy Fuel Engine with Peak Pressure Management without High-Pressure Fuel Injection
S. Paul Dev, D-STAR Engineering Corp.
Unmanned Ground Vehicles (UGVs), as well as UAVs, need heavy fuel engines that are compact and light-weight, while retaining the fuel economy of 4-stroke diesel engines. UGVs with hybrid- electric drives also need high engine operating speeds to minimize the size, weight and cost of the engine-driven electric power generators. The proposed presentation will report on a U.S. Marine Corps / D STAR effort to develop a high-speed heavy fuel engine that has been successfully operated at 8000+ RPM without the use of a high-pressure fuel injection system. The engine uses a proprietary system that automatically limits the peak combustion pressures, thus minimizes the stresses in and weight of the engine structural components. The system also enables operation of the engine with low Noise, Vibration and Harshness (NVH). Details of the peak pressure management system and results of the testing will also be presented. The presentation will conclude with an indication of future directions and applications such as hybrid drives for UGVs and USVs, fans for UAVs and APUs for the larger systems.
|
Events
