Constructive Simulation OneSAF and American's Army Game Technology

Transcription

1 Constructive Simulation OneSAF and American's Army Game Technology Ms Ha Ly, Mr. Lowell Baker US Army, Program Executive Office-Simulation Training and Instrumentation (PEO-STRI) Research Parkway Orlando, Florida U.S.A. Abstract. Robots are increasingly being used to assist the warfighter in a variety of missions to accomplish objectives in theater. The robots increase mission performance, combat effectiveness and personnel safety by allowing warfighters to conduct certain activities at safe standoff distances. In advancing the development and deployment of military robots, one of the main challenges to overcome is matching the unmanned systems capabilities with a warfighter s needs. There is a significant gap between the rapidly developing unmanned system technologies and the needs of the soldier on the ground. The Robotic Systems Joint Program Office (RSJPO) has identified that One Semi-Automated-Force (OneSAF) can simulate entity representations with varying levels of fidelity for unmanned ground vehicles that are composable and autonomous. They also recognized that OneSAF supports their need to be able to order these simulated entities and their aggregates to perform a particular course of action or behavior within the simulated environment. OneSAF was chosen by the RSJPO as an analytical tool to provide the ability to conduct exercises/experiments on robot assets in a constructive and virtual environment. These exercises/experiments are intended to help requirement developers refine Concepts of Operations (CONOPs) and Tactics, Techniques, and Procedures (TTPs) for unmanned ground vehicle. 1. INTRODUCTION This paper describes how One Semi-Automated-Force (OneSAF), a modern constructive simulation, is enhanced to fulfill this need for a tool to portray the influences of robots operating in the Contemporary Operating Environment (COE) through integration with the 3D front-end and First Person Shooter capabilities of American s Army, a multiplayer video game. This integration completes the 1 st person experience with HITL compatibility in a 3D stealth view. By simulating the COE including environmental conditions and battlespace circumstances, OneSAF can represent a variety of scenarios allowing the analyst the ability to examine several improvements to robot capabilities and the warfighter s tactics, techniques, and procedures in performing missions. composability aspects of the architecture allow users to attach as many sensors and weapons to an entity as desired. The robot entity compositions are composed of behavior and physical capabilities to represent the MARCbot, Packbot, Talon, MAAR, and Gladiator. The robot unit composition allows the user to create hierarchical collections of entities. The user can specify command and support relationships, formation location, as well as communications roles on the various radio networks. The Unit Composer allows embedding previously composed entities and/or units into other organizations. For example: Explosives Ordnance Disposal (EOD) platoon units can be composed of a platoon leader and two squad units. The squad units are composed of a squad leader and two teams: one team contains a truck and three infantry, and other team contains a truck, three EOD specialized infantry, and a talon robot. 2. ROBOTIC COMPOSITONS OneSAF simulates entity representations with varying levels of fidelity for unmanned ground vehicles that are composable and autonomous. The OneSAF Entity Composer allows a user to construct battlespace entities: tanks, aircraft, individual combatants, etc. from models such as turrets, hulls, sensors, and weapons. A user interacts with the Entity Composer by selecting specific models to represent that entity. Users select various models and model resolutions to meet their specific simulation needs. For example, a high resolution human mobility model could be used in an Entity Composition to support an urban operations scenario with few entities. Alternatively, a low resolution human mobility model might be used in modeling a crowd, where large entity counts are needed to provide context for the issues under study. The 2.1 Construct Entities Currently, OneSAF has a library of various fidelity (ultra low, low, medium, and high) physical and behavior models. These adequately perform modeling needs for a constructive simulation in responding to the immediate users of OneSAF. However, after conducting a straight forward mapping of current compositions, it was determined that OneSAF did not have any robotic entities. At the request of RSJPO, the development team created entity representations of various individual robotic systems: 1) Talon 3B Robotics System weighs about 80 pounds and has a range of 1.6 kilometers, capable of being fitted with seven image-relay cameras and can operate in sand or mud and even climb rocks.

2 2) Packbot EOD revision B- a rugged, lightweight robot designed to conduct EOD, HAZMAT handling, search, surveillance and reconnaissance, hostage rescue and other tasks. 3) MARCbot IV - Multi-function Agile Remote Control Robot, is a small robotic platform aimed squarely at inspection of suspicious objects during IED Sweeps, provided as a remote inspection platform allowing soldiers to achieve stand-off when trying to determine if an object is an IED. 4) MAARs - Modular Advanced Armed Robotic System, use the more powerful M240B medium machine gun and has significant improvements in command and control, situational awareness, maneuverability, mobility, lethality and safety compared to its SWORDS predecessor. 5) Gladiator - includes a rugged mobile base unit, interchangeable mission payload modules, and a wireless data link; it s operated with a hand-held controller and helmet-mounted view screen. 2.2 Generate Physical Models After reviewing the current OneSAF physical model, the development augmented the physical model library in support of robotic modeling. Combat physical models provide the representation of combat system and their interactions with the environment and other entities. Robotic Mobility To determine the position, velocity, direction, and orientation of robot entities over time. The steady state model is used to produce the terrain adjusted speed for the robot while the concepts of normal and free fall movement states from the Dismounted Infantry Mobility model are used to calculate the robot's new position and orientation. Remote Control - To determine the type of control currently in use (RF wireless, tethered, etc.), damage of controller or robot that would prevent communications, and whether the controller and RC platform are within range. Robot Vulnerability - To determine the vulnerability of a robot target as a result of various munitions (mines, direct fire munitions, indirect fire munitions). AMSAA models were used for the vulnerability from direct fire, indirect fire, conventional mines, and improvised explosive devices. Robotic Sensor To facilitate the sensing agent shares sensing/perceptions between a robot and its controller to support the robot camera fixed focus and zoom capabilities. Consumable Model To maintain persistent battery consumption on a robot. The development team found an opportunity here to leverage the battery consumable existing framework on a concurrent project and was able to reuse the code to support effort without starting from the ground up. The development team also integrated mobility and sensing control to allow the operator an immersive experience of robot movement, camera selection, and camera pivot through the control of the keyboard (or other mapped peripheral controls). Through the Headup Display (HUD) status information such as hull heading compass, lat/long location of robot, and currently selected camera are presented. 2.3 Form Behavior Models After reviewing the current OneSAF behavior model, the development team extended the behavior model library in support of robotic modeling. Minor updates to existing OneSAF behaviors were made to: 1) allow the robot to mount and move with the rest of its ordered unit, command a robot to mount and dismount from another ground vehicle in its unit, and 3) allow another ground vehicle to move to and pickup a robot. Other major behaviors created include: Robotic Reconnaissance - allows a section or team with a robot to perform a point, route, or area recon. The behavior also allows termination upon first detection of a suspected IED. Robotic Explosive Disposal - orders a team or section to conduct unexploded ordinance disposal. The disposal techniques are explosive (charge emplacement) and relocation; an unsuccessful disposal attempt is possible. Identification of Explosive Devices - a persistent behavioral capability given to the EOD trained entities to further identify explosives as a hazardous or nonhazardous. Robot Driver Behavior - performs obstacle avoidance for a robot; avoids sensed IEDs and terrain features such as urban clutter and buildings. Robotic Self-Recovery - This is a persistent capability of RF-controlled robotic entities to attempt to return to a position where communications with its controller are possible, in the event of a disruption or termination of the robot-controller communications link. 3. REINFORCE GUI OneSAF is known for being composable; it provides tools across the pre-runtime, during runtime, and post runtime phases of a simulation event. The OneSAF battlespace composers support a form of composition specialized for the user s different needs. The battlespace composers know about the types of model components that make up an entity, unit, or behavior as well as how model components represent data, so they can offer services to users to distinguish different types of equipment such as an M1A2 from a T80 tank. All of the battlespace composers represent their compositions in XML. There are three battlespace composers: Entity Composer, Unit Composer, and the Behavior Composer. OneSAF also extends the existing tool set to support growing new products with a Graphical User Interface.

3 3.1 Enhance Tools To empower end-users to modify robotic needs, the following tools were provided prior to completion of this project. Remote control Allows an operator to create a link between a controller and robot entity while controller entity is assigned to the same workstation as the robot. Properties/Status Panel Updates Displays the controlling/controlled entity when a link is established, the status of the link between a robot and its controller, and the battery status of a robot entity. Reports - Displays situation messages sent at the end of the robot reconnaissance behavior and unexploded ordnance (uxo) disposal behavior. I Plain View Display (PVD) Enhancements - Display urban clutter representation and tool tip updates to display mines being carried by the robot (i.e. uxo relocate). Technical Capability to Display Articulated Parts. Decent API or Customizable Interface Synchronized Multi-Channel Low Cost to acquire Cross-Haired, HUD, and Other Overlays Many Visual Models Included or Available Good Night Vision and Infrared Good Quality Visuals 3.2 Update Terrain The terrain development team stepped up the terrain capabilities to support robotics terrain database to show the impact on simulated behaviors and activities. The terrain database is 1-degree latitude by 1 degree longitude of South West Asia region. The dimensions of the downtown area with buildings are 10km x 3.2km. The dimensions of the downtown area with urban clutter are 3km x 2.6km. This terrain contains the following but not limiting to features: buildings 29000, roads 65000, trails 9500, Urban clutter (Oil drums, dumpsters, barbed wires, rubbles, vehicle barriers. 4. SELECT A VIRTUAL FRONT END To support the need for an Image Generator for sensor views and camera views for realistic maneuvers, the analysis team conducted a trade study of several Image Generators to meet the requirements of the project. The objective was to train robot operators using medium and high fidelity entity representations of the robotic systems. The Image Generator selected must provide sensor views to operator, the robot effectively for situational awareness and reconnaissance. The Image Generators we looked at included: MetaVR VSRG, NVESD NVIG, MultiGen VegaPrime, and Quantum3D Mantis. There were eight criteria used as discriminator for IG selection, listed in Table 1, and ranked in the order from most to least important.

4 Robotic IG Trade Study Meta VR VRSG MultiGen VegaPrime Display Articulated Parts Decent API Synchronized Multichannel Low cost Overlays Many Visual models Good night vision and IR Good Quality Visuals Windows Very strong technical solution, high cost Windows XP and Linux Total Score Operating Systems Comments Very high cost NVESD NVIG Linux Capable. Support is proven and reuse important. Quantum3D Mantis Windows XP Good API Table 1. Trade study of different Image Generators

5 There are eight (8) criteria used as discriminates for IG selection, ranked in order from most to least important. For each vendor product, scores of 0 to 4 have been assigned for each of the eight. Weights are assigned in priority order, with a weight of 8 for the first and most important criterion, 7 for the second, and so on. So a perfect score would be 4 s for all discriminators, or 8x4 + 7x4 + 6x x4 + 1x4 = 144. According to the trade study, NVIG is the IG of choice, based on the scores and code reuse from an existing project. However, at the same time that this trade study was conducted, OneSAF had an independent team working separately on developing an internal 3D frontend. The benefits of using the internally developed 3D front end outweighed the integration effort needed to re-establish NVIG connection to OneSAF. The majority of OneSAF code is rendered in Java, and the JMonkey open-source software was java based extension as a result of a Game like interface Usability study [5],[6]. Ares, name of God of war, is given to this internally developed 3D front-end. 4.1 Ares One of Ares s requirements is to make the 3D view composable. The 3D design abstracts OneSAF out into separate virtual management interfaces (terrain, battle space objects, warfare interactions, atmospheric, input, HUD, etc.). The 3D design depends on the virtual components. The team then creates an implementation of the interfaces and links them to the virtual interfaces using the system composer. This allows the team to create different applications by linking in different input or HUD managers. For instance, at the beginning of the effort the team was working on a strategy handler to control the entities by executing behaviors/missions. This corresponds to using a specific HUD, but in supporting this Robotic project the team went on to have a separate application that uses a first person HUD and first person input through an ownship framework so operator can take control over entities in the simulation and be immersed in the virtual world. The 3D visual supports first-person perspective from robot cameras, rendering of robot s articulated parts from the entity rigid body data, presentation of IEDs on 3D visual according to the perceived truth, and Robot UXO and reconnaissance behaviors orderable through the 3D Viewer. This is a significant accomplishment for OneSAF considering the open-source merit from a small team operating to meet a tight timeline. 5. OneSAF AND AMERICA S ARMY INTEGRATION The Robotic Systems Joint Program Office (RSJPO) Robotic Systems Collaborative Environment (RSCE) uses the America s Army platform to provide the simulation environment for the Robotic Systems Collaborative Operations Initiative. Utilizing the Unreal game engine, America's Army leads players through scenarios that model basic Army training, teaching teamwork and tactics. The objective of the RSCE is to provide a simulation environment to support testing, prototyping and demonstration of interoperability technologies for Robotic System (RS) with Intelligence, Surveillance and Reconnaissance (ISR) missions. Due to the nature of the experimentation that occurs within these networks, there is a need to include non-deterministic elements in the experiments being performed. Existing simulations and applications provide more deterministic solutions based on established and programmed rules. The desire in evaluating things like doctrinal changes or analysis of changes to organizations is to include an element that is more subjective. To insert this subjective, human element, the usual solution is to include human interaction and intervention at some level. This may be people interacting in a role playing environment without significant computer simulation, a combination of live role players in a virtual environment and machine driven constructive simulations, or simply having human intervention in controlling the constructive simulation entities. The environment leverages the existing virtual simulation capabilities of America s Army (AA) as well as the constructive simulation capabilities of OneSAF. The development team integrated OneSAF with America s Army (AA) over Distributed Interactive Simulation (DIS) protocol as shown in Figure 1. During this process, new configuration files were created to match OneSAF s PDU fields to AA s PDU fields, and new robot entities get appended within OneSAF s DIS framework. In facilitating affects of IED in OneSAF the team created the explosive, rendered the detonation, and processed the removal of an OneSAF IED through DIS protocol to appear in AA s

6 display. The terrain supported during this AA and OneSAF integration is McKenna database, as this is the only database AA is supporting with OneSAF. previous project and even robotic capabilities from our parallel FCS work, highlighting the open source and extensible architect framework of OneSAF. This is a good use case for the OneSAF business model as well because through Program Executive Office for Simulation, Training, & Instrumentation (PEOSTRI), we proved to be able to keep pace with the necessary changes and our extensible framework supported collaboration and shared resources within the modeling and simulation community. 7. REFERENCES Figure 1. OneSAF and AA RSJPO also requested that OneSAF provide an After Action Review (AAR) capability. The development demonstrated the ability of OneSAF s After Action Review (AAR) to data collect on external entities (AA s entities) represented in the OneSAF simulation. The AAR component in OneSAF receives simulation events through its simulation interface. Entity movements, entity state changes, firing events, and detonation events are examples of events handled though the simulation interface. A simulation interface might connect through DIS/HLA, or through a simulation specific protocol like the OneSAF s Simulation Object Runtime Database (SORD). OneSAF s AAR component provides a take home package that allows for mobile viewing of the battlespace after action review. The challenges in integrating with AA include the developmental aspect of America s Army perfecting its requirement and software. America s Army only communicated via (DIS), so we had to interoperate via this protocol. Since this developmental functionality is not well documented, we had to communicate regularly with the AA and RSJPOs representative to establish solid use cases. 1 OneSAF library of Physical Knowledge Acquisition Document and Domain Behavior Document, The OneSAF.net Developer Portal. < 2 OneSAF Program Office and Robotic System Joint Program Office. Robotic Training Support Vision Statement, America s Army Visualization Platform Operator s Manual. P 4-9, R. Wittman and J.R. Surdu, "OneSAF Objective System: Toolkit Supporting User and Developer Lifecycles within a Multi-Domain Modeling and Simulation Environment," SimTecT 2005 Simulation Conference and Exhibition, Syndey, Australia, 9-12 May Stephen Lopez-Coute and Rodney Figaroa, An Analysis of Game Design to Improve Complex User Interfaces in Constructive Simulations, NATO Modelling and Simulation Group Conference, UK, February Dwyer, T. Usability Study for OneSAF Graphical User Interface. PM OneSAF Internal Development Document CONCLUSION This project was a customer-funded effort, which took us one year to develop and integrate the robotic entities with a virtual environment via our internal SORD protocol and DIS protocol. While supporting this effort, we were able to reap the benefit of leveraging the ownship framework from a