Intelligent Vehicle Systems Symposium CAT/RF Simulation Lessons Learned Christopher Mocnik Embedded Simulation Team Email: [email protected] (586) 574-5491/DSN 786-5491 Fax: (586) 574-5008 RDECOM TARDEC Vetronics Technology Area (AMSTA-TR-R, Mailstop 264 Warren,
Download ReportTranscript Intelligent Vehicle Systems Symposium CAT/RF Simulation Lessons Learned Christopher Mocnik Embedded Simulation Team Email: [email protected] (586) 574-5491/DSN 786-5491 Fax: (586) 574-5008 RDECOM TARDEC Vetronics Technology Area (AMSTA-TR-R, Mailstop 264 Warren,
Intelligent Vehicle Systems Symposium CAT/RF Simulation Lessons Learned Christopher Mocnik Embedded Simulation Team Email: [email protected] (586) 574-5491/DSN 786-5491 Fax: (586) 574-5008 RDECOM TARDEC Vetronics Technology Area (AMSTA-TR-R, Mailstop 264 Warren, MI 48397-5000 11 June 2003 UNCLASSIFIED Tank-Automotive Research, Development & Engineering Center 1 Agenda • Introduction • Vetronics Technology Test bed (VTT) • Crew integration and Automation Test bed (CAT) /Robotic Follower (RF) Overview • Unmanned Combat Demonstration (UCD) Overview • Embedded Simulation System Overview • Lessons Learned Schedule and Requirements ESS Process Overview Distributed Control Inter-process Communications Unmanned Ground Vehicle (UGV) Process Reconnaissance, Surveillance, Target Acquisition (RSTA) and Automatic Target Recognition (ATR) Image Generation Scenario Development Terrain Database Development Hardware Implementation • Conclusion 2 Introduction The Embedded Simulation Team from TARDEC, along with DCS Corp. Successfully designed, developed, and integrated an Embedded Simulation System (ESS) that supported both the CAT/RF experiments as well as the Unmanned Combat Demonstration (UCD) with the Lead Systems Integrator (LSI). The following presentation will provide some background information on previous and current programs and then discuss lessons learned from CAT/RF and UCD development. 3 Background Information 4 Vetronics Technology Test bed (VTT) • Goal: Improve the war fighting capability of ground combat vehicle systems. • Approach: Develop, integrate and field test advanced Vetronics technology for ground combat vehicles. Advanced Crew Station Interface • • • • Indirect Vision Driving Multi-function Displays Speech Recognition 3-D Audio Advanced Electronics Architecture Embedded Simulation System • Modified M2A0 Bradley Fighting Vehicle used as test platform. 5 Crew integration and Automation Test bed (CAT) • Goal: Design an advanced 2-man crew station for a system < 20 tons incorporating the FCS fight, carrier, reconnaissance, and C2 of unmanned systems. • Approach: Build on VTT technologies and provide developments for integration into the FCS demonstrator. Additional capabilities include: Robotic Mission Planning and Control Robotic Assisted Driving Decision Aids RSTA Control • Prove out technology developments using a FCS class chassis - Interim Armored Vehicle (IAV) Infantry Carrier Variant (ICV) or Stryker. • Enhanced ESS to support Robotic planning and control, and RSTA operation. 6 Robotic Follower (RF) • Goal: Develop, integrate and demonstrate the technologies required to achieve unmanned follower capabilities for future land combat vehicles • Key Requirements: Dismounted or Mounted Following. Semi-autonomous perception. Significant separation times and distances. Map data and sensor terrain feature registration. Road detection On-coming traffic detection. • H/W & S/W design based on Demo III. • Chassis: Stryker representative of FCS mounted systems. XUV representative of mule system 7 LSI Unmanned Combat Demo (UCD) ARV-1 RF ATD (w/ COUGAR turret) RSTA and Engagement RSTA and Engagement ARV-2 Demo III XUV Targets Simulated targets for Maneuver Real targets for Live-Fire •Surrogate Platform • Mobility (~16T) • Semi Autonomous Nav • Simulated Objective Capability • Turret/Weapons • RSTA Demonstrating: • 1:1 Operator to ARV Control • ARV Engagement Control Vehicle (CV) CAT ATD •Surrogate Platform • Mobility (~2.5T) • Semi-Autonomous Nav • Simulated Objective Capability • Turret/Weapons • RSTA •Surrogate Platform • Mobility • Semi-Autonomous Nav • 2 Crew Stations • C2 8 Embedded Simulation System (ESS) MISSION APPLICATIONS Embedded Training Mission Rehearsal Mission Planning Crew Stations FCS Class Vehicle Vehicle and crew interaction data Embedded Simulation System SIMULATION BASED ACQUISITION Simulated Turret Virtual Lethality Virtual Sensors Simulated ATR Simulated ATT Simulated C2 VEHICLE SIMULATIONS Mobility Survivability Virtual OPFOR Virtual Friendlies Virtual Battlefield OPERATIONAL APPLICATIONS Battlefield Visualization Terrain Registration Virtual Sensor Coverage Virtual Lethality Coverage 9 CAT/RF and UCD Simulation Lessons Learned 10 Schedule and Requirements Specification • Schedule: In a perfect world, appropriate development time should be allocated in the program schedule to adequately design, develop, and fully test systems in order to meet the customer’s needs. The CAT/RF ATD was originally a two year development effort. However, in order to support the FCS program and meet the FCS milestone B decision, the amount of available development time was reduced to one year. In that time, the ES Team and DCS had to support not only the CAT/RF simulation requirements, but that of the UCD as well. • Requirements Specification: Requirements specification may be the most important part of any software engineering process. • Captures the customer’s needs. • Basis for system design. 11 Schedule and Requirements Specification Firm requirements were difficult to capture and document: • FCS and LSI unmanned systems concepts were still evolving. • Physical assets available for demos such as RSTAs, vehicle platforms, weapons platforms etc. were still being negotiated. • Results: DCS and the ES Team were able to successfully demonstrate an embedded simulation capability, though many non-critical requirements had to be dropped. Given a finite amount of time, a finite number of human resources, and “creeping” requirements, trade-offs have to be made in order to meet hard deadlines. • Lessons Learned: Work closer with internal and external customers to lock base system requirements as early in the development process as is possible. This will allow for more effective use of the available time. 12 ESS Process Overview • Processes functionally partitioned by the service they provide. RSTA Server CAT Vehicle SimControl Manager AKIT/BKIT ICD Akit Interface UGV PIU Data PIU Comm Object AAR • VTT code reused to largest extent possible, with new processes for UGV and RSTA control added. Mobility NIU World • Reused processes modified to incorporate new functionality DIS PDU’s CGFI OTB Sight Weapons • Inter-process communications performed through PIU Comm Object. 13 Distributed Control • The CAT ESS software expanded on VTT capabilities by adding support for dynamic control of any unmanned asset (controlled by the ESS) from either of the crew stations. • However, the VTT ESS code was designed around control of one ownship vehicle and therefore was tightly coupled with VTT ownship vehicle states and modes. As a result, some limitations were present in the CAT implementation. It was not possible to control two independent battlefield visualization eye-points for example. Crewstation 1 Vehicle 1 Vehicle 2 Crewstation 2 Vehicle 3 Crewstation N Vehicle M Embedded Simulation • Lesson Learned: A more flexible software architecture will be needed to be less dependant on vehicle states and modes. • In the future it is required to not only dynamically control unmanned ground assets, but unmanned air assets as well. • The notion is to control any asset from any crew station at any time. 14 ESS Inter-Process Communications • The PIU Comm Object was an effective tool for managing interprocess communications. Proven and well understood. Code Reused from VTT program • However, an extra “layer” of management is needed at the A-kit/BKit interface level. • Lesson Learned: Allowing each process to write directly to a common shared memory area with the vehicle will eliminate the need for an extra layer of management. • Will perform trade analysis on the impact of moving to the WSTAWG OE for this purpose. 15 ESS UGV Process • UGV Process Instantiates a UGV object for each UGV under ESS control in the scenario, and communicates to other processes via PIU Comm Object. • Plan Object parses incoming UGV mission plans from the vehicle. (Communicates directly with vehicle via its own socket connection, not through PIU). Plan UGV UGV Process … … UGV Platform Mobility UGV T/A Sensor (RSTA Class) RSTA Weapon Platform • UGV Object controls the virtual UGVs Interprets UGV Mission Plans Controls when data is passed to the UGV Platform Object. • UGV Platform Object starts a thread where in each functional object is instantiated. Passes data to each object. 16 ESS UV Process • Lesson Learned: UGV process performed very well. It was also an object based design instead of functional based design as was the earlier VTT code. • If possible, the second phase of the VTI will expand on this design philosophy with the inclusion of UAVs. Plan UGV UV Process … … UGV Platform Mobility UGV T/A Sensor … … UAV RSTA Weapon Platform • UGV process may become Unmanned Vehicle (UV) process and instantiate objects for all unmanned assets. • Future design efforts will utilize such methods as the Unified Modeling Language to identify, describe, and model system components and behavior 17 ESS RSTA Architecture • RSTA architecture utilized one RSTA server video channel to service multiple RSTA Clients. Caused severe lag when serving more than one RSTA client request. • Lesson Learned: Envisioned RSTA architecture may apply one video channel to each client. Uses more channels but increases performance. • Will investigate a trade of dedicated servers and clients versus available video channels. IG synchronization with Other channels RSTA Sim Server RSTA Sim Client UGV Sim 1 RSTA Sim Server RSTA Sim Client RSTA Sim Client UGV Sim 2 UGV Sim N Current RSTA Configuration RSTA Sim Client UGV Sim 1 RSTA Sim Server RSTA Sim Server RSTA Sim Client RSTA Sim Client UGV Sim 2 UGV Sim N Future RSTA Configuration 18 ESS Image Generation Architecture • Currently, each individual output channel is controlled through a master channel. This concept was carried over from the VTT where switching of the eye-point was very limited because of the single vehicle environment. • Lesson Learned: Moving to a distributed IG control approach would solve several problems (such as RSTA updates) and better supports the notion of a multi-crew station/vehicle architecture. • Will also perform trade study to look at other IG alternatives to X-IGTM. C2 Simulation C2 Simulation DIS / HLA I/F Perceived State DIS / HLA I/F Real State Perceived State Real State IG Manager Process “World” X-IGTM Slave 1 X-IGTM Master X-IGTM Slave 2 X-IGTM Slave N Current IG Configuration IG Channel 1 Manager Channel Function & Sync option IG Channel 2 Manager Channel Function & Sync option IG Channel 1 IG Channel N Manager Channel Function & Sync option IG Channel 2 IG Channel N Future IG Configuration 19 Scenario Development • Lengthy process involving a number of factors: • CAT and UCD employed one master scenario that was adjusted as needed for a particular experiment. Subject matter experts design the vignettes. A difficult task as no one has experience with an FCS soldier/robot team in combat. • Originally the intent was to develop a series of scenarios with each exercising a different facet of the FCS scout mission. Vignettes approved by the user community. SAF users convert vignettes into digital scenarios. Scenario – Live Maneuver Exercise Zone Recon Fort Bliss South 94 L D 1,2 R A 2 Lane s AA1 R A IG users run through the scenario from different points of view. • Dependent on digital terrain database. 17,1 8,19, 20, 21,2 2 R A 3,4,5 12,13,14,15 ,16 Blis s Meyer Range Rd 9,10,11 L D 6,7 • Lesson Learned: Scenario development is more time consuming than one would think. Enough time should be budgeted for the process. 20 Digital Terrain Database Development • No high-res data was available for Ft. Bliss Texas (desired DTED level 5). • Fly-overs had to be conducted to capture the elevation data and feature data for the test site. Company needs to be scheduled and flight time over the site authorized. Raw data then needs to be converted into DTED like data. • A trial and error method of finding the right mix of terrain resolution and rendering performance with the IG may need to be conducted. Ft Bliss digital terrain area was roughly 13km x 9km. At 1 meter postings, this is a large amount of data for the IG to render. For CAT, postings were put at 10 meters to get the desired performance. • Lessons Learned: Depending on the overall quality you desire (resolution of database, quality of feature data, performance in rendering), digital terrain database development can also take a large amount of time. At the core of the virtual world is the terrain database. Without you don’t have a simulation. Therefore, the appropriate amount of time should also be budgeted for this activity. 21 ESS Hardware Architecture • ESS Used Commercial-off-the-shelf (COTS) hardware. Each channel is a RacksaverTM 1U box containing a TyanTM dual processor (1.6 GHz) motherboard and a TI 4600 graphics card It also contains a National Instruments Field PointTM unit that can shut down the ESS if temperatures inside any of the boxes reach a programmable threshold level. • Overall, the hardware performed well. Some problems were the general size of the embedded box, coming in at approximately 24”w x 29”d x 17”h. Initial placement of the box in the vehicle was difficult. • Field Point unit control software was shutting down the ESS erroneously. • Lesson Learned: Will investigate reducing size of box. Hyper-threading is a possibility. This would allow microatx boards to be used. Dual video outputs from one graphics card. • Will performing trade study to investigate Non-COTS alternatives 22 Conclusion • The Embedded Simulation Team and DCS Corp. successfully developed an embedded simulation system to support both the CAT/RF program and the LSI’s UCD. Several Conclusions can be drawn from this effort. Properly defining system requirements will greatly reduce design and development time as well as produce a better product. Distributing services such as image generation and RSTA operations will increase performance. It also more closely models the distributed nature of a multi-vehicle, multi-crew station environment. Loosen dependencies on states and modes. Allow enough time for the development of scenarios and digital terrain databases as these can be time consuming tasks that have a high impact if not done properly. Continue to design and develop code using an object based approach and preferably using such practices as the Unified Modeling Language. 23