TARDEC Robotics Update to the Joint Robotics Program

Document Sample
TARDEC Robotics Update to the Joint Robotics Program Powered By Docstoc
					  TARDEC Robotics Update
to the Joint Robotics Program
       Curt Adams (810) 574-6160
            Associate Director
            Vetronics Technology Area

U.S. Army Tank-Automotive RD&E Center (TARDEC)
             Vetronics Technology Area
           (AMSTA-TR-R, Mailstop 264)
              Warren, MI 48397-5000

             8 November 2001
      Tank-automotive & Armaments COMmand

 • Vetronics Technology Integration (VTI) Contract
 • Crew integration & Automation Testbed (CAT) ATD
 • Robotic Follower ATD
 • RDEC Federation CalEx Experiment
 • Intelligent Mobility Omni-Direction Inspection System

Vetronics Technology Area
Mission: To conduct research in the Vetronics technology areas of
  crew stations, electronics architecture, embedded simulation and
  robotics while leveraging advanced automotive technology to
  provide our soldiers with the world’s most advanced ground
  vehicle systems and logistics support equipment.
                                    General Shinseki and MG (P) Caldwell Visit

       Crew Stations                                                             Embedded Simulation

• Cognitive Aids                                                                    • Mission Planning
• Route Planning                                                                    • Mission Training
• Auto Driving                                                                      • Battlefield Visualization
• 3-D Audio
• Speech Recognition
• Indirect Vision Driving

          Robotics                                                               Electronics Architecture

• Semiautonomous
Perception                       Vetronics Technology Area

                                                                                                                         Improved hardware and software reusability
• Soldier-Robot Interface
• Intelligent Situational
Behavior                         “Intelligent Systems for the
• Leader-Follower Technology           Objective Force”
                                                                                       Reconfigurable component

                                Unmanned Systems
                                                                                        based Software Ref Arch

                                Warfighter Interfaces
                                Warfighter Decision Aids

                                                                                     Open Interface based Sys Ref Arch
Vetronics Technology Integration Objectives

                       • Advance ground vehicle
                         “cockpit” and robotic follower
                         technology state-of-the-art
                       • Develop, integrate and test
                         CAT and RF ATD programs
                       • Provide technology risk
                         mitigation to FCS
                       • Provide technology readiness
                         to FCS Milestone B and
                         Block Upgrade Schedule
                         (Feb03 and Feb06)
                     VTI Contractor Team
Team Member                       Focus Areas
•General Dynamics Land Systems -Program Management,
                                      Systems Integration, CAT Lead
•General Dynamics Robotic Systems -Technology Insertion, RF Lead,
                                      Field Experiments
•Applied Systems Intelligence     -Intelligent Control Architectures,
                                      Decision Aids
•Micro Analysis and Design        -Human Performance Modeling
•Carnegie Mellon University       -Geometric Planning, Map
                                      Registration, Road Following
•Oasis Advanced Engineering-Embedded Training Systems, SMI,
                               Modeling and Simulation
•General Motors Defense           -CAT and RF Platforms (IAV
                                      Infantry Carrier)
Vetronics Technology Testbed Crewstation
  – Crew Station Design and SMI
  – Speech Recognition
  – Indirect Vision
• ARL’s Demo III
  – Hardware Architecture
  – Software Architecture
  – Autonomous Mobility
  – Symbolic Planning and Agent Based Associate - ASI
  – Geometric Based Planning and Map Registration -
  – Obstacle Avoidance
  – Communication and Navigation

• Two Ways To Improve Indirect Driving
     • Driver Cued
     • Tele-reflexive
                       Robotic Follower ATD
                              (STO III.GC.2000.04 )

                                                      Pacing Technologies:

                                                      Intelligent Situational

Mature & Demonstrate Robotics                           Leader-Follower
Technology Required for Early
     Insertion into FCS
                                               Affordability Metrics
          Solution Approach                    Total Sensor Cost <$370k
• Manned leader “proofs” path to reduce
perception & intelligence requirements         Technology Protection Plan
• Rapidly mature & integrate perception        Completed April 2001.
technology to enable higher speed &            Modeling and Simulation
enhanced decision making capabilities          • April 2003 M&S Demonstration of end ATD
• Successively demonstrate maturing            Exit Criteria
capability for FCS                             • Integrates mobility, sensor and terrain
                     Robotic Follower Systems Solution
• H/W & S/W (obstacle avoidance algorithms and AM sensors) design based on Demo III
• A separate LADAR at the rear for safe back-up maneuvers.
• In addition to sending back GPS waypoints, leader will use its onboard sensors to
collect and send back higher resolution terrain data (1 meter or less) to RF.
• Sensor terrain data will be registered to coarser onboard DTED map (4-10 meters) using
advanced map registration techniques.

 Interim Armored Vehicle (IAV) LAV III
  infantry carrier for RF platform.

               Demo III world model generated from LADAR sensor.

                                                  The Digital 14-
                                                   bit output of
                                                    the Indigo
                                                 Phoenix Imager
                                                    will greatly
                                                   increase the
                                                 and robustness
                                                   of the stereo

                                                 GDRS’s 77GHz
GDRS LADAR Generates High Fidelity 3-D Terrain   Radar Scans 90
   Elevation Information at 30 Frames/sec           degrees 10
                                                  times/sec in 1
                                                   degree steps

                   The Sony DXC-
                    390 camera,
                   with a 640x480
                     pixel array
                     ON CAT AND RF

The geometric
planner receives
information from
multiple sources
including the
onboard sensors
and external map                        Digital map     Local elevation map
data. All the data             DTED                   from on-board sensors
must be
registered to
ensure integrity
of the internal
                      Features from                   Obstacle map from
                     on-board sensors   Geometric     on-board sensors

• Terrain
• Terrain
• Landmarks
                 3-D REGISTRATION

Example of registration
of map using elevation
information: A low-
resolution DTED (upper
right) is registered with a
local, high-resolution,
elevation map from on-
board range sensing
(upper left) to produce
an updated map
                       ROAD DETECTION

Typical result of road tracking on              Typical result of road detection on highways.
unimproved road. Edges are tracked by           The algorithms can cycle at frame rate on
looking for transitions from road to non-road   conventional hardware
terrain type.

            Field of Interest is Used to
            Minimize Processing
            Requirements and Maximize
            Visual Servoloop Rate

              Feature Classifier Locates
              Lead Vehicle Within the
Data Collection & Playback
     Robotic System Simulation

  High Fidelity
Terrain Database

                          Imagery **         Mobility
Sensor Models

                                         Mobility Commands
         Vehicle Mobility Information

        * * Line   Level Equivalence to Real Sensors
                                                ATD Exit Criteria
                         Sp eed o n             Sp eed X -                                                                    Ob st acle
                         Pr imar y R o ad -     C o unt r y -                       M ax T ime          Sep ar at io n -      D et ect io n       -
M et r ic                ( kp h)                ( kp h)           R ang e - ( km)   D elay - ( hr s)    ( m)                  ( m)
D ef init io n           Sustained speed on     Open & rolling,   Distance          Time between        Distance between      Size of non-
                         paved or improved      highly            follower can      lead vehicle and    the lead and          engineered or
                         road with firm base.   trafficable for   travel using      follower vehicles   following vehicles,   camouflaged
                         Followers to stay in   equivalent        onboard           crossing same       dependent on          obstacles system
                         proper lane starting   manned system.    intelligence.     piece of terrain    communication         can detect.
                         in 2003.                                                                       range and latency.
C ur r ent                                                                                                    M in: 50          Positive: .5
( D emo IIIb )                   30                    15               160                 1                M ax: 500        Negative: 1x2x2
     A p r il, 2 0 0 3                                                                                       M in: 20           Positive: .3
    ( X U V chassis)             55                    30               160                12               M ax: 2 km        Negative: 1x2x2
                                                                                                             M in: 10           Positive: .3
    EN D   M inimum              65                    30               160                24               M ax: 5 km        Negative: 1x2x2
    ATD                                                                                                      M in: 1            Positive: .3
              Go al              100                   65               750                24             M ax: 200 km        Negative: 1x1x1

    Affordability Metric: Specific efforts to reduce sensor cost are not part of this ATD. While reduction in cost for
    computing capabilities is expected, additional computing capabilities will be required to meet goals. This ATD
    will strive to keep the total cost of the autonomous mobility suite at or lower than the current Demo III cost of
1 Difference between achieved performance in 2003 and End ATD will be demonstrated via modeling &
2 Parameters for obstacle detection:

   Positive obstacles Height above ground plane
   Negative obstacles Depth, Width, and Span in direction of travel (expressed as DxWxS)
              TRL Milestone Chart Accelerated Robotic Follower ATD
                     FY01                       FY02                   FY03                FY04                     FY05

                TRL= 3                           TRL= 6                         TRL= 6                           TRL= 6
Semi-           Demo III Baseline Perception     Road Following and Collision   Improve Obstacle Detection       Improve Obstacle Detection
                METRIC                           Detection (Pedestrian &        Algorithms; subset of Demo       Algorithms
autonomous      •Obstacle Detection              Vehicles)                      III Sensor Suite.                METRIC
Perception      Positive        .5 m (H)         METRIC Lane maintenance        METRIC Obstacle Detection        •Obstacle Detection
                Negative        1 m (W)          Lateral error    0.1 m max     Positive          .3 m (H)       Positive       .3 m (H)
                                                 •Collision avoidance 100%      Negative          .5 m (W)       Negative       .5 m (W)
                                                TRL= 6                          TRL= 6                           TRL= 6
Soldier-                                                                        Dismounted controller            Vehicle Interface Testing
                                                CAT Interface Testing
Robot                                           METRIC                          METRIC                           METRIC
Interface                                       •Workload/vehicle - 50%         •Workload/vehicle - 50%          •Workload/vehicle - 50%
                                                Reduction over Demo III         Reduction over Demo III          Reduction over Demo III

               TRL= 3                            TRL= 6                         TRL= 6                           TRL= 6
               Demo III Baseline Intelligence    Incorporate onboard terrain    Incorporate onboard terrain      Intelligent Situational Behavior
Intelligent    METRIC                            database                       database                         METRIC
Situational    •Planning Capabilities - Plan     METRIC                         METRIC                           •Planning Capabilities - Plan to
Behavior       around 5 m W obstacles using      •Limited planning capability   •Planning Capabilities - Plan    prevent communication loss or
               onboard perception only           since LOS operation only       around 10 m W obstacles using    mobility kill.
               •6 Operator Interventions/km      •1 Operator Intervention/km    onboard database                 •1 Operator Intervention/km
                                                                                •1 Operator Intervention/km

              TRL= 3                             TRL= 6                         TRL= 6                           TRL= 6
              XUV Follower Demo w/GPS            On-road LOS convoying          •Dismounted follower using       •Improved Mobility Follower
Leader-       Waypoints                          METRIC                         waypoints augmented with         •Waypoints Augmented with
Follower      METRIC                             •Speed                         terrain intelligent navigation   Terrain Intelligent Navigation
              •Speed                             On road       65 kph           METRIC                           METRIC
Technology    On road       30 kph               X-Country     N/A              •Speed                           •Speed
              X-Country     15 kph               •Separation   20 - 100 m       On road            0-20 kph      On road           65 kph
              •Separation   500 m                                               X-Country          0-20 kph      X-Country         40 kph
                                                                                •Separation        5 m - 1 km    •Separation        5 km
   Description of CalEx Simulation
• The Vehicle Dynamics Mobility Server (VDMS)
  is a high resolution model that using the High
  Level Architecture (HLA) to communicate, can
  portray robotics vehicles. This allows for soldier-
  in-the-loop experimentation using a Robotics
  Operator Control Unit (OCU).
• Such variables as payload and tire pressure can be
  modified and changed to help determine the
  optimum design.
• This tool can be used to evaluate contractors
  concepts against one another in military
  significant exercises in the support of FCSS.
 RDEC Federation CalEx Data Results
• The RDEC Federation ran a road march scenario
  with nine robotics vehicles with varying payloads
  and tire pressures on a Bosnia database. The
  following was determined:
   – Payloads: An increase in payload caused the vehicles
     to slow down as expected.
   – Tire Pressures: An increase in tire pressure did not
     represent an increase in speed. TARDEC vehicle
     dynamics engineers concluded that the higher pressures
     caused larger vibrations and deviations from the ideal
     course. This in turn may cause the vehicle to be a little
     tougher to control and it may actually travel a larger
     distance overall.
   – Recommendation for this vehicle on this type of terrain
     would be to use 30 lbs. of tire pressure.
Omni-Directional Inspection System (ODIS)

   Funded by the JRP Man-Packable Robotic Systems
                   (MPRS) program

              AUVSI Demonstration
        Under vehicle Inspection using the ODIS Robot
              POC – Grant Gerhart/TARDEC

ODIS Robot – Current
•   Low profile platform fits under vehicles
    – man-packable, weighs less than 50lbs.
•   Replaces “mirror on a stick” inspection
•   2-3 hour run time with “hot-swappable”
    batteries – 4 mph top speed
•   Tele-operated video inspection with
    active LED lighting
•   Com link transmits real time video data
    to base station operator
•   Uses omni-directional drive technology
    for high mobility/maneuverability                    Assemblies
                                                                              Pan/Tilt Camera
•   Inspects vehicle underbodies for              Battery
    bombs, contraband, etc.                       Packs

•   Demonstrated and tested at AUVSI, Ft.
    Leonard Wood and TARDEC
•   Three ODIS robots available for User                                      Vetronics
                                               Sonar, IR, and Laser Sensors
    testing in January 2002
        Under vehicle Inspection using the ODIS Robot
              POC – Grant Gerhart/TARDEC

ODIS Robot – Future
•   Advanced sensor suite including video,
    thermal, acoustic and chem/bio sensors
•   OCU will feature fully autonomous and
    tele-operational behavior
•   OCU will use wearable or hand held
    computers for imaging & navigation
•   Operate on military & law enforcement
    communication frequencies
•   Marsupial deployment and improved
•   Image enhancement and object
    recognition – license plate reading
•   Extensive testing at the 2002 Winter
    Olympics and Hill AFB
Under vehicle Inspection using the ODIS Robot
      POC – Grant Gerhart/TARDEC

Shared By: