ISTAR Concepts & Solutions

Transcription

1 ISTAR Concepts & Solutions CDE Call Presentation Cardiff, 8 th September 2011

2 Today s Brief Introduction to the programme The opportunities ISTAR challenges The context Requirements for Novel Integrated ISTAR Solutions The questions

3 ISTAR Concepts & Solutions (ICS) MOD research programme Dstl Programme Office - ISTAR & Sensors Domain New 5-year research programme Collaboration between UK Government, Industry & Academia and allies Focus on short, medium and long term capability solutions Intention to run a yearly CDE themed call Exploitation roadmaps established across the domain

4 ISTAR Concepts & Solutions Identify & Design Test & Evaluate Exploit & Prioritise Systems Concepts Technology Concepts Experimentation Modelling & simulation Validated Solutions Research Requirements Quick wins

5 Industry engagement (to March 2012) I&S Domain Industry Day (invitation) IET, Savoy Place, London, 20 th Sept Prime contractor support (pilot year only) Quick start Embedded team, experimentation & infrastructure FATS competition Technology Concepts CDE Call ( Novel Integrated ISTAR Solutions ) Today s Briefing (Cardiff on 8 th September)

6

7 Enduring Framework ( ) Current pilot year is intended to be 1 st of 5 Establish way of working, quick start Collaborative framework to cover years 2-5 Strategy under development to cover: Industry construct Advertise, compete and establish by April 2012 Consortium supply anticipated

8 Summary Concept design Assessment Experimentation Applied R&D Prioritisation & context Coherence & coordination Cross-DLOD implications Exploitation through applications areas

9 ISTAR Concepts and Solutions Context & challenges

10 Military Operations Revolution in Threat No longer characterised by physical properties (size, shape, velocity, etc) Now characterised by intent and activity No longer geographically clustered Now disparate, sparse and hidden Attacks are no longer coordinated and focussed Now seemingly random with high intermittency Response to the threat Needs to easily adapt to the unplanned or unexpected Constrained by poor / incomplete / intermittent information and environment conditions Typically operates with limited connectivity / cooperation between disadvantaged users

11 ISTAR Intelligence, Surveillance, Target Acquisition and Reconnaissance The Decision Cycle Direct Disseminate Collect Process

12 ISTAR Intelligence, Surveillance, Target Acquisition and Reconnaissance In Practice: Massive Complexity Multiple simultaneous Decision Cycles operating Multiple interrelated and parallel decisions taking place Need to maximise: Situational Awareness Timeliness, accuracy and flexibility of decision making Ability to operate under uncertainty, information loss and contradiction Integrated ISTAR has the potential for: More rapid traversal of DCPD loops Maintaining Decision cycle performance with minimal operator input Less reliance on pre-programmed plans Reduction in Information overload

13 Challenges in the ISTAR area To demonstrate remote and autonomous sensing and collecting capabilities; To demonstrate association and combination of multiple off-the-shelf (OTS) sensor modalities to improve detection, tracking, recognition and identification of targets; To demonstrate autonomous (or with minimal analyst intervention) cross-modality cueing of sensors and information sources; To demonstrate techniques and algorithms which enable inference and decision making across information from multiple platforms, between hard and soft information sources, over multiple ISTAR layers with multiple information collection times; To demonstrate improved endurance/persistence/coverage with smaller, lighter, cheaper and lower-power sensors

14 Tranche 1: Novel Integrated ISTAR Solutions The requirement

15 What do we want? Innovative solutions to ISTAR challenges Technology Readiness Level 3 5 Break point with demonstration in Feb 2012 Ideally (loosely) integrated with February experiment Low cost technology (e.g. off the shelf sensors) What do we not want? Market surveys Requirements capture 3 year PhDs Assessed against: Builds Critical Technical Capability to meet UK Needs, Scientific Quality / Innovation Military relevance, Likelihood of Exploitation Science, Innovation and Technology Risk

16 Topics of interest 1. Autonomous Sensor and Information Cueing 2. Accurate, automatic 3D map generation 3. Automatic (assisted) target recognition a) Multi-mode sensor approach to address difficult targets (vehicles) b) Assisted recognition of people and gestures in urban scenario 4. Autonomy for Tactical Situational Awareness

17 1. Autonomous Sensor and Information Cueing Smart use of sensors to optimise coverage (e.g. switch to sensors that have a better view of a new detection) To reduce cognitive burden, To reducing information overload For information prioritisation To provide the correct information for the current task There are at least 3 key elements: 1. A local tactical picture a single construct that harmonises and contains all geospatial information, blue and red force positions, etc Track formation and harmonisation from multiple sensor detections (different modalities) Sparse temporal sampling (due to platform revisit or obscuration) is an issue, therefore there is a need for tagging/id

18 1. Autonomous Sensor and Information Cueing 2. Sensor/platform management Knowledge of sensor positions and capabilities/errors Sensor position inaccuracy Sensor pointing inaccuracy Decision making and optimisation to assign sensors to tasks Need for ATR to know what sensor to use Need to understand environment/weather/clutter/range/target characteristics, and how they relate to sensors available Need to interact with human command chain, if using soldier as sensor, soldier is best ATI asset 3. Information management (e.g. prioritised threat warning, info appropriate to task) ATR to guide decision making

19 2. Accurate, automatic 3D map generation From airborne imagery or Ground-based (eg 3D street-view) To support: Training, environmental familiarisation Mission planning and rehearsal Identification of points of access and egress Mission support line of sight analysis 3D change detection for threat detection and battle damage assessment Image matching for geo-location and navigation Enable advanced sensor operation in complex environments by enabling accurate beam propagation modelling Electronic surveillance, imaging radar, hyper-spectral imaging (shadow prediction), acoustic

20 2. Accurate, automatic 3D map generation Sensor mix and collection fidelity? Structure from video, lidar, Inertial Navigation System (INS) requirements, 3D burst illumination (BIL) imaging Structure from video offers the simplest sensor solution but also the most complex processing problem. Is structure from video sufficiently accurate, can it be automatic? Vehicle mounted sensor system Robust no moving parts Cameras + INS + fan lidar? Exploitation of current camera systems (LSAS) 3D model generation from rapidly changing perspective Automatic removal of movable objects (vehicles & people)

21 2. Accurate, automatic 3D map generation Requirement Software package (executable) delivered to MOD for assessment Input Airborne video stream INS and pointing data 2D map Output 3D point cloud, polygon model, rendered imagery. Exploitation of previous research (SEAS DTC) Automatic alignment and integration of airborne and ground based 3D imagery Efficient processing, compression and storage techniques.

22 3. Automatic (assisted) target recognition Vehicles and people To support Reduction on workload enhanced search Tracking and target reacquisition Removing mobile targets from 3D map generation Air-to-ground and ground-to-ground target identification

23 3. Automatic (assisted) target recognition Vehicles: Current algorithms can be confused by Camouflage Shadows Clutter Obscuration Articulation Training data Can be made available where appropriate

24 3a. Multi-mode sensor approach to address difficult targets (vehicles) Opportunity to combine (e.g.): Visible (colour) Thermal BIL Silhouette BIL 3D Requirement: New ATR concepts/algorithms Algorithms delivered in a format that can be tested against standard datasets Exploitation of multi-mode imaging visible/ infrared and active imaging 3D/silhouette.

25 3b. Assisted detection & recognition of people and gestures in urban scenario Sensors mounted on mobile platforms Combine (e.g.) face, gait, shape features Identify individuals or reacquire targets from their known signature. Exploitation of 3D imaging (e.g. X- box kinect ) sensors Simple (robust) person detector

26 4. Autonomy for Tactical Situational Awareness To reduce the need for detailed intensive supervision of ISTAR assets Sensor/platform management system Natural language tasking e.g. search area, examine specific location, support specific activity Plan platform position/trajectory for optimum sensor coverage Process sensor info to extract object info (ATR/ATI, track, count, ID, etc) Feed local tactical picture Re-plan, repeat Using knowledge of environment/sensor/target location to plan waypoints/trajectory

27 4. Autonomy for Tactical Situational Awareness Control of sensor modality to match object of interest E.g. radar GMTI selected to track moving objects System should respond to current activity Support patrol by suggesting surveillance plan to commander When patrol is static choose a perimeter surveillance mode When hostile fire is reported, investigate location Output should be information not data (e.g. video that requires analysis) This call is not about auto-pilots or another UAV, it is about a whole system approach

28 Summary Timescales Call closes: 27 th September 2011 (at 12:00 noon) Successful proposals (expected) on contract: End October 2011 Demonstration event: February 2012 Contract ends: 31 st March 2012

29 ISTAR Concepts & Solutions CDE Call Presentation, Cardiff, 8 th September 2011 Proposals for funding must be submitted by 1200 (noon) 27 th September 2011 using the Centre for Defence Enterprise Portal ( All proposals must be clearly marked Novel Integrated ISTAR as a prefix in the title. Queries relating to the technical aspects of the call should be addressed to DstlSensors@dstl.gov.uk Queries relating to the submission process (including how to use the Portal) should be addressed directly to the CDE at science-enterprise@mod.uk or by phone on