Unmanned Aerial Vehicle Application to Coast Guard Search and Rescue Missions Allison Ryan July 22, 2004 The AINS Center for the Collaborative Control of Unmanned Vehicles 1
US Coast Guard Search and Rescue Scenario of Interest Coast Guard pilot flies helicopter in search pattern over assigned area Height and pattern determined by weather and search conditions Target must be visually detected by human Probability of survival depends on speed of search Extremely high operation and maintenance costs 2
Concept Animation Simulation of UAVs flying in formation with search and rescue helicopter 3
Motivation for UAV Participation Increased coverage allows faster search, higher chance of survival No increase to human risk because no extra pilots Marginal increase in cost (UAV: thousands of dollars helicopter: millions of dollars) Does not require changes to Coast Guard operating procedure 4
C3UV Search and Rescue Goals Develop a control and software platform to increase the efficiency of SAR by flying UAVs in formation with a US Coast Guard helicopter Control: formation flight controllers that ensures safety and search coverage Software: coordinate UAV control, sensor processing, and pilot interface UAV UAV control control Look Look for for target target Software platform comm comm Pilot Pilot input/ input/ display display 5
Specific Mission Requirements Function with commercial off the shelf UAV and autopilot systems Ensure safety of all aircraft and increase search efficiency while preserving accuracy Process data from UAV onboard sensors to find target in water and then respond accordingly Require no modification of Coast Guard helicopter and minimal modification of search procedure Aid helicopter pilot without adding to his or her responsibilities or distracting from search procedures 6
Challenges for New Research UAV control: force UAV formation to track helicopter although UAV flight dynamics are more limited Formation control: -maintain strict formation for sensor coverage during search -allow dynamic reconfiguration -ensure safety of all aircraft at all times Sensing: detect target in real-time sensor data from multiple UAVs Autonomy: use high-level hybrid control to manage a group of UAVs in various maneuvers with minimal human input 7
Proposed Development Framework Off the shelf UAVs carry commercial autopilot system with GPS Piccolo autopilot from CloudCap Technologies Sig Rascal UAV from ACR Control software runs on laptop, interfaces with autopilot ground station to send GPS waypoint commands to UAVs 8
Proposed Development Framework Off the shelf UAVs carry commercial autopilot system with GPS Control software runs on laptop, interfaces with autopilot ground station to send GPS waypoint commands to UAVs Groundstation 9
Areas of C3UV Research Related to Search and Rescue- formation flight Control of individual UAVs Formation flight control Computer vision processing UAV autonomy 10
Work Specific to Search and Rescue Dynamic formation spacing adjustment: Adjust formation spacing to ensure sensor coverage based on altitude Altitude step response improvement Suggest helicopter altitude change maneuver to allow better tracking from UAV Sensor coverage study Compare effectiveness of a number of methods to improve sensor coverage and search speed in spiral pattern 11
Formation Flight Control for SAR Formation control developed for truck following has been adapted for helicopter search and rescue: Lemniscate orbits adapt to helicopter forward speed, as for truck Altitude control tracks helicopter altitude Inter-aircraft spacing determined by altitude UAVs match helicopter s forward speed while maintaining their optimal air speed 12
Dynamic Spacing Adjustment Dynamic formation spacing adjustment has been implemented in hardware in the loop simulation θ x = 2* h* tan( ) 2 θ = view angle of sensor OR Height (h) Spacing (x) x = P θ 2* h*(1 )*tan( ) 100 2 θ x When P = desired percent overlap of neighboring UAV sensor fields h 13
Altitude Step Response Improvement UAV has poor response to step altitude change UAV altitude tracking, HIL simulation Suggest modified altitude change maneuver to allow better tracking from UAV Theoretical basis: L.E. Dubins optimal trajectories HIL Simulations courtesy of Anna Williams, Masters Thesis, UC Berkeley, 2004 14
Dubins Result and Application to UAVs Dubins result: optimal path for vehicle with minimum turning radius r min consists of connected arcs with radius r min r min r min Adaptation for UAV altitude change maneuver r min Maximum UAV climb/dive slope r min L.E. Dubins. On curves of minimal length with a constraint on average curvature and with prescribed initial and final positions and tangents. American Journal of Mathematics. 79: 497-516. 1957. 15
Result of Improved Altitude Change Maneuver Overshoot reduced from 25% to 10% HIL simulation result Similar settling time Minor change in helicopter maneuver greatly improves tracking Better tracking results in better formation stability and sensor coverage 16
Sensor Coverage Study for Spiral Search Patterns Rescuing a target in a search area depends on both search speed and completeness of coverage Sensor coverage from helicopter and 2 UAVs UAV flight dynamics cause gaps in sensor coverage during sharp turns All sensor coverage results from HIL simulation of altitudeadapting formation controller 17
Metrics for Search Performance Coverage factor (c): Ratio of sensor sweep width to track width Probability of detection (POD): Ratio of imaged area to total search area Search rate: square kilometers per minute Standard expanding spiral with c = 1.1 has POD = 0.93 and search rate = 0.99 square km per minute Courtesy of USCG SAR Manual 18
Techniques for Improving Sensor Coverage increase coverage factor Increase coverage factor Round corners of spiral Sensor coverage with increased coverage factor Change to round spiral C = 1.4 POD = 0.96 Rate = 0.78 km/min 19
Techniques for Improving Sensor Coverage- round corners Increase coverage factor Round corners of spiral Sensor coverage with rounded corners Change to round spiral C = 1.1 POD = 0.9 Rate = 1 km/min 20
Techniques for Improving Sensor Coverage- round spiral Increase coverage factor Round corners of spiral Sensor coverage with round spiral pattern Change to round spiral C = 1.1 POD = 0.92 Rate = 1.4 km/min 21
Comparison of Techniques for Search Coverage Improvement Expanding square Coverage factor 1.1 Probability of detection 0.93 Search rate (square km / min) 0.99 Further study of varied search patterns is called for Expanding square with round corners 1.1 0.9 1.0 May be investigated in conjunction with improved lateral controller High coverage expanding square 1.4 0.96 0.78 Circular spiral 1.1 0.92 1.4 22
Future Work Lateral control to improve corner turning in expanding square-type search pattern Formation reconfiguration control to safely add or remove UAVs from formation Supervising controller for mode switching, link with vision processing, and user interface 23
Related Publications S. Spry, A. Vaughn, X. Xiao, and J.K. Hedrick. A Vehicle Following Methodology for UAV Formations. Proceedings of the fourth International Conference on Cooperative Control and Optimization, Destin, FL., Nov. 2003. Anna Williams, Search and rescue augmentation using unmanned aircraft, Masters Thesis, UC Berkeley, 2004. A. Ryan, M. Zennaro, A.S. Howell, R. Sengupta, and J.K. Hedrick, An Overview of Emerging Results in Cooperative UAV Control, to appear in Proceedings of the IEEE Conference on Decision and Control, Dec. 2004. A. Ryan, A. Williams, S. Spry, and J.K. Hedrick, Cooperative control of UAVs for Coast Guard Search and Rescue Missions, in progress. 24
References United States National Search and Rescue Plan U. S. Coast Guard Addendum to the National SAR Supplement L.E. Dubins. On curves of minimal length with a constraint on average curvature and with prescribed initial and final positions and tangents. American Journal of Mathematics. 79: 497-516. 1957. 25
Questions? 26