Maritime Autonomous Navigation in GPS Limited Environments

Transcription

1 Maritime Autonomous Navigation in GPS Limited Environments 29/06/2017 IIR/University of Portsmouth

2 GPS signal is unreliable Tamper Jam U.S. stealth UAV captured by Iranian government by means of GPS spoofing. Coverage area of the GPS jamming unit at 25m above ground level on maximum power of 1.58W ERP.(Image courtesy of DSTL) Pritt, Mark D., and Kevin J. LaTourette. "Aircraft navigation by means of image registration." Applied Imagery Pattern Recognition Workshop (AIPR): Sensing for Control and Augmentation, 2013 IEEE. IEEE, Grant, A., Williams, P., Ward, N., and Basker, S.: GPS jamming and the impact on maritime navigation, Journal of Navigation, 2009, 62, (02), pp We need a reliable positioning scheme for Unmanned Surface Vehicle! Page 2

3 Framework for Virtual GPS Extend Kalman filter based framework Average positioning deviation <35m under extreme weather condition. Page 3

4 Principles for virtual GPS DR N β v boat N β v boat N β v boat Trajectory of boat N β v boat φ 2, λ 2 vboat N β Rhumb line distance Start point with position coordinates φ 1, λ 1 v boat : the boat speed over ground φ 2 = asin( sin φ 1 cos δ + cos φ 1 sin δ cos θ ) λ 2 = λ 1 + atan2( sin θ sin δ cos φ 1, cos δ sin φ 1 sin φ 2 ) Where: φ is latitude, λ is longitude, θ is the bearing (clockwise from north), δ is the angular distance d/r; d being the distance travelled, R the earth s radius Using DVL/IMU/Compass as inputs; Providing the state predication information for boat Page 4

5 Test results of DR based positioning Ideal condition Choppy condition Deviation: Ideal condition: Choppy condition: 2400s; Disadvantage of DR based positioning is the cumulative error. Only suitable for short-term positioning. Page 5

6 Principles for virtual GPS Radar Challenge: Image registration with different viewing angle. Page 6

7 Principles for virtual GPS Radar Radar Deviation = 14.2m Google Earth Factors that influence the accuracy: Shape of the coastline; Quality of radar chart; Environments around the boat. Page 7

8 Test results of radar based positioning Page 8

9 Principles for virtual GPS Camera Landmarks are used to calculate the boat positions 0s 60s 0s 60s Using two pictures taken from a monocular camera at different locations Or using the image from a stereo camera. Page 9

10 Principles for virtual GPS Camera Water drop on camera Landmarks Process 1.Separating the sky and water 2. Noise Reduction 3. Label connected components 4. Possible Landmarks Page 10

11 Test results of camera based positioning Ideal condition Choppy condition Deviation: Ideal condition: 180m; Choppy condition: 420m; Only suitable for ideal sea condition Need to have clearly visible landmarks Page 11

12 Data fusion of Radar and DR results Kalman filter based data fusion method Can significantly reduce the random error of radar. Page 12

13 Test result of the Virtual GPS system Cases definition: As the right table Record Frequency Ship dynamics: 5s Radar: 60s/15s Cameras:60s/15s Cases definitions Ideal Choppy Foggy Choppy with tide Vehicle Mission Choppy + Foggy with tide Variable Case 1 Case 2 Case 3 Case 5 Case 6 Autonomy Level Full Full Full Full Full Max Speed 5 knots 5 knots 5 knots 5 knots 5 knots EO Camera On On On On On DVL speed On On On On On Radar On On On On On Depth sensor On On On On On MEMS On On On On On Wind speed reader On On On On On GPS On On On On On AIS Sensor On On On On On Mission Phase Transit Transit Transit Transit Transit Duration of Mission 1800s 1800s 1800s 2400s 2400s Start Point Lat-Lon Lat-Lon Lat-Lon Lat-Lon Lat-Lon Waypoints(s) Lat-Lon Lat-Lon Lat-Lon Lat-Lon Lat-Lon End Point Lat-Lon Lat-Lon Lat-Lon Lat-Lon Lat-Lon Day / Night Day Day Day Day Day Time of Day Sunrise Sunrise Sunrise Sunrise Sunrise Weather Clear Clear Foggy Clear Foggy Sea State Calm Choppy Calm Choppy Choppy Sea Obstacle Type Large Large Large Large Large Environment Wind Speed None Breezey None Breezey Breezey Wind Direction None S/W None S/W S/W Tide LAT LAT LAT Real Real Page 13

14 Test result of the Virtual GPS system Sea Conditions Average Deviation (m) Maximum Deviation(m) Case1 Ideal Case2 Choppy Case3 Foggy Case4 Choppy with tide Case5 Choppy + Foggy with tide Test results can meet the basic positioning function of unmanned surface vessel even in extreme condition. There is a big potential to improve the current positioning accuracy by: Enhance the signal synchronization between the sensors. The radar scanning range could be dynamically adjusted according to the distance between the ship and the coastline. The expected average error is less than 20m in extreme weather. Page 14

15 Thank You Q&A Aviation Systems Energy Efficiency Industrial Applications Marine Systems