State estimation with fleeting data

State estimation with fleeting data

18/04/23 2

> Plan

1. Introduction

2. Approach used: fleeting data and tubes

3. Demo

4. Conclusion

State estimation with fleeting data

18/04/23 3

Introduction

State estimation with fleeting data

18/04/23 4

Context: offline SLAM for submarine robots using interval arithmetic and constraint propagation (without outliers)

Introduction

State estimation with fleeting data

18/04/23 5

Redermor and Daurade, submarine robots of the GESMA

Introduction

State estimation with fleeting data

18/04/23 6

Sensors of the submarines :– GPS (position at the surface)– DVL (speed and altitude)– IMU (Euler angles and rotating speeds)– Pressure sensor (depth)– Lateral sonar

Introduction

State estimation with fleeting data

18/04/23 7

Introduction

State estimation with fleeting data

18/04/23 8

Experiments with marks in the sea

Introduction

State estimation with fleeting data

18/04/23 9

Goals:– Get an envelope for the trajectory of the robot– Compute sets which contain marks– Make a cartography of the area

Introduction

State estimation with fleeting data

18/04/23 10

Introduction

Input:– Navigation data of the submarine (Euler angles,

depth, altitude, speed, some GPS positions) – Marks detections on the sonar image (distance)– Raw sonar image– Time and max error for each data

State estimation with fleeting data

18/04/23 11

Introduction

Output:– Trajectory (envelope and center)– Position of marks in the sea (envelope and center)– Map of the area

State estimation with fleeting data

18/04/23 12

What has already been done:– Compute an envelope for the trajectory of the robot– Compute sets which contain marks (detected by a

human operator by scrolling the sonar image)

Introduction

State estimation with fleeting data

18/04/23 13

What has already been done:– Generate a reconstitution of the sonar image showing

the estimated position of the marks on it (predicted stain)

– Help the human operator for the detection/identification of the marks

– Check the consistency of input data

Introduction

State estimation with fleeting data

18/04/23 14

Introduction

State estimation with fleeting data

18/04/23 15

What could be done:– Use a little bit more the sonar image to eliminate

zones where we are sure there is no object (no bright points)

• Should improve the precision of the positions estimation• Could help the user to see which parts of the sonar data

might contain objects

Introduction

State estimation with fleeting data

18/04/23 16

Introduction

State estimation with fleeting data

18/04/23 17

What could be done:– Use a little bit more the sonar image to eliminate

zones where we are sure there is no object (no bright points)

• Should improve the precision of the positions estimation• Could help the user to see which parts of the sonar data

might contain objects

– Generate a cartography of the area using sonar data

Introduction

State estimation with fleeting data

18/04/23 18

Approach used: fleeting data and tubes

State estimation with fleeting data

18/04/23 19

Demo

State estimation with fleeting data

18/04/23 20

Conclusion

State estimation with fleeting data

18/04/23 21

Conclusion

We can use the sonar/telemetric data to improve the estimation of the trajectory

Tubes are well suited to deal with estimation of trajectories

State estimation with fleeting data

18/04/23 22

Conclusion

Prospects : – Do SLAM instead of localization– Apply the method on real sonar data of submarines