Indoor Follow Me Drone

Transcription

1 Indoor Follow Me Drone Wenguang Mao, Zaiwei Zhang, Lili Qiu, Jian He, Yuchen Cui, Sangki Yun Presented By Stephen Xia for Columbia University ELEN 9705 Spring 2018 Instructor: Prof. Xiaofan (Fred) Jiang January 24, 2018 Proceedings of the 15 th Annual Internation Conference on Mobile Systems, Applications, and Services MobiSys 2017, Niagara Falls, NY, USA June 19 th 23 rd, 2017

2 Motivation Follow Me Drones Problem: There is no robust and efficient mobile solution for indoor follow me

3 Components of Follow Me User Tracking Control Drone Control User movement Prediction

4 Components of Follow Me User Tracking Control Drone Control User movement Prediction

5 Current Methods for User Tracking

6 GPS Current Methods for User Tracking

7 Current Methods for User Tracking GPS Computer Vision

8 Current Methods for User Tracking GPS Computer Vision RF

9 Prior Works for User Tracking GPS Computer Vision RF Acoustics

10 Limitations in Indoor Scenarios

11 Limitations in Indoor Scenarios GPS Availability

12 Limitations in Indoor Scenarios GPS Computer Vision Availability Computation and Accuracy

13 Limitations in Indoor Scenarios GPS Computer Vision RF Availability Computation and Accuracy Hardware

14 Limitations in Indoor Scenarios GPS Computer Vision RF Acoustics Availability Computation and Accuracy Hardware

15 Acoustic Tracking Distance Orientation and Angle

16 Acoustic Tracking Prior Art Phase Shift Methods (e.g. Doppler Shift) Angle of Arrival and Time Difference of Arrival (e.g. Cross Correlation) *

17 Acoustic Tracking Prior Art Frequency Modulated Carrier Wave (FMCW) (e.g. Doppler Shift) MUltiple Signal Classification (MUSIC) *

18 Acoustic Tracking Prior Art Frequency Modulated Carrier Wave MUltiple Signal Classification (MUSIC)

19 Acoustic Tracking Distance Orientation and Angle

20 Acoustic Tracking Distance Orientation and Angle

21 Frequency Modulated Carrier Wave

22 Improving Robustness of FMCW - MUSIC

23 Improving Robustness of FMCW - Other Filtering the MUSIC spectrum

24 Improving Robustness of FMCW - Other Frequency Compensation and Signal Chopping Kalman Filter

25 Rabbit

26 Acoustic Tracking Distance Orientation and Angle

27 Acoustic Tracking Angle

28 Components of Follow Me User Tracking Control Drone Control User movement Prediction

29 Components of Follow Me User Tracking Control Drone Control User movement Prediction

30 Prior Work on Control

31 Prior Work on Control Manipulated Variables: Pitch Yaw Roll Controlled Variables: Distance to user Orientation to user Lateral Velocity

32 Prior Work on Control Controller Types: Model Predictive Control PID Vector Field Control Many more.

33 System Control Sensing: Rabbit Robust Acoustic-Based Indoor Tracking

34 System Control Manipulated Variables: Pitch Yaw Roll Controlled Variables: Distance to user Orientation to user Lateral Velocity

35 Components of Follow Me User Tracking Control Drone Control User movement Prediction

36 User Movement Prediction Controller: Model Predictive Control Sensing Lateral Velocity Doppler Shift

37 Full System

38 Full System

39 Evaluation

40 Evaluation - Rabbit

41 Evaluation Full System

42 Evaluation Full System

43 Evaluation Real World Trajectory

44 Evaluation Resource Usage and Delay Power: 3% per 10 minutes on a Galaxy S7

45 Discussion