RIO: A Pervasive RFID-based Touch Gesture Interface Swadhin Pradhan - PowerPoint PPT Presentation

RIO: A Pervasive RFID-based Touch Gesture Interface Swadhin Pradhan 1 , Eugene Chai 2 , Karthik Sundaresan 2 , Lili Qiu 1 , Amir Khojastepour 2 , Sampath Rangarajan 2 2 1 1 MobiCom 2017

LOTS of Connected Devices by 2020 2

LOTS of Connected Devices by 2020 ~30 Billion* 2 *Ericsson Research

Smart Spaces 3

Smart Spaces Smart Retail Smart Hospitals Smart Offices 3

4

Can we effectively transform everyday object into a touch Interface ? 4

5

Wireless Capacitive IMU Vision 5

Issues with these modalities 6

Issues with these modalities • Not easy to customize • Not low cost • Extensive training phase • Need to be recharged frequently … 6

Passive RFID tags as touch interface 7

Passive RFID tags as touch interface COTS Tags COTS Reader/Antenna Custom Tags 7

And that interface … 8

And that interface … • Customizable and less costly • Battery-free • Supports fine-grained tracking • Multi interface support 8

9

Button 9

Keyboard Button 9

Slider Keyboard Button 9

Slider Keyboard Button Dial 9

Slider Keyboard Button Multiple Dial Sliders 9

Slider Keyboard Button Multiple Shapes Dial Sliders 9

Key Idea and Reasoning Touch and Single Tag Tracking Tracking within Multiple Tags Evaluation 10

Key Idea and Reasoning Touch and Single Tag Tracking Tracking within Multiple Tags Evaluation 11

Antenna Tag Reader

Predictable phase change at reader side Antenna Tag Finger Swipe on Tag Reader

Why does it happen ? 13

Why does it happen ? 13

Why does it happen ? ZAntenna ZChip 13

Why does it happen ? ZAntenna ZChip ZBody 13

Why does it happen ? ZBody 13

Why does it happen ? Backscattered Electric field and Phase ZBody 13

VNA based measurements 14

VNA based measurements 14

VNA based measurements 15

VNA based measurements 15

VNA based measurements 15

Key Idea and Reasoning Touch and Single Tag Tracking Tracking within Multiple Tags Evaluation 16

How to detect touch ? 17

How to detect touch ? 17

How to detect touch ? Sudden Jump in Phase 17

How to detect tracking ? 18

How to detect tracking ? 18

How to detect tracking ? Predictable Phase Pattern 18

This change is also present in NLoS 19

This change is also present in NLoS Behind a Door Behind a Wall 19

This change is also present in NLoS Remains Similar Behind a Door Behind a Wall 19

Single tag tracking algorithm sketch 20

Single tag tracking algorithm sketch Phase Tag Position 20

Single tag tracking algorithm sketch Phase Calibration Phase Tag Position 20

Single tag tracking algorithm sketch Phase Calibration Phase Last Known Position 20

Single tag tracking algorithm sketch T-second Swipe Phase Phase Calibration Phase Last Known Position 20

Single tag tracking algorithm sketch T-second Swipe Phase Phase Calibration Phase Last Known Position 20

Single tag tracking algorithm sketch T-second Swipe Phase Phase Calibration Phase DTW Matching Potential New Position Last Known Position 20

Key Idea and Reasoning Touch and Single Tag Tracking Tracking within Multiple Tags Evaluation 21

Tracking within Multiple Tags 22

Tracking within Multiple Tags 22

Mutual coupling phenomenon 23

Mutual coupling phenomenon Neighboring tags show reverse phase pattern 23

Why does it happen ? 24

Why does it happen ? 24

Why does it happen ? Coupled mutual impedance has reverse impact in back-scattered phase 24

Multi tag tracking algorithm sketch 25

Multi tag tracking algorithm sketch 25

Multi tag tracking algorithm sketch Apply joint DTW matching 25

Key Idea and Reasoning Touch and Single Tag Tracking Tracking within Multiple Tags Evaluation 26

Experimental Setup 27

Experimental Setup Camera RFID Tag Antenna Reader Laptop (Processing) 27

Single Tag Tracking Accuracy 28

Single Tag Tracking Accuracy 28

Single Tag Tracking Accuracy Error ~10 mm 28

Multi Tag Tracking Accuracy 29

Multi Tag Tracking Accuracy 29

Multi Tag Tracking Accuracy Error ~18 mm 29

Custom Tag Tracking Accuracy 30

Custom Tag Tracking Accuracy Custom Tags 30

Custom Tag Tracking Accuracy Error ~10 mm Custom Tags 30

Related Works 31

Related Works • RFID based gesture recognition – Data-driven learning • ShopMiner (SenSys ‘15), FEMO (SenSys ‘15), CBID (ToN ‘16) … • RFID tag tracking – Tracks tag movement • RF-IDraw (SIGCOMM ‘14), Tagyro (Mobicom ‘16), PolarDraw (CoNext ‘16) … • RFID based UI (Closest) – ML on a fixed set of gestures • PaperID (CHI 2015), IDSense (CHI 2016) … 31

Future works • Multi-touch tracking • General 2D gesture tracking • Tracking in different environment scenarios • Building customized tags for other use cases 32

Key takeaways • Impedance Tracking based Touch Primitive • Mutual coupling can be exploited • Custom designed tags may enable different applications 33

Thanks! Questions ? swadhin@cs.utexas.edu 34