Konark: A RFID based system for enhancing in-store shopping experience

Transcription

1 Konark: A RFID based system for enhancing in-store shopping experience Swadhin Pradhan 1, Eugene Chai 2, Karthik Sundaresan 2, Sampath Rangarajan 2, and Lili Qiu 1. 1 UT Austin 2 NEC Labs America WPA 2017, MobiSys 17 June 19, 2017

2 Motivation Consumer Retailer 2

3 Motivation Immersive Personalized Faster More Insights Fine-grained.. Consumer Retailer 2

4 Overview 3

5 Overview Creating a RFID based system to provide Better consumer shopping experience Richer retail analytics 3

6 Overview Creating a RFID based system to provide Better consumer shopping experience Richer retail analytics Queue-free checkout Which items customers put in the cart at a particular time window? 3

7 Overview Creating a RFID based system to provide Better consumer shopping experience Richer retail analytics Queue-free checkout Items bought? Items interested? Which items customers put in the cart at a particular time window? Detecting which items users are browsing 3

8 Our Work 4

9 Our Work Building a RFID reader and antenna equipped shopping cart and developing algorithms to detect items inside the cart, and to detect users browsing interests on-the-go. 4

10 Detecting Items Inside the Cart 5

11 Detecting Items Inside the Cart Key idea 5

12 Detecting Items Inside the Cart Key idea 5

13 Detecting Items Inside the Cart Key idea 5

14 Detecting Items Inside the Cart Key idea 5

15 Detecting Items Inside the Cart Key idea 5

16 Detecting Items Inside the Cart Key idea Reference Tags 5

17 Detecting Items Inside the Cart Key idea Reference Tags Algorithm sketch: Cart is Mobile? Yes Tag IDs Time Hi-dimensional Feature Space K-Means Clustering WPA 2017, MobiSys 17 Feature Values 5 June 19, 2017

18 User s Item Browsing Detection Key idea : 6

19 User s Item Browsing Detection Key idea : The phase variation of the tags of interacted items is higher. 6

20 User s Item Browsing Detection Key idea : The phase variation of the tags of interacted items is higher. Algorithm sketch: Cart is Mobile? No Filter tags according to nearest reference tags Track phase variation across tags Find the tags with highest phase variation 6

21 Setup Impinj Speedway Revolution RFID Reader Reads phase, RSSI, doppler of RFID tags 300 tags/second 50 channels, (25 MHz bandwidth) 6 dbi gain circular polarized Antennas Dogbone Monza 6 RFID tags (Suited for MHz reading) 7

22 Experimental Setup Reference Tags Laptop The Cart Antennas RFID Reader 8

23 Evaluation metrics 9

24 Evaluation metrics Modules : Detecting items inside the cart. Detecting user interest in a particular item. Metrics : Accuracy (What percentage of items predicted correctly?) ofalse Positive % (What percentage of items are outside but tagged inside?) ofalse Negative % (What percentage of items are inside but tagged outside?) Detection Latency (How much time it takes to detect?) 9

25 In-cart item Detection Accuracy 10

26 In-cart item Detection Accuracy 10

27 In-cart item Detection Accuracy Accuracy reaches ~100% after Detection latency of 60 seconds Accuracy remains good even after putting many items in the cart 10

28 Detection Latency vs Accuracy 11

29 Detection Latency vs Accuracy 11

30 Detection Latency vs Accuracy False positive rate is not too high even in low detection latency and high number of items 11

31 Item Browsing Detection 12

32 Item Browsing Detection For number of items ~ 200, to achieve ~100% accuracy, we need only ~20 sec latency. 12

33 Key takeaways Trade-off between detection latency and accuracy. Speed and benefits compared to traditional self-checkout system or vision based system (Amazon Go). (NLOS/ Occlusion). Infrastructure RFID solutions (ShopMiner [SenSys 15] or CBID [Infocom 14]) which work with smaller number of tags, and lacks user level information. 13

34 Future Works Making the retail analytics richer. Testing with multiple carts at different mobility. Collaboration among shopping carts etc. Field-testing the system in real shopping malls. 14

35