october 2015 industrial indoor positioning prototype ...beacons with different hardware...

2
2015 International Conference on Indoor Positioning and Indoor Navigation, 13 th -16 th October 2015 Industrial Indoor Positioning Prototype Implementation for Field Service Automation Ching-Ling Huang, Bo Yu, Piyush Modi GE Global Research 2623 Camino Ramon San Ramon, CA, USA 94583 {chingling.huang, yu, modi}@ge.com Abstract— GE’s Industrial Internet aims to connect machines & people at work with data analytics to improve performance of all assets in the field. In this paper, we report a prototype platform, called MyWorld, built on top of a software framework incorporating two types of BLE beacons to provide key features like location-aware information push, indoor navigation support, and real-time personnel tracking. This framework aims to support high-level applications with heterogeneous beacons. Keywords— Industrial Internet; indoor positioning; iBeacons; Bluetooth low engergy; field service automation I. OVERVIEW Industrial environments have undergone profound changes as the business assets that make up the manufacturing and service chain - people, products, equipment, and raw materials - are increasingly on the move. GE’s Industrial Internet vision [6] belongs to one of the latest efforts to connect machines & people with the aid of data analytics to improve overall performance of all assets in the field. In this paper, we report our indoor positioning experiment results and the prototype, called MyWorld, built on top of the software layer for field service automation. A. Indoor Positioning Using BLE Beacons We first present the empirical channel models derived from measuring the RSSI and the ground-truth distance between a BLE iBeacon [3] and an iPad with BLE. In this small-scale experiment, we collected 1000 samples of RSSI data at the incremental separation distance of every 30 centimeters. The collected RSSI data samples are then MMSE optimally fitted into a two-segment exponential curve with respect to distance, which is a common approach for modeling wireless channel for WiFi and Bluetooth PHY layer [2]. Fig. 1 summarized our experiment results as two empirical channel models for two types of beacons: One is from Gimbal [4] and another from Estimote [5]. Following the ranging convention in iBeacon, the proximity definition is color-coded below the distance axis. Due to different hardware and implementation, two kinds of beacons show slightly different statistical behaviors in the same indoor space. However, the estimated distance is then classified into the same proximity definition. B. Software Framework To provide uniform APIs for indoor positioning at high- level application level, a software framework is proposed in our approach. This proposed framework is driven by two important requirements: It must be able to accommodate different types of beacons to coexist, as each site may have different types of sensors. It must provide the seamless indoor positioning experience as the user switches between online and offline modes. In order to meet these requirements, our framework in Fig. 2 includes several important components explained below. Figure 1: RSSI vs. Distance: exponential fitted channel models for Gimbal & Estimote [4,5] beacons. The proximity is color-coded as Immediate (<1m), Near (1~5m) and Far (>5m) under Distance axis. Figure 2: Software framework to support different kinds of beacons and hybrid network environment.

Upload: others

Post on 23-Mar-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: October 2015 Industrial Indoor Positioning Prototype ...beacons with different hardware implementation while keeping the application level agnostic of the underlying hardware, beacon

2015 International Conference on Indoor Positioning and Indoor Navigation, 13th-16th October 2015

Industrial Indoor Positioning Prototype Implementation for Field Service Automation Ching-Ling Huang, Bo Yu, Piyush Modi

GE Global Research 2623 Camino Ramon

San Ramon, CA, USA 94583 {chingling.huang, yu, modi}@ge.com

Abstract— GE’s Industrial Internet aims to connect machines

& people at work with data analytics to improve performance of all assets in the field. In this paper, we report a prototype platform, called MyWorld, built on top of a software framework incorporating two types of BLE beacons to provide key features like location-aware information push, indoor navigation support, and real-time personnel tracking. This framework aims to support high-level applications with heterogeneous beacons.

Keywords— Industrial Internet; indoor positioning; iBeacons;

Bluetooth low engergy; field service automation

I. OVERVIEW

Industrial environments have undergone profound changes as the business assets that make up the manufacturing and service chain - people, products, equipment, and raw materials - are increasingly on the move. GE’s Industrial Internet vision [6] belongs to one of the latest efforts to connect machines & people with the aid of data analytics to improve overall performance of all assets in the field. In this paper, we report our indoor positioning experiment results and the prototype, called MyWorld, built on top of the software layer for field service automation. A. Indoor Positioning Using BLE Beacons We first present the empirical channel models derived from measuring the RSSI and the ground-truth distance between a BLE iBeacon [3] and an iPad with BLE. In this small-scale experiment, we collected 1000 samples of RSSI data at the incremental separation distance of every 30 centimeters. The collected RSSI data samples are then MMSE optimally fitted into a two-segment exponential curve with respect to distance, which is a common approach for modeling wireless channel for WiFi and Bluetooth PHY layer [2]. Fig. 1 summarized our experiment results as two empirical channel models for two types of beacons: One is from Gimbal [4] and another from Estimote [5]. Following the ranging convention in iBeacon, the proximity definition is color-coded below the distance axis. Due to different hardware and implementation, two kinds of beacons show slightly different statistical behaviors in the same indoor space. However, the estimated distance is then classified into the same proximity definition. B. Software Framework To provide uniform APIs for indoor positioning at high-level application level, a software framework is proposed in

our approach. This proposed framework is driven by two important requirements: • It must be able to accommodate different types of beacons

to coexist, as each site may have different types of sensors. • It must provide the seamless indoor positioning experience

as the user switches between online and offline modes. In order to meet these requirements, our framework in Fig. 2 includes several important components explained below.

Figure 1: RSSI vs. Distance: exponential fitted channel models for Gimbal & Estimote [4,5] beacons. The proximity is color-coded as Immediate (<1m), Near (1~5m) and Far (>5m) under Distance axis.

Figure 2: Software framework to support different kinds of beacons and hybrid network environment.

Page 2: October 2015 Industrial Indoor Positioning Prototype ...beacons with different hardware implementation while keeping the application level agnostic of the underlying hardware, beacon

2015 International Conference on Indoor Positioning and Indoor Navigation, 13th-16th October 2015

Hardware Abstraction Layer (HAL) handles the different channel characteristics and data format of Estimote & Gimbal beacons and passes a list of all detected beacon IDs with estimated distance to that beacon (based on RSSI readings and the corresponding channel model). This HAL entity allows all the coexisting beacons to be represented in a single list with the same attributes. The low-level differences of beacon formats, IDs, types, and raw messages will be handled by HAL and thus the application level software can be designed and developed without knowing the details of underlying BLE hardware implementations and corresponding channel models of heterogeneous beacons. On top of HAL, we provide two types of communications between server and client. The WebSocket connection enables real-time location queries from client to server. When the client operates in online mode, it keeps posting the raw sensor data to the server, which computes the positioning results to the client in return. Meanwhile, every time the user gets online, it also uses the REST API to download a local version of RSSI lookup tables for each type of beacon, which can later be used for offline access. During online mode, all computation is conducted on the server side, and the client always gets the most accurate positioning results. During offline mode, the client makes use of the last updated local RSSI lookup table to estimate the location. C. MyWorld ProtoType MyWorld [7] is a prototype for field engineers to enhance operation efficiency and to utilize adaptive work environments by using technologies like wearable computing, mobile devices, collaborative sensing, augmented reality and natural user experience technology. MyWorld is intended to provide the seamless presentation and access to relevant asset data, contextual information sources, insights and connect people with various modalities. MyWorld is built on top of the software framework depicted in Fig. 2. The main advantage of this software framework is the flexibility to work with BLE beacons with different hardware implementation while keeping the application level agnostic of the underlying hardware, beacon IDs, and empirical channel models. We incorporated the indoor positioning module into the prototype to provide real-time location-aware capabilities and proximity detection in several features like Location-Aware Information Push, Indoor Navigation Support, and Real-time Personnel Tracking. See screenshots in Fig. 3 – 5.

REFERENCES [1] R. Mautz, “Indoor Positioning Technologies”, Habilitation Thesis

submitted to ETH Zurich, Feb. 2012. [2] H. Liu, et. al., “Survey of Wireless Indoor Positioning Techniques and

Systems”, IEEE Transactions on Systems, Man, and Cybernetics – Part C: Applications and Review, Nov. 2007.

[3] iBeacon: http://en.wikipedia.org/wiki/IBeacon [4] Gimbal (previously Qualcomm) beacons: http://gimbal.com [5] Estimote beacons: http://www.estimote.com [6] GE’s Industrial Internet: https://www.gesoftware.com/industrial-internet [7] R. S. Silva Filho, et. al., “Using Wearable and Contextual Computing to

Optimize Field Engineering Work Practices” to appear in HCII 2015.

Figure 3: Nearby assets are listed in the order of proximity. When the engineer approaches the asset, a summary of asset is proactively shown and can lead to detailed maintenance instructions. See [7].

Figure 4: The field engineer is marked as the blue circle. In order to reach the Tool Depot #1, the engineer is instructed to route through Milling Station with its image shown on the left upper corner.

Figure 5: The digital map shows all assets and personnel in the same space. This shared view by mobile users and the remote observers (e.g., control center or supervisors) can enable intuitive interactions, better collaborations among human & machines.