by Carol Politi
TRX NEON sensor fusion and mapping algorithms augment GPS to provide both 2D and 3D information for devices while indoors. The algorithms are layered and architected to fit within ever evolving platform architectures that now include sensor hubs, sensor-specific processors, as well as application processors. This blog post outlines the functions performed within each layer.
As shown in the diagram, NEON Indoor Location Algorithms have three layers:
1. Sensor Processing - The lowest level, sensor processing includes power-efficient calculations designed to run as an always-on background process. Drift processing, step detection, step length/step direction computation, confidence estimation, feature detection, context identification, sensor calibration.
2. Navigation Processing - Navigation processing is performed at the application processor level, and is currently designed to run as a foreground application. This includes additional drift estimation and removal, constraint filtering (e.g., GPS, BLE, other external constraints), indoor/outdoor detection, elevation processing, application of map landmark/feature constraints, additional sensor calibration.
3. Map Processing - Map processing includes the processing and fusion of map features and signatures required to develop 3D navigation maps. These maps are used by the navigation processor to enhance location calculations.
The location services API provides information including X, Y, Z and error bounds to applications, and accepts constraints provided by applications or third party processes. This open architecture allows flexible integration of constraints detected by applications.
This architecture allows implementation of the solution on a broad array of devices, as long as the devices have a basic set of sensors (accelerometer, gyroscope, pressure, compass), and is easily extendible to accept constraints from user input or additional RF, audio, or optical sources.
by Carol Politi
TRX's fourth US indoor location patent recently issued (Patent US 8751151). (TRX has additional patents that have issued in Singapore and Australia). This patent covers a system and method for recognizing features for improving indoor location accuracy using Simultaneous Location and Mapping (SLAM). In this blog post, I'm going to give a bit of background on the requirements for and development of this technology.
TRX began developing indoor location solutions early, with a focus on delivering location for public safety personnel operating indoors, in buildings they did not control. As a result, the algorithms and techniques the company developed to support location were ones that did not rely on infrastructure. However, the algorithms relied upon sensors that were not common in commercial mobile or public safety devices. TRX built and now delivers sensor accessories that link to radios and cellular devices.
Flash forward a few years, and these sensors are now being embedded in mobile phones, industrial devices, and radios. The cost, size, and power requirements for the sensors are dropping rapidly. This means TRX algorithms can be used to power indoor location on an array of devices, and TRX wearable accessories can be far smaller and lower power.
The specific challenge TRX has worked to overcome is the fact that GPS does not work indoors or underground, and it often has large errors, especially when people are moving about in urban areas. Most approaches to indoor location require that facility owners install infrastructure. In these approaches, the applications requiring the location information must rely on information provided by the facility owner/operator. These approaches don't satisfy the strong requirements for infrastructure-free location from public safety, mobile, or defense markets.
Embedded sensors (e.g., accelerometer, gyro, compass, pressure, etc.) have long been used for navigation, however, they have historically been extremely expensive and limited to military platforms and industrial systems. Gaming, camera, and other mobile/portable device applications are driving the use of these sensors, providing a high volume market and increasing the quality available at low costs. However, while these sensors provide valuable navigation data, just embedding them in a device does not allow the device to deliver accurate indoor location. The sensors are subject to inertial drift, magnetic interference, and other errors that significantly erode accuracy. These errors have prevented the use of these low cost sensors in determining accurate indoor location without frequent aiding from GPS or beacons at known locations.
TRX has developed an approach that enables calculation of location indoors (and other locations without reliable GPS) using these low cost sensors. In this approach, features and landmarks detected using data from magnetic, inertial, RF, and other sensors are fused to establish navigation maps that augment sensor fusion algorithms implemented on the devices. Information from multiple users is merged to create navigation maps which are used to constrain errors. This combination of embedded sensor fusion and collaborative mapping allows the use of low cost sensors to calculate accurate indoor location over extended periods, without relying on frequent access to GPS or other corrections.
Thanks to the entire TRX team for the dedication, innovation, and hard work devoted to solving this really hard problem.
by Carol Politi
TRX demonstrated and is presenting its micro-location solution (indoor location, underground location) using collaborative mapping at the Joint Navigation Conference this week in Orlando. This is an excellent conference covering major areas of research in location determination and navigation for a broad range of platforms (pedestrians, vehicles, ships, etc). Conference papers describe advances in inertial measurement units, collaborative navigation techniques, and programs focused on reducing reliance on GPS and supporting location determination in areas without access to GPS.
Yesterday, TRX demonstrated its approach to implementing extended location determination for pedestrians in indoor environments using very low cost sensors (smartphone through low grade industrial). Today, Dr. Carole Teolis is presenting "Indoor Navigation using Collaborative Mapping". The public abstract of this paper is provided below:
Indoor Navigation Using Collaborative Mapping, Dr. Carole Teolis and Dr. Kamiar Kordari, TRX Systems
Indoor location technologies can pinpoint a person's location inside a building or in GPS-denied area; however, these technologies cannot do so effectively without accurate indoor maps. Indoor maps are a major component in enabling location-based applications by constraining error in location calculations, delivering a context for the location, and allowing for routing to a destination.
For the majority of buildings, indoor maps are not publically available, and in regions of military conflict, obtaining maps is even more problematic. The map data that is publically available is typically in image format and does not contain critical navigation routing in a format that can be used by the applications to correct location estimates. While the companies working on indoor mapping have the data formats defined that include navigation routing information, they are simply not populated. A growing number of companies (e.g., Google, Micello, Point Inside) are now creating excellent maps of indoor spaces but the majority of their effort is focused on high use environments such as malls, airports, hospitals, and museums.
The process of building and updating indoor maps is different from that of outdoor maps since neither satellite imagery nor GPS traces for referencing the images are available for indoor spaces. The current state of the art in indoor mapping requires a manual labor-intensive process to create and then to maintain maps as they change over time, which is difficult to scale. A system is needed to create and automatically maintain/update map navigation map data without hiring an army of mapping personnel to manually survey sites. One solution to the problems of scale in creation and maintenance of an accurate global database of indoor maps is a combination of mobile mapping and crowd sourcing where many people can use a indoor location-enabled mobile device to contribute to the creation of indoor maps.
These algorithms are being used to create an indoor mapping product to automate the indoor mapping process through sensor fusion and crowdsourcing. The generated database of indoor information will be made available through a cloud-based API for indoor location-based applications. This presentation will describe TRX’s methodology for map building and map based correction.
The foundation of the system is map discovery. Building features (e.g. hallways, elevators, exits, stairwells) and their connectivity relationships, signatures (RF, WiFi, magnetic), and navigable passageways are discovered as a person walks though a building while carrying their Smartphone or wearing a TRX tracking unit running the NEON mapping application. The features are discovered by classifying person's movements such as walking, taking stairs, or riding in an elevator based on the data from the suite of embedded sensors. The person's location, which is simultaneously being tracked using the same sensors, is used to identify the location of the discovered building features.
The discovered map data contains topological information on the building features that the subject can reach. This type of high level connection diagram of the environment is valuable for routing. The map we infer also has metric information for each link based on the subject's path that gives an estimated distance between features.
The technical challenge in crowd sourcing is developing the capability to merge and fuse data coming from many users to incrementally update a global map of indoor spaces. If the tracked subject has been globally initialized, their path information can provide global location estimates for the inferred features also. In order to be able to merge maps from different users, it is important to have global initialization (or at least relative initialization). The location accuracy of the discovered building features is dependent on the tracked person's location accuracy when the features are discovered. If the accuracy is low, the feature may not be useful. The fusion algorithms on the TRX server merge and fuse data coming from users to incrementally update a map of indoor spaces. By using techniques for automatically filtering poor quality map data, we improve both the completeness and accuracy of a map as the number of people contributing to the map increases in a building. To improve the robustness of the fusion process, we create descriptors that are used to differentiate discovered features. After map of a building is created, the map data can be used to provide for map-based location corrections in real-time.
Results will be presented using data collected in different types of buildings and underground structures to demonstrate the improved accuracy of this map-aided location system and how reliably it can provide accurate location for long periods of time without using GPS data.
This dynamic map-building tool provides an efficient mechanism for collaboratively creating and updating indoor maps. The maps improve situational awareness and enable the location system to provide a level of self correction without requiring infrastructure.
by Carol Politi
TRX will be showcasing its NEON Indoor Location and Indoor Mapping at the InvenSense developers conference this Wednesday and Thursday in Santa Clara, CA. The conference held by Invensense brings together System OEM's, Ecosystem Partners, and Application Developers focused on motion solutions for consumer electronic devices such as smartphones, tablets, wearables, gaming devices, optical image stabilization, and remote controls for Smart TVs. TRX wearable accessory and cellular device indoor location and indoor collaborative mapping solution will be demonstrated.
by Carol Politi
TRX had the pleasure of attending the Red Herring Top 100 Conference last week. Two hundred companies attended to network, discuss capital raising, company growth, and strategic partnering. The companies ranged from early stage startups doing several million dollars in revenue to later stage companies doing 50-100M+. Each company had the chance to pitch its business and each was evaluated based on a range of criteria including financial performance, quality of management and execution of strategy, technological prowess, IP creation, and the disruptive impact they bring to their respective industries. Everyone attending ended up giving elevator pitches about 50 times during informal networking events and delivering and defending a 12 minute business presentation. TRX was voted to be one of the Top 100 companies! This was a great event both for the networking and for the feedback on the company presentations.
by Carol Politi
TRX was fortunate to have been recognized by TEDCO as the winner of its Corporate Excellence Award during the 3rd annual ICE awards. TEDCO recognized TRX for its growth in the number of employees, revenue and investments, as well as community involvement and industry recognition.
TEDCO held its Fourth annual ICE Awards today and in the Corporate Excellence category, recognized the following Maryland companies as finalists:
- CSA Medical – Specializing in spray cryotherapy technologies, CSA Medical developed truFreeze®, which freezes and removes bodily tissues in general surgery with or without an endoscope.
- i-Lighting – Providing LED lighting solutions that create secure, lasting and attractive environments both indoors and outdoors.
- Looking Glass – Developers of information architecture, Looking Glass’s platforms enable users to asses and manage threat intelligence, monitor ecosystems and support decision-making.
CSA Medical was recognized today as the winner, and Dr. Carole Teolis of TRX presented them with the TEDCO Corporate Excellence Award.
TEDCO fills a critical role in supporting early stage businesses in Maryland - they were an invaluable asset to TRX as we initially conceived of our indoor location business and as we subsequently raised additional capital.
by Carol Politi
The US Patent and Trademark Office has issued three TRX Systems patents for systems and methods for determining 3-D indoor location using unique sensor fusion and building feature and signature mapping. Patent numbers US8,712,686, US8,706,414, US8,688,375 were issued, all directed toward unique TRX innovations for locating people, mapping indoor spaces, and unique methods for using known and discovered map data to deliver accurate user location.
The patents include use of accelerometers, gyros, magnetic field detectors, pressure, light, RF, GPS and other common sensors to support indoor location and indoor mapping. The TRX NEON solution uses inferred features and signatures to establish full three-dimensional maps of buildings and underground structures.
by Carol Politi
TRX Systems was selected as a finalist for Red Herring's Top 100 North America award, a prestigious list honoring the year’s most promising private technology ventures from North America. Red Herring selects the most exciting and promising start-ups and "scale ups", evaluating them individually from a large pool of candidates. Twenty major criteria underlie the scoring and process. They include, among others: the candidate company's addressable market size, its IP and patents, its financing, the proof of concept, trailing revenues and management's expertise. Each company goes through an individual interview after filling out a thorough submission, complemented by a due diligence. The list of finalists often includes the best performing and prominent companies of that year.
TRX is headed out to Monterey, California this week to present at the Red Herring North America Forum. You can follow the conference at: https://twitter.com/digitalherring hashtag #RedHerring100
by Carol Politi
Dr. Carol Teolis, Chief Technology Officer and Cofounder of TRX, is being honored today by the University of Maryland Department of Electrical and Computer Engineering in its third annual ECE Distinguished Alumni Awards.
The indoor location innovations spearheaded by Dr. Teolis and the TRX team are highlighted in her presentation on GPS-denied Location and Mapping at the University Of Maryland School of Engineering (link here). This discussion highlights the challenges associated with delivering seamless location using small RF, inertial, and other sensors in cellular devices and low-cost wearable accessories. The presentation highlights the broad array of approaches available to solve the challenge of infrastructure-free indoor location. Dr. Teolis reviews TRX's unique approach using very low cost embedded sensors (magnetic, accelerometer, gyroscope, pressure, Bluetooth, WiFi, light), fusing sensor information, and detecting unique structural, magnetic, and RF features within the building to develop collaborative map information that can be used to improve location estimates.
Congratulations to Dr. Teolis from the TRX team!
by Carol Politi
William English of TRX will be presenting TRX NEON Indoor Location Solutions at the NDIA 15th Annual Science & Engineering Technology Conference from April 8-10th. That presentation includes a video demonstration of NEON algorithms and technology as well as a technical overview of the NEON indoor location and mapping technology.
The conference is being held at the Marriott Inn & Conference Center in Hyattsville, MD.