Intelligent Positioning in Cities using GNSS and Enhanced 3D Mapping

Lead Research Organisation: University College London
Department Name: Civil Environmental and Geomatic Eng


Poor positioning performance in dense urban areas is a major obstacle to the practical realisation of new technologies such as navigation for the visually impaired, tracking people with chronic medical conditions, augmented reality, advanced lane control systems for vehicles and advanced railway signalling systems.
The Global Positioning System (GPS) provides metres-level positioning in open environments. However in dense urban areas, buildings block, attenuate, reflect and diffract radio signals, limiting the real-time positioning accuracy to 10-50m when enough signals can be received to calculate a position. Other radio positioning technologies are typically no more accurate, while position obtained from dead reckoning degrades with time. Optical techniques developed by the robotics community are more suited to some applications than others and are still undergoing research to make them more reliable and efficient.
Using the new global navigation satellite systems (GNSS) constellations (i.e., GLONASS and, in future, Galileo and Compass) in addition to GPS improves the availability of satellite-based positioning in urban areas. However, to improve the accuracy, a new approach to positioning is needed and the increasing availability of 3D mapping provides an opportunity to achieve this.
The aim of this project is thus to improve the accuracy of real-time mobile positioning in urban areas to within a few metres by combining multi-constellation GNSS with 3D mapping, a concept known as intelligent urban positioning. By exploiting knowledge of the surroundings provided by 3D city models and rebuilding the positioning algorithms from the bottom up to make use of all available information, a step change in positioning performance can be achieved, unlocking the potential for a host of new positioning applications.
This research will build on UCL's track record of innovation in urban positioning, including the development of a brand new GNSS positioning method known as shadow matching. This project will investigate new ways of using 3D mapping to aid ranging-based GNSS positioning and then combine this with shadow matching to obtain the best overall position solution. Testing will be conducted under a wide range of scenarios to assess how the performance varies as a function of the urban environment, the class of GNSS user equipment used and the characteristics of the 3D mapping. Finally, context detection algorithms will be developed to determine when the positioning system is in an environment suitable for the algorithms developed under this project and when it is in an environment where conventional GNSS algorithms or an indoor positioning technique should be deployed instead.
By improving the accuracy and reliability of urban positioning, a successful outcome of this project would unlock the potential for many new applications that can both contribute to the economy and provide solutions to societal problems, while improving the reliability of many existing technologies. Positioning technology that can determine the correct side of the street and identify adjacent buildings is a key component of automated guidance for visually impaired pedestrians. More accurate emergency caller location and tracking of people with chronic medical conditions enables response teams to arrive more quickly. Augmented reality will benefit from a more efficient overlaying of information on the surrounding environment. Researchers mapping patterns of air pollution or wheelchair accessibility in cities will be able to quickly and cheaply geolocate information to within a few metres. More reliable identification of traffic lanes and railway tracks will support the development of advanced intelligent transport systems. Route guidance for visitors to cities, location of friends and business associates in complex or crowded urban environments, and location-based advertising will also benefit.

Planned Impact

According to the European GNSS Agency (GSA) GNSS Market Report - Issue 2, dated May 2012, the worldwide global navigation satellite systems (GNSS) market size is expected to grow from £54bn in 2012 to £137bn in 2020. This will be dominated by the road and location-based services (i.e. mobile device) sectors, both of which are affected by the problem of poor real-time positioning in dense urban areas. The GNSS industry has an extensive presence in the UK, including CSR, Forsberg, Intel, QinetiQ, Spirent, ST Microelectronics, UBlox, Veripos and others. There are also many companies that use GNSS in their products. By improving the accuracy and reliability of urban positioning, a successful outcome of this project would unlock the potential for many new applications that can both contribute to the economy and provide solutions to societal problems, while improving the reliability of many existing technologies. ST Micro-electronics and Ordnance Survey will directly benefit as project partners.
Automated navigation and guidance for blind and visually impaired pedestrians will significantly improve their independence, boosting quality of life and enabling them to contribute more to the economy. Positioning technology to enable them to easily determine the correct side of the street and identify adjacent buildings, noting that many shops and cafés are only a few metres wide, is an important part of this. This is reflected in the participation of the Royal National Institute for Blind People (RNIB) as a project partner.
Emergency caller location and tracking of people with chronic medical conditions enables response teams to reach patients and other emergencies more quickly. If people can be located reliably to the correct building, response times will be shortened and lives potentially saved.
Many location-based services will benefit from better urban positioning, generating economic benefits. More reliable location and route guidance services for visitors to cities wilmprove the user experience, increasing their take up and making them more attractive to advertisers. It will also enable reliable location of friends and business associates in complex or crowded urban environments. Location-based advertising will benefit from reliable knowledge of which shop or business a potential customer is visiting. This will enable advertising to be better targeted, minimising intrusion from irrelevant adverts and thus making consumers more amenable to receiving advertising on mobile devices.
Reliable identification of traffic lanes is an important requirement for advanced intelligent transport systems that can direct individual vehicles to maximise traffic flow, prevent gridlock and prioritise emergency vehicles and buses, making best use of the available space in crowded cities. Similarly, advanced rail signalling systems require a system for identifying which track a train is on. By combining the new techniques developed under this project with dead reckoning and/or optical techniques, reliable real-time identification of traffic lanes and railway tracks should become a realistic prospect.
The potential for augmented reality to enhance the lives of both professionals and consumers is huge. However, the information must be correctly overlaid on the surrounding environment. A more accurate position initialisation will make the overlay process less reliant on image-matching, which can be confused by repeating features, improving both the robustness and the processing efficiency.
Other applications with economic benefit include high-value asset tracking, road-user charging, pay-as-you-drive insurance and identification of "white space" radio spectrum for short-range communication. Applications with social benefit include enforcement of exclusion zones and curfews and spatial data gathering in cities to produce maps of pollution or accessibility for the disabled.
Description Global navigation satellite systems (GNSS) positioning in dense urban areas is degraded because the signals are blocked and reflected by the surrounding buildings. By using 3D mapping of the buildings, we have achieved a factor of 5-6 improvement in real-time positioning accuracy. Results across several sites in dense urban areas within London have shown that our single-epoch 3D-mapping-aided (3DMA) GNSS algorithm achieves a horizontal RMS position accuracy of 4.7m using a consumer-grade GNSS receiver and 4.9m using a tablet (equivalent to a smartphone) . This compares with an equivalent single-epoch conventional GNSS position accuracy of 26.4m using the consumer-grade receiver and 31.0m using the smartphone. We have also shown that our algorithms can run in real-time on a Raspberry Pi 3 and an Android smartphone
In more detail, we have shown that terrain height aiding significantly improves GNSS positioning in dense urban areas and that weighting GNSS ranging measurements according to predicted GNSS signal availability in a least-squares positioning algorithm also improves performance. We have developed a new hypothesis-based 3D-mapping-aided (3DMA) GNSS ranging algorithm and shown that this improves performance over our earlier algorithm. We have integrated 3DMA GNSS ranging with GNSS shadow matching using both position-domain and hypothesis-domain approaches. We have found that best performance is obtained from hypothesis-domain integration of hypothesis-based 3DMA ranging with shadow matching. The 3DMA least-squares algorithm is used for initialisation. We have conducted a thorough performance assessment to determine the impact on performance of sky visibility, street width, building size and type, mapping quality, search grid spacing and passing vehicles, helping us to understand where future research should be directed.
Final project results have been accepted for publication as a two-part journal paper in NAVIGATION.
Exploitation Route Our research findings could be used to significantly improve the accuracy of GNSS positioning in dense urban areas. For example, our algorithms could be implemented on a smartphone following the 2016 update to the interface between the GNSS chip and the Android operating system. For specialist applications, a dedicated GNSS receiver could be used.
Our research on this topic will continue under other funding. We will be looking at multi-epoch positioning to improve accuracy, together with ways of improving computational efficiency. There are many more ways in which performance could potentially be further improved.
Sectors Aerospace, Defence and Marine,Digital/Communication/Information Technologies (including Software),Environment,Healthcare,Leisure Activities, including Sports, Recreation and Tourism,Culture, Heritage, Museums and Collections,Security and Diplomacy,Transport,Other

Description We have some initial interest from industry and collaborated with Swarm Systems for possible micro air vehicle implementation, starting in 2016 and culminating with a small flight trial in March 2018. We have discussed possible smartphone implementation with Google since late 2017. They were initially potentially interested in using our software, but are now considering implementing these techniques themselves, making use of our published research. During 2018 and early 2019, we tested our algorithms using Google-supplied data and shared the results with them. Further collaboration is ongoing, but remains at the research level, not development. Uber have implemented a version of Shadow Matching partially based on work at UCL prior to this EPSRC project so may also be interested in making use of the results. We are also aware of interest in our research at Apple; however, they tend not to collaborate with universities.
First Year Of Impact 2017
Sector Aerospace, Defence and Marine,Digital/Communication/Information Technologies (including Software)
Description Dstl CDE
Amount £50,000 (GBP)
Organisation Defence Science & Technology Laboratory (DSTL) 
Sector Public
Country United Kingdom
Start 04/2017 
End 03/2018
Description Faculty Research Award
Amount $82,000 (USD)
Organisation Google 
Sector Private
Country United States
Start 09/2017 
End 12/2018
Description Google Faculty Research Award
Amount $80,632 (USD)
Organisation Google 
Sector Private
Country United States
Start 01/2019 
End 12/2019
Description Impact Acceleration Award (a project within a larger IAA awarded to UCL as a whole)
Amount £15,000 (GBP)
Funding ID EP/K503745/1 
Organisation Engineering and Physical Sciences Research Council (EPSRC) 
Sector Academic/University
Country United Kingdom
Start 11/2016 
End 03/2017
Description Collaborate with Google 
Organisation Google
Country United States 
Sector Private 
PI Contribution Currently confidential
Collaborator Contribution Currently confidential
Impact None yet
Start Year 2017
Description Collaboration with Swarm Systems 
Organisation Swarm Systems Ltd
Country United Kingdom 
Sector Private 
PI Contribution Under the Impact Acceleration Award, we have adapted our 3DMA GNSS algorithms to work in the air and tested them using data collected on a long pole. Under subsequent funding from a Dstl CDE, via Swarm Systems, we demonstrated the algorithms running in real-time using GNSS data from a Swarm Systems micro air vehicle.
Collaborator Contribution Swarm Systems have modified their micro air vehicle to enable real-time demonstration of our algorithms using it and flew it during the live demonstration.
Impact Demonstration of UCL algorithms in real-time using GNSS data from a Swarm Systems micro air vehicle.
Start Year 2016
Description U-Blox Intelligent Positioning for Cities 
Organisation U-Blox Melbourn Ltd
Country United Kingdom 
Sector Private 
PI Contribution Provision of early access to results and an opportunity to influence the direction of the project.
Collaborator Contribution Supply of three GNSS receiver development modules. Provision of advice through attendance at 6-monthly project progress and steering meetings. Active participation in outreach events. Note that U-Blox have replaced ST Microelectronics as a project partner.
Impact Results obtained using the U-Blox GNSS receivers have been published in two conference papers.
Start Year 2014
Description Intelligent Positioning in Cities - Solving the Urban Positioning Problem 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Industry/Business
Results and Impact This one day event at UCL was held as part of the Intelligent Positioning for Cities project. Presentations were given by project team members, project industry partners and invited guests. There was a group workshop activity brainstorming new urban positioning applications and a panel-audience discussion. About 55 people from Industry and Academia attended.
Year(s) Of Engagement Activity 2015