Between the Poles

Deep learning enables automated extraction of building footprints and road networks from satellite imagery

Automated feature extraction from satellite imagery has made major progress in the last year. Accurate building footprints extracted from high resolution satellite imagery are available from companies such as Ecopia (which has just announced a partnership with DigitalGlobe).  Also NVIDIA has demonstrated the ability to automate detection of many road networks using sophisticated algorithms and multi-spectral high resolution imagery.

Building footprints

SpaceNet is a repository commercial satellite imagery and labeled training data being made available through Amazon Web Services at no cost to the public to foster innovation in the development of computer vision algorithms to automatically extract information from remote sensing data.  It is the result of a collaboration between DigitalGlobe, CosmiQ Works, and NVIDIA.  Two competitions for automated building footprint extraction from high resolution satellite imagery and a third for road network extraction have been held based on SpaceNet data. 

The first SpaceNet Challenge was launched in November 2016.  Imagery of Rio de Janeiro from the WorldView 2 satellite at 50cm resolution using eight spectral bands was made publicly available on Amazon Web Services (AWS).  42 developers competed to create algorithms that extract building footprints from the imagery.  The source code of the winning algorithms was made available on the SpaceNetChallenge GitHub repository. 

The winning implementation was developed by Brazilian developer using an implementation based on random forests with brute force polygon search and did not leverage deep learning frameworks.  The organizers' conclusion based on the results of Round 1 was that automated building footprint extraction remained a challenge.

Round 2 winner. Most building footprints are precisely detected. Number on each building footprint represents the overlap between the prediction and the ground truth.

Round 2 of the SpaceNet challenge involved higher resolution 30 cm imagery from WorldView 3 for four cities across the globe. The competition challenged developers to improve performance from the first competition using the higher-resolution imagery and more geographically diverse training data samples. SpaceNet on AWS held imagery and building footprints for Las Vegas, Paris, Shanghai, and Khartoum with over 180,000 buildings181,619 footprints.  In addition additional bands were made available (panchromatic, 3-band pan-sharpened, and 8-band pan-sharpened).

Results from Round 2 winning solution: top left: Las Vegas, top right: Las Vegas, lower right: Khartoum, and lower left: Paris. Blue outlines: ground truth, green outlines: correct identifications, red outlines: false positives and yellow outlines: false negatives.

The winner of Round 2 applied a  deep neural network model developed originally for medical image segmentation called U-Net.  Successful training of deep networks requires many thousand annotated training samples.  Training was aided significantly by augmenting the imagery with OpenStreetMap data.  The winning solution used OpenStreetMap layers and Worldview-2 multispectral layers as the input of the deep neural network algorithm.  In Las Vegas, the winning solution was outstanding in outlining suburban homes, but also did well in identifying odd-shaped buildings. In Khartoum, the solution was very good in identifying the footprints of stand-alone apartment complexes, but had trouble with small buildings, with building footprints that are close together, and with horizontally large buildings.  One of the problems was inconsistent ground truth which affected training. The solutions offered by the developers in the competition were sufficiently successful that the organizers concluded that the winning algorithms achieved performance with potential for automated mapping tasks such as keeping maps up-to-date and to assist first responders during natural disasters.  The source code for the winning implementations can be found in the SpaceNetChallenge GitHub repository.

DigitalGlobe has just announced a partnership with Ecopia, which has established an automated process to create building footprints quickly and at scale by leveraging DigitalGlobe's Geospatial Big Data platform (GBDX) and advanced machine learning in combination with DigitalGlobe’s imagery library. The two companies plan to automatically extract accurate 2D building footprints globally, then refresh the datasets periodically to find and track changes over time.  These datasets would be valuable to municipal governments for permitting purposes and for first responders after disasters.   Ecopia is developing a database for all of Australia (7.6 million square kilometers) which will include every building with a roof area greater than 9 square meters.  The database will contain building footprints for about 15 million buildings. The database includes linkages to other datasets including geocoded address, property data and administrative boundaries. Ecopia intend to update the dataset regularly to be able to track changes, especially urban sprawl.

Road networks

In November 2017 CosmiQ Works, Radiant Solutions, and NVIDIA announced a third round competition, “Road Detection and Routing Challenge” to explore automated methods for extracting map-ready road networks and routing information from high-resolution satellite imagery.  

Detecting road networks in Paris. Photo image, ground truth, and the results achieved with three deep learning algorithms

NVIDIA has released the results of several deep learning algorithms that illustrate just how difficult identifying road networks is and the sophistication of the available tools.  These include using multi-band spectral imagery to identify the material properties of the road surface itself (asphalt, gravel, packed earth) and using more sophisticated processing including conditional random fields (CRF), percolation theory, and reinforcement learning. 

Detecting road networks in Las Vegas. Photo image, ground truth, and the results achieved with three deep learning algorithms

NVIDIA utilized high performance GPU compute resources provided by the NVIDIA GPU Cloud and Amazon Web Services.

Studies find geospatial proliferating across the world economy

The latest estimate of the annual contribution of the geospatial industry to the world economy of US$400 billion represents a substantial increase over a previous estimate of US$150-270 billion released in 2013.  Both reports agree that the total contribution of geospatial data and technology annually to the world's economy is to be calculated in trillions of US dollars.  There is evidence that geospatial's contribution is increasing.

 AlphaBeta used consumer surveys from 22 countries, big data analysis of online job postings and other analytics to determine the impact of geospatial data, technology and services on the world economy. The study (The Economic Impact of Geospatial Services, AlphaBeta, September 2017) estimates that geospatial services generated annual revenue of US$400 billion per year in 2016 - this directly contributed to increasing the world's GDP. In addition geospatial data, technology and services save time and fuel for the travelling public - this contributes to increasing labour productivity. The study estimates that these efficiency benefits save the public over US$550 billion per year. Together the contribution in efficiency gains and direct revenue are estimated to reach almost a trillion US dollars. This is viewed as a conservative estimate of the contribution of geospatial to the world economy since the study was able to quantify only a few of the ways that geospatial data, technology and services benefit the economy.  The ways that AlphaBeta was able to quantify in which geospatial contributes to the world economy include productivity gains for the travelling public, direct revenue generated by the geospatial industry, and revenue gains and cost savings in other industries from using geospatial data and technology.

Productivity - digital map data and technology shorten traveling time by 12% (estimated to be worth US$ 264 billion). It also saves US$ 20 billion in fuel. Geospatial also enables consumers to make buying decisions faster, saving more than 21 billion hours (estimated to be worth US$283 billion).

Direct revenue - the geospatial industry generates US$400 billion in revenue per year.

Indirect revenue and savings - the study suggests that the increased revenue from geospatial-enabled services is several times higher that direct revenue to the geospatial industry. For example, the study estimates that digital maps drives US$1.2 trillion of sales worldwide. Furthermore the study estimates that geospatial technology improves revenues and costs by at least 5 percent in sectors contributing more than 70 percent to global GDP, for example, retail and mining. This has an estimated value of about US$2.7 trillion.

Other non-quantified benefits - other benefits that are difficult to put a price tag on include creating 4 million jobs, reducing automobile emissions by 1,686 million metric tons, reducing emergency response time by up to to 20% in some countries, and improving emergency preparedness.

To put this in context, according to a report prepared by the Boston Consulting Group (BCG) in June 2012 the geospatial services industry in the United States generated annual revenues of $75 billion. The BCG report estimated the economic impact of the geospatial services industry on government, business, and consumers to be $1.6 trillion in revenues (greater efficacy) and $1.4 trillion (about 8.7% of the U.S. GDP) in cost savings (greater efficiency).  In 2013 Google released a report What is the economic impact of Geo services ? prepared by Oxera Consulting Ltd. Global revenues from geospatial products and services as defined by Oxera was estimated to be $150-$270 billion per year. The $270 billion estimate was computed by scaling up the BCG estimate for the United States to the world economy.  The latest report considerably increases the estimated direct revenue to the geospatial industry, but is inline with the 2013 report in estimating that the total benefits of geospatial to the economy lies in the range of several trillions of US dollars.  The improvements in revenue and costs in all industries which use geospatial data and technology is estimated to amount to roughly 3% of the world's GDP.

For comparison the most quoted estimate of geospatial data an technology to a national economy is a 2008 a report prepared for the CRCSI & ANZLIC by ACIL Tasman estimated that the spatial information sector contributed between 0.6% and 1.2% of the Australian GDP in in 2006-2007.  The recent Google results suggest that the proliferation of geospatial data and services has increased the penetration of geospatial in the world's economy.

There is other evidence that geospatial is proliferating in the world's economy.  McKinsey has estimated that by 2020 the benefits (difference between willingness to pay and the cost of applications) that customers, consumers and businesses, will enjoy from location-based applications will reach as much as US$700 billion/year. Most of this, about 70 %, will be the consumer benefit from time and fuel saved by using GNSS navigation systems including those with real-time traffic information and mobile LBS applications. The remaining 30 % of the benefits will be through location-enabled applications such as geo-targeted mobile advertising.  Smart navigation applications are estimated to provide $500 billion in value to global consumers in time and fuel saved by 2020. Geo-targeted advertising could represent more than 5 percent of total global advertising spending by 2020.

A major industry where geospatial is already having a significant impact which may accelerate is construction.  Globally construction is expected to grow by almost 70% from $7.2 trillion today to $12 trillion by 2020.   Construction productivity has been stagnant in many advanced economies for decades.  It has been slow to adopt process and technological innovations and its R&D spending has lagged other industries. In a recent report, Imagining construction’s digital future, McKinsey suggests compelling reasons why the construction industry is ripe for disruption and how digitization might contribute to its transformation.  Two of the technologies that McKinsey identifies that will be key in the anticipated transformation are geospatial and BIM. 

Elevated risk in road construction lacking knowledge of the subsurface

The challenge of increasing urbanization in a world where anthropogenic pressure on the environment has become threatening is driving the digitalization of urban planning and design, construction and maintenance and operations of urban infrastructure. Building information modeling and laser and photogrammetric scanning of entire cities is resulting in comprehensive digital twins of the above ground infrastructure of cities. But the below ground has been neglected. There is a growing recognition that the subsurface needs to be included in urban planning and construction to reduce the risks of conflicts with subsurface infrastructure and to maximize the benefits of subsurface resources.

One of the challenges identified by the European COST Sub-Urban program is a communications gap between urban planners and subsurface experts. To begin to improve the level of communication between urban planners and subsurface experts, Geodan and TNO organized a preconference workshop “Utilizing subsurface data for urban planning, design, construction and maintenance” at the GEO|Design+BIM conference in Amsterdam.

Alternative routes for proposed highway

The workshop was a hands-on event with real data from a part of the Netherlands. It was led by Cindy Teeven and Ceciel Fruijtier from Geodan and Rob van der Krogt from TNO. In the simulation in which all participants in the workshop participated, a decision had to be made about the preferred route for a new highway. The alternatives were a route that skirted existing buildings and followed the route of an existing railway and regional road and a route that went through the countryside. An online GIS system provided layers of typical 2D information from public databases that might be relevant to the decision process including the location of existing residential and business buildings, historical and conservation sites, existing infrastructure such as roads and railways, and so on. Each of the participants was assigned a role to play: people of the place, traffic engineer, civil engineer, regional government, environmental authority, and local business council. After an hour with the data we were asked to report on the decision that had been reached on a preferred route. Primarily for reasons relating to reducing traffic through populated areas, the choice of most of the stakeholders was the route through the countryside. However, the “civil engineers” raised a red flag about this route because of the risk of poor soil conditions that could make this route challenging and expensive. Clay and peat, which are prevalent in the Netherlands, are susceptible to subsidence when a load is placed on them and are therefore to be avoided whenever possible when building any kind of structure. In contrast a great deal was known about the soil conditions of the alternative route because infrastructure and houses had been built along it for many years.

Surface elevations from lowest (blue) to highest (red)

After the first hour it was announced additional information was provided including a digital terrain model (DTM) showing surface elevation for the area and a model of the subsurface composition of the terrain – whether sand, peat or clay - to roughly 50 meters interpolated from bore holes and groundwater sampling sites in the region.

The DTM showed that the route skirting existing infrastructure was generally higher by a couple of meters than the alternative route through the countryside where the land was low lying suggesting subsidence. The subsurface soil composition model was used to create profiles showing the probability of clay and peat in the subsurface along the proposed routes. The 3D model revealed a high probability of clay and peat in the subsurface along the route through the countryside. In light of the new information about the subsurface, participants were requested again to arrive at a decision for the preferred route. Armed with the new subsurface information several of the stakeholders changed their minds. For the “civil engineers” the new subsurface information confirmed their fears about the risk of the route through the countryside.

The cost of an elevated highway or viaduct over the extensive peat and clay fields would be much more expensive than the alternative route skirting existing infrastructure. If the simulation had included cost and it had been made clear that the cost would have to be borne through a rate increase, I would expect that more of the stakeholders would have changed their minds.

Probability of clay in the subsurface along route from lowest (blue) to highest (red). Vertical dimension is depth.

It was explained at the end of the workshop that the workshop was based on a historical event in the north of Holland. At the time it was decided to build the route through the countryside. After extensive construction delays and rising costs, it was ultimately resolved in a high profile court case. The outcome was that the construction contractor and the regional government agreed to split the much higher cost of building a viaduct through the countryside.

The workshop was effective in illuminating the risk of decision making without information about the subsurface. In this case there was some data about the subsurface although historically it was only consulted after the initial decision. The reality in much of the world is that there is little information publicly available about the subsurface. Many construction projects proceed blindly with respect to what lies below the surface. In the United States poor information about what lies below the surface is the leading cause of construction delays and cost overruns. Construction bids are routinely inflated significantly to accommodate the risks associated with the lack of knowledge about the subsurface. Poor information about the subsurface costs the world economy trillions of dollars. What is underway in the Netherlands with the Key Registry for the Subsurface or BRO is the beginning of a practical solution to this problem. The world should be watching.

Fully automated classification of road and highway features for autonomous vehicles

Point clouds are fine for visualization to provide context, but to be really useful they need to be classified.  Feature extraction, as the process is often referred to,  is currently a mostly manual process that has been used for many years in the remote sensing sector. A trained person visually inspects the point cloud and graphically selects features which are identified by type of feature, sign, light pole, lane marking, and so on. 

Automated classification of point clouds

Image processing of 3D point clouds has advanced to the point where it has been possible to semi-automate this tedious feature extraction process.  The reliability of the automation process depends on the point density of point cloud, often expressed as points per square meter or per square centimeter.  Point density depends on the technology used to capture the point cloud. Terrestrial lidar, for example, typically captures points at the rate of hundreds of thousands to millions of point per second and is capable of millimeter to centimeter precision. Phodar generates lower point densities and is less precise. Aerial lidar captures many points per second but with less precision because it is covering much greater areas. Single photon lidar is a new technology that can collect 6 million points per second which means that planes can fly quite high, cover very broad areas and still collect 12 - 30 points per square meter.

At the Year in Infrastructure 2017 (YII2017) conference in Singapore, Sharad Oberoi of Sanborn gave an insightful overview of the state of the art for automated feature extraction from combined aerial and mobile scans of highways and adjacent areas.  Different types of laser scanners are appropriate for capturing different types of features. Terrestrial and mobile laser scanners can capture point clouda that are suitable for extracting curbs, gutter drains, traffic signals, road and local terrain information, parking meters, walls, obstructions in the right of way, manholes, sidewalks, overhead clearances, lights, utility wires and connections, garbage cans, benches, fences, guardrails and barriers, retaining walls, and line of sight vegetation obstructions. Aerial lidar will only be able to get gutter drains, road and local terrain information, walls, obstructions in the right of way, and sidewalks with limited ability to get curbs, lights, utility wires and connections, retaining walls, and vegetation.  Lidar does not do well for metal structures such as communications towers and fir these a combination of lidar and phodar is preferred.

Sanborn has developed feature extraction technology that extracts transportation objects from combined aerial and mobile data sets in a semi-automated manner. The mobile and aerial data sets meet the accuracy and information content required for geospatial information for mapping applications as well as the additional information that can be mined for potential asset inventory and infrastructure information content. Mobile technology allows for the low-risk, rapid collection of geospatial information, limiting safety impacts to workers. The aerial LiDAR allows for the rapid collection of elevation information for detailed surface modeling as well as feature extraction. Both technologies collect enormous amounts of point-cloud data that can be merged using the Terrascan capability in Bentley MicroStation. The combined data set is then passed through the automated feature extraction modules to identify transportation objects.  Sanborn feature extraction modules achieve a high proportion of correct feature extraction for road and highway scans.

In the future Sharad expects that it will be possible to improve data collection efficiency by using terrestrial robots or autonomous UAVs equipped with SLAM technology to enable them to operate in GPS-denied areas such as under bridges and in tunnels.  Sanborn has been working with deep learning to improve the automated extraction process.  So far this has yielded mixed results, but Sharad expressed optimism that with AI full automation of the classification process will be achieved.

High precision maps for autonomous vehicles

An application of this technology is high precision maps for the autonomous vehicle market.

Sanborn develops high precision maps for the autonomous vehicle market.  Autonomous vehicles will require high precision maps which contain significantly more detailed information and true-ground-absolute accuracy than current road maps. Sanborn has developed proprietary mapping technology that leverages aerial imagery, aerial lidar data, and mobile lidar data to create standardized, high-precision 3D base-maps focusing specifically on the self-driving vehicle market. These maps include detailed inventories of all stationary physical assets related to roadways such as road lanes, road edges, shoulders, dividers, traffic signals, signage, paint markings, poles, and all other critical data needed for the safe navigation of roadways and intersections by autonomous vehicles.

Intriguing new technology developments for improving construction productivity

Throughout media day at the Year in Infrastructure 2017 (YII2017) conference I was on the edge of my chair because there were such tanatalizing hints of spatial technologies that might address major challenges of 21st Century construction. We were inundated with a phenomenal amount of information about new technology developments and by examples of how these developments are actually being used by customers in the real world.  It is breathtaking to experience how rapidly new technology is being developed, rolled out and applied in the real world. This is happening in diverse disciplines as evidenced by the projects nominated for the roughly 20 different domains that are broken out for Be Inspired Awards categories at YII2017.

Today's presentations and the latest press releases absolutely intrigued me and I was tantalized by what I heard. The presentations stimulated many questions about the new technologies that Bentley was developing and I was very relieved to be able to sit down and chat with Francois Valois, Senior Director of Software Development Reality Modeling at Be‎ntley. Real world experience has shown that reality modeling is relevant to every part of the construction lifecycle. As a result reality modeling has grown rapidly in importance at Bentley and now has its own software development group. It joins four other groups which reflect the construction business cycle; design,  analytics, project delivery, and operations.  We chatted about a number of different topics that had been discussed during the day's sessions.

Comparing reality and design

During a highway construction project, estimating how much earth has been cut and filled is critical because it determines for a large part payments to contractors. With laser scanning and UAVs it is now becoming possible to scan an entire project with greater frequency. Converting the scanned point clouds to reality meshes produces a manageable 3D image of the status of the construction project at a given time. Seeing changes from one week to the next, for example, requires a tool that can compare two reality meshes and identify where differences exceed a specified threshold. Depending on the type of project and the 3D scanning technology this could be millimeters or centimeters. Bentley has developed a product called ContectCaptureEdit which allows two reality meshes to be compared. The difference between two reality meshes captured a week apart can be used to compute quantitatively the volumes of cut and fill on a highway or rail construction project in the last week. This approach can also be used to monitor construction progress for buildings and civil projects. Bentley's vision for the future is to enable comparing not only two reality meshes but also a reality mesh and a design model. This would automate the process of detecting divergences from design which is essential for assuring that what is designed is built.

Classification of reality meshes (aka feature extraction)

Feature extraction is a methodology that has been used for many years to identify and classify objects in 2D images captured by remote sensing using cameras on satellites or fixed wing planes and helicopters. In the utility sector, classification is still a manual process with some exceptions. The advent of 3D scanning technologies including lidar and phodar (capturing many photos with a consumer digital camera) makes possible the really differentiating focus at Bentley on classifying 3D objects. 3D classification can provide more information than 2D feature extraction because the extra dimension allows you to determine, for example, the depth of cracks in concrete or the extent of deformation damage to a transformer. Bentley is applying deep learning and other AI technologies provided by the Microsoft Azure cloud platform to enable semi-automated and even completely automated classification of reality meshes. For example, Bentley is working with a partner SiteSEE to automatically classify scans of communications towers captured by UAVs. With a smart UAV and automatic classification, the process of inspecting communications towers could be fully automated, saving a lot of time and increasing reliability by enabling more frequent, less expensive inspections.

Reality modeling for substations

As an example where automated classification is feasible and could provide significant benefits is in creating digital models of existing (brownfield) substations. Currently the process is quite laborious. The substation is scanned using terrestrial lidar or phodar with the option to include lidar or digital cameras on UAVs. The reality mesh generated from the scans is then classified manually – which involves selecting recognizable objects and identifying them as transformers, holding tanks, pumps, buildings, or other types of object. The final step is to link each object to an entry in an asset database – such as Bentley's AssetWise. Bentley refers to this process as “operationeering”. Operationeering makes it possible to use the classified scan plus the linked database – possibly together with augmented reality - to improve operational efficiency. The link between an object's location and the information about the object stored in an asset database esnsures that maintenance staff have immediate access to everything – location and condition - they need to know when there is a problem.  Or it could be used to simply conduct an inspection.

There are tens of thousands of brownfield substations in the U.S. If the classification step could be automated, and if QR codes are pasted to objects in the substation before scanning, the entire process could be semi-automated saving a tremendous amount of tedious and time-consuming effort.

Interoperability

Large construction projects invariably involve CAD, BIM and other types of products from multiple vendors. For example, Revit, Bentley BIM, Tekla, Sketchup, and others may be used on the same project. At YII2017 Bentley announced iModel 2.0 and iModelHub which enable consistent interoperability between different vendors' products. For example, changes made in Revit can be synced as “change sets” through iModelHub to products from other vendors. Bentley developed an iModel for Revit two years ago and iModelHub is a generalization of the concept to other design products. iModelHub is a complementery product for ProjectWise, which is a file based system widely used in construction for sharing project designs and related information.

Connected data environment enabling full lifecycle management

Perhaps the most tantalizing technology concept advanced during Greg Bentley's session was the connected data environment (CDE) which conceptually is intended to enable data to be reused throughout the entire lifecycle, As Francois explained it to me, it is designed to enable roundtripping data between ProjectWise, which is used during design and construction, and AssetWise, which is intended for optimizing asset performance during operations and maintenance. Reusing data collected during design and construction for operations and maintenance remains extremely rare. If CDE provides a foundation for doing this, it could represent a first step in resolving a major challenge hindering full lifecycle data management.

During today's sessions there were many tantalizing hints of the things Bentley is doing to address major business challenges like enhancing project collaboration, automatically classifying 3D scans, and enabling information sharing across the full construction lifecycle. I am anticipating further elucidation about these very exciting developments in the next few days at YII2017.

Geospatial imagery analytics market expected to grow rapidly

An very interesting report on the geospatial imagery analytics market in 2017 and its growth through 2022 has been released by MarketsandMarkets. The analysis looks at different types of imagery analytics (imagery analytics, video analytics), how the imagery is collected (GIS, satellites, UAVs), and how it is used by different vertical industries (defense & security, insurance, agriculture, healthcare & life sciences). The major players considered in the study include Hexagon AB, Harris Corporation, DigitalGlobe, ESRI, Google, Trimble, RMSI, Planet Labs, UrtheCast Corporation, Fugro N.V., KeyW Corporation, and Satellite Imaging Corporation.

The study estimates that the geospatial imagery analytics market was worth about US$ 3.41 billion in 2017. The video imagery segment is expected to grow at a higher rate than the image analytics segment. The study also sees a high growth rate for UAVs as a way of collecting imagery and video. Surprisingly for me, the healthcare and life sciences segment is projected to grow at the highest rate among different industry segments.  North America, followed by Asia and Europe, are expected to be the largest markets for imagery analytics in 2022.  MarketsandMarkets projects a high growth rate (31.1% CAGR) for geospatial imagery analytics reaching US$ 13.21 billion in 2022. Advances in artificial intelligence and big data are expected to be major contributors to this high growth rate. 

The analysis is based on a bottom-up approach by estimating the revenues of key market players. The data were verified through primary research by conducting extensive interviews with chief executive officers, vice presidents, directors, and other executives.

"Geospatial Imagery Analytics Market"

Do you really need a GIS to do spatial analytics ?

Geospatial technology - spatial data management and analytics - is being used in many industries outside of the traditional GIS sector. Analysts and IT folks in these industries may be surprised to find that the spatial analytics and data management they require is already available from their database management system whether it is Oracle, SQL Server, PostgreSQL, or SQLite

At the GIS in the Rockies conference in Denver Dave Murray from the City of Westminster reminded us how powerful the spatial analytics in modern relational database systems are.

In the late 1980s the first implementations supporting spatial data in a relational database were Vision* on Oracle RDBMS developed by Geovision and GFIS on Db2 by IBM. Subsequently the Open Geospatial Consortium defined a simple feature specification for spatial SQL. First Oracle and then Microsoft and the open source community have added spatial SQL based on the OGC standard to relational databases. Oracle Spatial, PostGIS+PostgreSQL, Microsoft SQL Server, and Spatialite+SQLite all support spatial SQL.

Folks wishing to employ geospatial techniques may find that they don't need a traditional GIS product and all the complexity involved in integrating a GIS with their database management system. They very likely will find that they can do all the geospatial processing they require using spatial extensions to standard SQL on their existing database. Finding the number of automobile accidents along a stretch of road, the census tracts a road passes through, or the burglaries in a neighbourhood are straightforward spatial SQL queries. The results can be visualized using Oracle's own visualization tools or open source applications like QGIS sitting on top of Oracle, SQL Server or PostGIS+PostgreSQL.  Dave mentioned spatial SQL courses on Udemy for those interested in learning more about spatial SQL.  Gordon Luckett's web site has a wealth of information about spatial queries, especially for integration with AutoCAD and Esri ArcMap.  Gordon also has developed courses on Lynda such as Using SQL Server with AutoCAD and Learning QGIS.

 

Open source geospatial application reduces copper cable theft by 70%

A highlight of the SA Geotech conference in Johannesburg was a presentation by Ryno Goosen of Suritec Geospatial.  Ryno, who was awarded the Best Paper prize at SA Geotec, gave a fascinating presentation about how he developed an open source GIS solution to fight copper cable theft.

In Canada copper cable theft costs the Canadian electricity sector an estimated $40 million annually.  In South Africa in 2014 there were 72,533 reported incidents of copper cable theft resulting in 10,736 arrests.  Copper theft incidents in the U.S. have tripled in the last 5 years.

Goosen Goosen's solution architecture was based on open source geospatial software.  He chose open source for several reasons.  First and foremost, it enabled massive scalability by providing an elastic IT infrastructure.  Secondly it provided open interoperability with other open source software as well as proprietary systems. He also emphasized the importance of using software that was "battle-tested" in both military and commercial environments. 

His solution architecture was built on Cesium, GeoServer, Boundless Desktop/QGIS, Protobuf, Cassandra, WildFly, Solr and GeoTrellis.  Using this solution he was able to combine data from different sources including ground patrol and remote sensors.  By analyzing this data geographically and temporally he was able to determine spatial and temporal patterns of copper cable theft.  This enabled optimum deployment of resources.  The results were impressive.  Over a 30-month period cable theft was reduced by 69%.

The success of Goosen's shows that open source GIS software has become a viable alternative for building complex and highly scalable solution-oriented architectures.  He also suggested that it levels the playing field for small to medium size enterprises by providing them with similar intelligence capabilities to their larger counterparts.

Are point clouds going away ?

At the SPAR3D 2017 conference in Houston, one of the interesting questions posed to a panel comprised of Greg Bentley (CEO of Bentley), Burkhard Boeckem (CTO of Hexagon Geosystems), and Shabtay Negry (Senior VP at Mantis Vision) is whether point clouds, whether from digital photography or LiDAR, will effectively disappear, being replaced by meshes for most reality modeling applications.  Meshes are mathematical constructs, typically 3D triangular networks, that are smaller and much easier to manipulate than point clouds, which are typically huge and a challenge to edit.
 
I blogged previously about a digital model comprised of meshes of downtown Philadelphia that was used to orchestrate and secure the Papal visit.  All the data was collected as photographs which were then used to create a single reality mesh using Bentley ContextCapture.  At SPAR3D there are presentations on two other projects which used a similar approach to create models of the city of Helsinki and the Cologne Cathedral.

Another fascinating example is the USS Arizona digital project which combined underwater lidar, sonar, aerial imaging, and existing photography and surveys to produce a survey-grade 3D model of the USS Arizona in the form of a mesh using Autodesk Remake.  Surveys were conducted in 2014 and 2016.  Just recently the two digital mesh models of the ship were compared to create a difference mesh to show areas where there has been change in the two years since 2014.

At SPAR3D there was a consensus among the panelists that meshes are replacing point clouds for most applications. Point clouds are just too big and have too little intelligence. They will remain as an intermediate data type because most scanners generate point clouds.  Replacing point clouds for many applications meshes will grow in importance in the future.  But realizing their full potential will require interactive, editable meshes that allow small bits to be changed or replaced without having to regenerate the entire mesh (which is often the case with triangulated surfaces).  Easy to use analytic tools like those used to create the difference map of the Arizona will also be required.  The panelists thought that with the appropriate edit and analytic tools people will be able to manipulate surfaces easily and may not  even be aware of what the form of the underlying data structures is.  Smaller files will also make cloud processing easier because in the future huge point cloud files won't have to be uploaded to the cloud.

Autonomous UAVs using SLAM to navigate without GPS

I have just spent an incredible three days in Houston at the SPAR3D 2017 conference.  There is so much going on right now in the 3D world with several key technologies just coming to fruition and finding broad application in the real world. 

 

One of them is Simultaneous Localization and Mapping (SLAM) technology.  In 2012 a company called GeoSLAM, which was a joint venture with Australia's National Science Agency (CSIRO), began marketing a SLAM-based solution developed by CSIRO.  I blogged about a backpack solution for indoor mapping that Avideh Zakhor from the University of California Berkeley had developed for navigating inside a building, generating floor plans, distinguishing rooms, identifying staircases, texture mapping surfaces, identifying windows, finding heat sources such as computers and human occupants, identifying lights, and even estimating plug load lower consumption - all automatically.

 

At this year's SPAR3D conference Stefan Hrabar from CSIRO described a project he has been working on that combines autonomous navigation with LiDAR-based SLAM to enable drones to safely navigate autonomously without GNSS in areas with complex geometrical structures.  This enables missions that the UAV can do with no pilot intervention.  Some examples that have been implemented include rainforest surveys - looking for invasive species, telco (cell phone) tower inspection and mapping, construction site monitoring including inside buildings, and underground mine mapping, especially, vertical, hard to access shafts called stopes.  For example, Stefan showed a demonstration of telco tower mapping where the UAV was placed near a telco tower, a button pushed to initiate the flight, and the drone mapped the tower without any human intervention.

 

Hovermap is the CSIRO application that provides a SLAM-based 3D LiDAR mapping payload for small UAVs.  It provides omnidirectional collision avoidance, advanced autonomy, and the ability to operate in GNSS-denied areas.  Hovermap uses a rotating LiDAR to produce a virtual spherical shield around the UAV.  Real-time SLAM makes it possible for the drone to navigate in complex geometric environments  without GPS, while the on-board autonomy makes possible mission execution without human intervention.  Some aplications include construction site mapping (inside buildings and outside), underground mine mapping and transmission line mapping and monitoring.