At Geo Week 2026 in Denver, one of the industry’s most experienced voices in multi-sensor positioning argues that the real transformation isn’t in any single instrument — it’s in how the instruments learn to work together.
Mohamed Mostafa has been thinking about sensor fusion longer than most people in the geospatial industry have been using the term. As a senior technical leader at Trimble — the company whose Applanix division has been building direct georeferencing and inertial navigation systems for aerial, land, and marine platforms for decades — he has watched the UAV boom, the mobile mapping expansion, and now the convergence of all three domains into something genuinely new. His perspective at Geo Week 2026, shared alongside Anna Jarvis, Product Manager for Land Products at Trimble Applanix Corporation, was less a product briefing than a philosophical argument about where the industry is and where the hard problems still live.
The Corridor Mapping Moment
Trimble invested significantly in the educational dimension of its Geo Week presence this year, co-organizing a corridor mapping session that Mostafa describes as one of the more substantive gatherings of the conference. The session brought together a cross-section of the mapping ecosystem — an aircraft frame manufacturer, a payload manufacturer, a researcher, and working end users — to address corridor mapping for high-definition maps, utility documentation, oil and gas pipeline inspection, and related applications.
What made the session noteworthy, in Mostafa’s telling, was who showed up and why. An audience member asked why the phrase “corridor mapping” had drawn him in. The answer was practical and revealing: the expectation that self-driving vehicles will require continuously updated, high-definition maps of highway networks at linear scale — hundreds of miles or kilometers, repeatedly — means that corridor mapping, long the domain of specialized geospatial firms, is about to become infrastructure. “We always lived in that area mapping or wide area mapping all our lives as geospatial professionals,” Mostafa noted. The autonomous vehicle industry is coming to a problem the geospatial world has been solving for years, and it is bringing new scale and urgency with it.
The workshop Trimble hosted on Monday of the conference — a four-hour session that Mostafa organized and paid for out of his team’s budget — went further into the technical weeds. Among the invited speakers was a representative of Plowman Craven, a UK-based surveying firm, presenting a railroad mapping application that illustrates just how demanding the leading edge of the discipline has become.
Sub-Millimeter on the Railroad
Railroad inspection presents a particular challenge: certain sections of track infrastructure cannot be accessed by train-mounted systems, and boots-on-the-ground inspection at the required frequency and scale is impractical. Plowman Craven’s solution involves drone operations configured to achieve sub-millimeter positional accuracy — a fraction of a millimeter, or roughly one-hundredth of an inch — in service of change detection at the level of individual fasteners and joints.
The application is worth dwelling on. At that level of accuracy, repeat surveys taken months apart can detect whether a nut has rotated by a degree or two from its previous position. Combined with high-resolution imagery and AI-assisted comparison, the system can flag anomalies that would be invisible to a walking inspector and impossible to quantify without a prior baseline. “If the nut is rotated by a millimeter or two, you can capture it,” Mostafa said, “because you get into sub-millimeter position.” The AI layer compares the current scan to the historical record and identifies what has moved, deformed, or shifted — producing documentation with the evidentiary standards required for regulatory compliance and, where necessary, legal proceedings.
This is not a research demonstrator. It is a production system, built around Trimble positioning technology combined with Plowman Craven’s own custom software, because no vendor has yet assembled a complete commercial solution for this class of problem. The regulatory driver — railroad infrastructure compliance in the UK — is what makes the business case work, and it is the same pattern Mostafa sees across the most technically demanding corners of the geospatial market: compliance mandates create the economic justification for accuracy levels that would otherwise be difficult to sell.
Platform Modularity: The CapEx Argument
The convergence of aerial, land, and marine data collection is reshaping how serious geospatial firms think about capital investment. Mostafa’s framework is straightforward: rather than buying separate sensor systems optimized for each platform, the direction of travel is toward high-quality, modular sensor assemblies that can migrate across platforms — drone pod in the morning, wing pod in the afternoon, vessel-mounted by the end of the week — without sacrificing the accuracy that each application demands.
The economic logic is that a single high-quality sensor assembly, properly integrated across platforms, is a better investment than multiple cheaper systems each optimized for one use case. “You have a nice pod that has all the best sensors in the world, and then you can move it around in different platforms,” Mostafa said. The capital outlay is higher upfront, but the ongoing cost of maintaining accuracy, calibration, and regulatory compliance across a multi-platform operation is lower than repeating that investment for each domain.
This is not a simple engineering problem. The challenge Mostafa identifies is in the geometry: when a sensor assembly is fixed, the vector from the GNSS antenna to the sensor center is a constant — calibrated once, stable indefinitely. When the platform or the sensor orientation changes, that vector changes, and the design matrix for position computation has to account for a moving reference frame. “When you’re moving the camera around, you’re moving the vector connecting that GPS distance from the GPS antenna to the camera center,” he explained. Constant recalibration, or sophisticated dynamic modeling, is the price of flexibility — and it is part of why truly platform-agnostic systems require more R&D investment than marketing materials typically convey.
On the regulatory side, the manned aircraft world provides a useful reference point. Certifying a sensor payload for installation on a manned survey aircraft requires an STC — Supplemental Type Certificate — a process with defined protocols and significant lead time. The drone world has not yet developed equivalent institutional infrastructure, which is simultaneously an opportunity and a source of risk for firms trying to move quickly.
The Land Perspective: Augmentation Over Replacement
Jarvis addressed the sensor fusion challenge from the ground up. For mobile mapping on roads and in urban environments — tunnels, tree cover, intersections with long signal interruptions at traffic lights — the problem is maintaining positioning integrity when GNSS degrades or drops. The answer is augmentation: using the data already on the platform to bridge the gaps and keep the trajectory solution robust.
“Sensor fusion is definitely the path that everyone’s looking at to get that better performance in difficult environments,” Jarvis said. The sensors in question are ones many platforms already carry — LiDAR point clouds that can yield relative motion estimates, cameras that provide visual odometry, inertial systems that maintain continuity through short outages. The question her customers increasingly bring is how to extract more value from the sensor stack they already own: “How can we use the data we already have on our equipment to augment the performance that we’re getting, either in real time or in post-processing?” The goal is not to replace GNSS but to ensure that the overall system degrades gracefully when any single input is compromised — and that the solution is as robust on a congested urban street as it is in open sky.
The land and aerial worlds are also converging operationally, not just technically. Firms that built their business around terrestrial mobile mapping are adding drone capability; UAS operators are diversifying into ground-based collection. “Worlds are colliding,” Jarvis observed — a point Mostafa reinforced with the observation that companies are now actively working toward landing drones on moving vehicles, integrating the two collection modes within a single operational workflow.
RTK as a Service and the Business Model Evolution
One of the more consequential shifts in the positioning market over the past several years has been the emergence of subscription-based accuracy — the ability to activate higher-precision GNSS corrections for the duration of a specific project and pay accordingly, rather than acquiring and maintaining permanent base station infrastructure. Trimble’s RTX correction service, applied in products like the APX RTX for UAV georeferencing, delivers centimeter-level accuracy over-the-air without a local base, priced as a service rather than capital equipment.
The significance goes beyond cost structure. By making high-accuracy positioning accessible on a per-job basis, the model extends the addressable market for precise geospatial data collection to organizations and projects that couldn’t justify the permanent infrastructure investment. And the quality of the data collected at capture time — the georeferencing accuracy built in at the start — propagates through every downstream deliverable. Higher-quality data from the beginning means fewer revisits, cleaner processing, and more defensible outputs. The business model innovation and the accuracy innovation are not separate stories; they are the same story told from different angles.
What Comes Next: Real-Time at Scale
Mostafa closed with a reference that captures the direction he sees the field heading. Among the drone datasets Trimble processed in a recent RTX performance evaluation — close to a thousand — one company stood out: Neo, a Norwegian firm led by Tron Bjørnstad, who serves as both CTO and CEO, building custom multispectral sensors for mining applications and flying them with real-time RTX embedded in the workflow. The output is a fully processed, georeferenced multispectral map produced in real time, in the field, at centimeter accuracy.
“From the generation when we used to have 500 feet real-time accuracy of GNSS before we removed selective availability,” Mostafa said, “now, when you see the maps producing in real time at centimeter accuracy — like science fiction.” The distance from selective availability, which the U.S. government eliminated in 2000, to real-time centimeter mapping is roughly 25 years of compounding improvement in receivers, corrections infrastructure, inertial systems, and processing algorithms. Neo represents a current example of what that compounding makes possible when all of those elements are assembled by people who understand each one deeply.
The multi-domain integration story — air, land, and marine data fused into a single coherent model — points toward a follow-on conversation. Michael Kot, CTO at MJ Engineering, presented at Trimble’s Geo Week workshop a workflow that integrates airborne, terrestrial, and underwater data into a unified deliverable, assembling components from vendors across multiple countries because no single vendor has yet built the complete pipeline. It is the kind of integration problem that defines where the industry is headed and where the next generation of commercial solutions will need to go — and one that warrants its own treatment.