Software Defined Vehicles
Interview with Brett Harrison, TERN
“The system has to be accurate and hyper-continuously available”
As vehicles become increasingly software-defined and automated functions demand continuous reliability, localisation systems face new expectations for resilience and continuity. In this interview, Brett Harrison from Tern explains why legacy positioning concepts are no longer sufficient.
Positioning has long been treated as a functional layer within the vehicle stack — important, but largely assumed to work. Yet as vehicles transform into high-performance computing platforms and automated systems move closer to real-world deployment, that assumption is being re-examined.
Brett Harrison brings experience from both operational and large-scale technology environments. As Co-Founder and President of TERN, he focuses on positioning systems designed to operate independently of satellite signals and to strengthen infrastructure resilience. At Automotive Computing Conference US 2026, Harrison will address how positioning must evolve beyond GPS. In this interview, he explains why legacy localisation concepts are reaching structural limits.
ADT: Why are current localization concepts no longer sufficient for future vehicle functions?
Harrison: Society has become ever increasingly reliant upon turn by turn navigation, for everything from simple runs to the grocery store to Amazon deliveries, Uber pickups, and Ambulance dispatch. The skillset of map reading for directions has been effectively lost. If GPS were to have prolonged disruption 20 years ago, the impact would be manageable. With society’s addiction and widespread reliance upon turn by turn directions, the impact of a disruption today would be completely devastating. Everything from food supply transport, medicine delivery, Uber/Amazon/police dispatch would all be grinding to a halt. Simply stated, as time progresses it becomes ever more important to make sure we are not operating as a society upon a single point of failure (GPS).
Advertisement
For years, positioning was mostly used for simple driver turn-by-turn navigation. If it drifted or dropped for a moment, it was frustrating, but it did not impact core vehicle function. Now vehicles are software platforms. Location is used continuously by multiple systems and workflows, not just turn by turn guidance. That changes the requirement. It is not enough to be accurate when conditions are ideal. The system has to be accurate and hyper-continuously available throughout all the environments people actually drive in. Accuracy matters. But you cannot have accuracy if the solution is absent, so resiliency matters even more. The real gap is that many legacy approaches can deliver strong accuracy at times, but at many times throughout the journey the solution drops coverage entirely or experiences the erratic nature of multi-path error. As vehicle functions become more software driven, that inconsistency becomes a platform limitation, not a user experience issue. Now we are able to deliver an intelligent solution that is far more advanced and resilient than a tri-lateration solution.
How does removing hardware dependencies change robustness and scalability in automotive positioning systems?
Hardware dependencies create rigidity. They tie positioning performance to specific components, supply chains, and vehicle configurations, which makes global deployment slower and uneven. They also limit how systems can evolve over time, because improving performance often requires redesign rather than iteration. When positioning is delivered as software using sensors and compute that already exist in the vehicle, scalability changes fundamentally. The same capability can be deployed across model years and regions without reengineering. Performance can improve through updates rather than hardware cycles. And the system can remain operational even when external infrastructure is unavailable or inconsistent. That architectural shift is why software-only positioning has become a serious area of evaluation at the government and industry level. TERN’s Independently Derived Positioning System (IDPS) has been tested by the U.S. Department of Transportation as part of its work on complementary positioning approaches, reflecting a broader recognition that resilience and scalability increasingly come from software, not added hardware. At a platform level, removing hardware dependencies eliminates many of the constraints that have historically limited where positioning systems can be trusted and how widely they can be deployed.
What impact does this shift have on redundancy and safety concepts in automated driving?
It changes the conversation from backup to continuity. Redundancy is often discussed as adding another component in case the first one fails. But if all redundant systems share the same external dependency, that redundancy is fragile. And not truly a redundancy. A positioning system that can operate independently – turning vehicles from consumers of positioning data into producers of it – creates a fundamentally different safety posture. It provides continuity when other inputs are degraded or unavailable. That continuity allows automated systems to behave predictably, maintain situational awareness, and avoid abrupt disengagements. From a safety perspective, the goal is to finally have consistency under real-world conditions.
What implications does this have for software-defined vehicle architectures?
Software-defined vehicles depend on reliable foundational layers. Positioning is one of those layers. If it is inconsistent, everything above it becomes more complex, more constrained, and harder to validate. As architectures evolve, we are seeing positioning treated less like an application and more like a core infrastructural foundation – a foundation that can no longer be assumed. When delivered as software, it can be integrated deeply into the vehicle stack, shared across systems, and improved over time. That aligns naturally with how software-defined platforms are built and maintained.
Which architectural principles are shaping next-generation vehicle platforms today?
Advertisement
Three principles stand out. First, independence from fragile external dependencies. Second, leveraging existing vehicle components rather than adding manufacturing and design complexity. Third, systems that improve, learn, and evolve over time rather than age out. Intelligence is what makes the third principle possible. Not as a feature layer, but as the mechanism that allows software systems to adapt to real-world conditions, interpret imperfect inputs, and maintain continuity over time. In practice, Intelligence enables platforms to improve through use rather than requiring redesign, and to remain reliable across environments that were never predictable or controlled. These principles are showing up across compute, perception, and now positioning. Platforms that use Intelligence to strengthen software-first infrastructure are easier to scale globally, easier to update, and better suited for long-term evolution. The competitive advantage increasingly comes from systems that get smarter with use while remaining dependable in the environments vehicles actually operate in.
Where do legacy concepts still slow down scalability and long-term upgradability?
Legacy concepts tend to rely upon fixed infrastructure, fixed hardware, and fixed performance envelopes. That works until conditions change, and conditions always change. When systems require new sensors, specialized maps, or constant external signals to function, scaling becomes expensive and uneven. Upgrades become episodic rather than continuous. The result is fragmentation across fleets and markets. Modern platforms are moving away from that model in favor of software-first approaches that can adapt as requirements evolve.
Partnerships are becoming essential across the automotive computing stack. Which type of partnerships will matter most in the next three years, and what should OEMs prioritize first?
The partnerships that will matter most are technological partnerships, rooted in shared architectural thinking rather than short-term integration wins. As vehicles become software platforms, the real challenge is not adding more capabilities, but ensuring that foundational systems work reliably together over time. OEMs should prioritize partners who think in terms of infrastructure, not features. That means partners who design software to be broadly deployable, resilient across environments, and capable of improving through updates rather than requiring constant rework. These relationships tend to be longer-term and more strategic by nature, but they are driven first by technical alignment. Regulatory and commercial partnerships will continue to play an important role, but they tend to follow technical reality. When the underlying architecture is sound, scalable, and resilient, those conversations become easier. In the near term, OEMs that focus on building a coherent, software-first foundation with the right technical partners will be better positioned to adapt as requirements evolve and expectations continue to rise.
