Tern AI: Affordable GPS Alternative That Delivers

The Limitations of Traditional GPS Technology

Many drivers have encountered the annoyance of a malfunctioning GPS, leading to missed highway exits. Tern AI, a company developing a cost-effective alternative to GPS, attributes these issues to the inherent limitations of current systems that depend on satellite positioning.

A Novel Approach to Vehicle Localization

Tern AI asserts it has developed a method for pinpointing a vehicle’s location utilizing only map data and the vehicle’s pre-existing sensor suite. This system is presented as an inexpensive solution, eliminating the need for additional, costly sensors.

During a demonstration at SXSW, the Austin-based startup showcased its ability to determine position “from nothing” exclusively for TechCrunch.

Breaking Free from Triangulation

“We don’t employ triangulation, satellites, or Wi-Fi; we simply ascertain our location while driving,” explained Brett Harrison, co-founder and president, as Cyrus Behroozi, a senior software developer at Tern, initiated the demo on an iPhone. “This represents a significant advancement, as it allows us to move beyond the constraints of triangulation-based technologies.”

The Broad Implications of Accurate Positioning

Harrison emphasizes the importance of this breakthrough for several reasons. Commercially, businesses reliant on GPS – including ride-sharing and delivery services – experience financial losses and wasted resources when drivers must retrace their routes due to inaccurate GPS data.

Furthermore, critical infrastructure – encompassing aviation, disaster relief, and precision agriculture – heavily depends on GPS functionality. The potential for adversaries to disrupt GPS signals has been demonstrated, posing substantial risks to both economic stability and national security.

Government Initiatives for PNT Resilience

The United States government has indicated a commitment to prioritizing alternatives to GPS. President Donald Trump’s executive order during his first term aimed to reduce dependence on single-source PNT (positioning, navigation, and timing) services, such as GPS. Additional initiatives are underway to ensure robust PNT capabilities through testing and integration of non-GPS technologies within agencies like the Department of Defense and the National Security Council.

Cost-Effective AI Development

“DeepSeek achieved results comparable to those requiring billions of dollars from companies like OpenAI and Anthropic for a fraction of the cost,” Harrison noted. “The government invested billions and decades to achieve real-time map positioning in vehicles. We accomplished this with under $2 million.”

Rapid Development and Seed Funding

Tern AI emerged from stealth mode in February 2024 and subsequently announced a $4.4 million seed funding round a few months later. This swift progress enabled the demonstration of their positioning technology experienced at SXSW this week.

Tern AI’s approach offers a potentially transformative solution to the vulnerabilities and limitations of traditional GPS systems, promising increased reliability and security for a wide range of applications.

Testing Tern’s System in Austin

The demonstration began with Behroozi establishing a Bluetooth connection between his 2019 Honda Civic and his smartphone. This allowed the Tern application to access data from the vehicle’s pre-existing sensors. It was highlighted that Tern’s technology is capable of direct integration into vehicles manufactured from the 2009 model year onwards.

Typically, Tern employs manual position setting for expedited testing. However, for this particular demonstration, the team opted for a “cold start” scenario. Behroozi disabled location services on his phone, challenging the Tern intelligent system to rely solely on a cached map—covering a 500-square-mile area surrounding Austin—and the vehicle’s sensor data.

As the vehicle was driven, the system actively collected road data, working towards achieving “convergence.” Full convergence from a cold start required approximately ten minutes, a timeframe Behroozi attributed to the presence of traffic, which restricted the vehicle’s movement. Harrison clarified that convergence generally occurs within one to two minutes without a starting point, and is instantaneous when one is provided.

Harrison emphasized the system’s ability to accurately localize vehicles in challenging environments. These include parking garages, tunnels, and mountainous regions, where GPS performance is often compromised. The specific methodology behind this capability remains confidential, described by Harrison as “proprietary” information.

Following the achievement of full convergence, a further drive allowed observation of the system’s consistent and precise tracking of the vehicle’s movements. The accuracy appeared comparable to, and at times exceeding, that of GPS. This was particularly noticeable during navigation through downtown Austin, where my Google Maps application frequently reported inaccurate location data throughout the week.

Harrison also pointed out the enhanced privacy benefits offered by Tern’s system.

“The system operates as a completely self-contained loop,” he explained. “Currently, no data is transmitted externally. Position is determined independently through on-device processing, eliminating any external dependencies.”

Designed for Expansion

From its inception, the company and its solution were engineered with scalability as a core principle. Harrison explained that the advanced hardware found in a Waymo vehicle is unlikely to be integrated into more common models like a Nissan Sentra due to cost considerations.

He highlighted a Waymo-Uber robotaxi as an example, emphasizing the current economic realities of such technology.

However, implementing Tern within a vehicle’s infotainment system represents a straightforward software download, offering exceptional scalability.

Modern vehicles are already equipped with the necessary sensor data, and existing map data is readily available from current providers. This simplifies the integration process considerably.

Potential Customer Base

Tern’s prospective clientele is diverse, encompassing automakers, mobile phone manufacturers, and technology giants like Google and Uber.

Harrison indicated the startup remains open to both independent growth and potential acquisition opportunities.

The overarching goal, he stated, is widespread adoption of the technology to address emerging threats and unlock the full capabilities of systems currently constrained by the limitations of triangulation.

Government Collaboration

Tern is actively pursuing collaborations with government entities.

Recently, the company was awarded a contract by the U.S. Department of Transportation following a successful demonstration of its technology alongside nine other international companies.

“We aim to showcase the advancements achievable through American innovation,” Harrison commented, expressing optimism about the demonstration’s impact.

Note: This article has been revised to provide clarity regarding the typical convergence time of Tern’s system and to remove ambiguous details concerning GPS functionality.