Tern AI’s low-cost GPS alternative actually works

We have all experienced this moment of frustration when the GPS disorders miss an outcome on the highway. The team at TearThe establishment of an inexpensive GPS alternative says that the current technology is limited by its dependence on satellite positioning.

Tern Ai says that it found out how the position of a vehicle can only be localized with card information and the existing sensor data of a vehicle. The company’s pitch: It is a cheap system for which no additional expensive sensors are required.

At SXSWThe startup based in Austin demonstrated exclusively for Techcrunch that it could “derive a position from nowhere”.

“No triangulation, no satellites, no WLAN, nothing. We only find out where we are while we drive, ”Brett Harrison, co-founder and president, told Techcrunch, while Cyrus Behroozi, senior software developer at Tern, who invited demo on his iPhone. “This has really changed, because if we move away from triangulation -based that restrict the technology, we can now have the ability to be completely this network.”

Harrison says that this breakthrough is important for several reasons. From a commercial point of view, companies that are abandoned on GPS-in one way-in one way of driving hail apps to delivery company, every time, money and gas if their drivers have to return due to incorrect GPS position.

It is even more important that our most critical systems such as aviation on disaster reaction to precision breeding on GPS are based. Foreign opponents have already shown that they can Spoof -GPS signalsWhat catastrophic effects could have on both the economy and national security.

The United States has signaled that it wants to prioritize alternatives to GPS. In his first term, President Donald Trump signed one Executive order To reduce trust in a single source of PNT services (positioning, navigation and timing) such as GPS. There are also several other initiatives Which direct agencies and bodies such as the Ministry of Defense and the National Security Council to ensure a resilient PNT by testing and integrating non-GPS technologies.

“Deepseek came out and said it cost 6 million US dollars to do what it needed [OpenAI, Anthropic, and other AI companies] To do billions, ”said Harrison. “To move this point in real time in a vehicle in real time, the government needed billions of dollars and a few decades. We did it with less than $ 2 million. “

Tern came from Stealth in February 2024 and announced it 4.4 million US dollars seed round A few months later. This is a quick handling to achieve the type of positioning that I experienced this week at SXSW.

Test from Tern’s system in Austin

In order to start the demonstration, Behroozi Associated his Honda Civic 2019 via Bluetooth with his phone, so that the Tern application was able to access data from the existing sensors of the vehicle. He found that Tern’s technology can be integrated directly into vehicle models from 2009.

Tern usually uses the position manually to speed up things, but the team wanted a “cold start” for our demo. Fehmroozi switched off the location services of his phone, so that the medical intelligent system only had an intermediate card of a 500 square miles around Austin and Vehicle sensors with which they can work.

When the car drove, the system started road data to work towards “convergence”. It took about 10 minutes for the system to be fully converged by a cold start, but Harrison assured me that it usually takes about one to two minutes if it has a starting point. In addition, we stuck in a bit of traffic, which slow things, added Fehri.

Harrison noted that the system of Tern can also localize vehicles in parking garages, tunnels and mountains that have difficulty GPS. Harrison would not explain exactly how the information “proprietary” says that the information.

We drove a few more minutes after the system had reached the full convergence, and I watched as our exact movements pursued in a way that seemed better than GPS in some cases. This became clearer when we drove into the city center of Austin, where my Google cards regularly misuse me during the week when I navigated with high -towering buildings.

Harrison said that the system of Tern from the perspective of data protection is also safer because “If someone knows their ID, they can find them at any time.”

“Our system is a total closed loop,” he said. “At the moment we don’t emit anything. It derives its own position independently of one another [via on edge computing]So there are no external points of contact. “

Scaled

“We have set up the company and the solution scalable from the start. If you look at this Waymo car and the entire hardware that is embedded, we do not see that this will soon be possible in a Nissan Sentra. It’s just too expensive. Said Harrison and pointed to A in front of us Waymo-Uber Robotaxi.

“At the manufacturer level, if [Tern] is implemented in the infotainment system, it is just a software Download, so extraordinarily scalable. All new vehicles have the sensor data we need. The card data already exist with all providers today. So it’s very simple. “

Tern’s potential future customers could be anyone, from car manufacturers to mobile phone manufacturers, from Google to Uber. Harrison said the startup was open to the cultivation of the company, but also an acquisition.

“The main thing is to put this into the hands of the economy, with the growing threats and the development of technology, which, due to the limits of triangulation, do not recognize the full potential,” said Harrison.

He noticed that Tern examined opportunities with the government. The startup recently received a contract award by the US Transport Ministry after a week of proof of his technology together with nine other companies from all over the world.

“We hope that we have done a good job for the government, which is now possible with American innovation,” said Harrison.