[[{“value”:”
TAMPA, Fla. — Californian software startup PiLogic announced $4 million in seed funding May 22 to develop satellite diagnostics and other artificial intelligence tools tailored for space applications.
The first satellite equipped with the venture’s electrical power system diagnostics model is slated to launch this year for an undisclosed customer, according to Johannes Waldstein, PiLogic’s founder and CEO.
The software would integrate onboard sensor data with detailed knowledge of engineering principles and the satellite’s parts, enabling the operator to predict component health and potentially resolve issues autonomously.
“It understands the cause and effect in a healthy system, and then it reasons to what’s broken or what’s wrong,” Waldstein said.
Validated with satellites on the ground through work with NASA, PiLogic’s “exact AI” model is designed to be faster and more accurate than restrictive rules-based systems and broader generative AI tools like ChatGPT that rely on large language models.
“You lose the satellite if [diagnosis] takes too long,” he added, “if you have to send information back down, and you have an operator trying to figure out … and then suddenly, either you lose contact, or it runs out of charge or doesn’t recharge.”
By running inference and remediation directly onboard, the AI gives operators a chance to prevent cascading failures.
“It will go to a backup battery or it will change sensors automatically,” Waldstein continued.
“And it will contact the operators saying: ‘I’ve done this, I think there’s a failure in this sensor, I think space debris has hit it, and now this portion of the satellite is not working — we should reroute around that.’”
The model is designed to run efficiently without massive datasets or graphics processing units (GPUs), requiring minimal onboard computing power. That makes it suitable even for even small cubesats, Waldstein said, although PiLogic sees most demand coming from customers with more complex, high-value spacecraft.
PiLogic has also developed an AI model that tracks and predicts objects in 3D space, designed to distinguish between entities such as airplanes and drones much faster than conventional methods.
Waldstein, a data science and AI innovator who has co-founded five startups over the past two decades, said PiLogic’s software is currently being tested by several large satellite operators and dual-use commercial/government companies.
The startup plans to use the seed funding to grow its engineering team, scale its software platform and accelerate integration with early customers.
Early-stage investors Scout Ventures and Seraphim Space led the funding round, joined by Sovereign’s Capital, Flex Capital, FN Fund and angel investor Gokul Rajaram.
“}]]
Source: Read More