Dublin-based startup Ubotica has teamed up with IBM to streamline the use of artificial intelligence (AI) applications on low-Earth-orbit (LEO) satellites. Ubotica, founded in 2016, offers its CogniSAT edge computing platform that enables AI inference to be performed directly on satellite data in space.

The partnership aims to address the challenge of costly and time-consuming data storage and processing on the ground. Both companies state that their mutual customers will be able to deploy their AI models directly to satellites using the CogniSAT platform with a single click, allowing for data insights to be generated in space.

This approach is expected to provide two key advantages. Firstly, it will enhance autonomy and decision-making capabilities at the edge, reducing reliance on ground systems. Secondly, it will result in significant cost savings for satellite constellation operators, both in terms of capital and operations.

By reducing the amount of raw data that needs to be downlinked for processing, capital expenses are lowered through savings in communications infrastructure. Furthermore, operating expenses are reduced by minimizing the bandwidth needed for communication of insights from onboard processing.

Ubotica will leverage IBM’s cloud infrastructure and watsonx.ai components as part of the partnership. It will begin by installing a Red Hat OpenShift Kubernetes Service (ROKS) cluster and building components to create a hybrid cloud AI platform.

The initial architecture will align with the underlying implementation of watsonx.ai, incorporating elements from the Open Data Hub. In the future, Ubotica may integrate additional IBM services like Watson Studio, Watson Open Scale, or tools from Cloud Pak for Data.

This collaboration solidifies Ubotica’s position in the global space tech race, building upon its previous successes. The startup has already provided AI solutions to the European Space Agency (ESA) and NASA JPS since 2020. This year, it achieved a significant milestone with the launch of CongiSAT-6, the first satellite to utilize space AI for autonomous image tasking and onboard processing of image sensor data to generate insights.