Launch of Collaboration with Liquid AI to Develop Edge AI Solution

Aiming to enable machine learning with minimal power consumption and improve processing performance on edge devices.

ITOCHU Techno-Solutions Corporation (headquartered in Minato-ku, Tokyo; Ichiro Tsuge, President and CEO; hereinafter “CTC”) has launched a collaboration with Liquid AI, Inc, (headquartered in Massachusetts, USA; Ramin Hasani, Co-founder & CEO; hereinafter "Liquid AI"), a Massachusetts Institute of Technology (MIT) spin-off start-up, to develop edge AI solutions. The goal is to improve processing performance on edge devices by leveraging Liquid AI's machine learning methods such as Liquid Neural Network (LNN), which enables highly adaptive machine learning with minimal processing power.

In recent years, plenty of attention has been focused on edge AI, which enables immediate processing of data acquired by edge devices, such as cameras and IoT sensors, with computers mounted on the devices. By placing AI close to devices, edge AI allows making analysis and decisions in real time, while reducing communication costs with servers. It is therefore expected to be used for self-driving ground and aerial vehicles, customer behavior analysis with in-store cameras and anomaly detection in factories.

The collaboration aims to develop edge AI solutions that process large volumes of data in real time at the edge by utilizing Liquid AI's machine learning technology. In particular, LNNs are expected to be used for autonomous navigation of drones and vehicles because the model can flexibly learn adaptation to unknown environments and unexpected situations not included in previously learned data, better than other modern alternative AI solutions. Today’s machine learning models require large neural networks with millions of parameters to navigate ground and aerial vehicles from high-dimensional visual data. LNNs on the other hand, could perform the task by 1 to 3 orders of magnitude smaller number of parameters. The small size and the algorithmic efficiency of LNNs allow them to run on edge devices or small computers installed close to users. As a result, it can reduce the AI infrastructure cost proportionally, leading to reduction of power consumption and CO2 emissions.

CTC will utilize AI of Liquid AI to develop state-of-the-art edge AI solutions for robust and autonomous navigation of ground and aerial vehicles. We perform this by combining Liquid AI’s technology with our expertise obtained through providing camera-based edge AI solutions and building data analysis infrastructures.

CTC launched the NAPP (North America Partnership Program), an initiative to build stronger partnerships with overseas start-ups to co-create businesses with them, in April 2023 and this collaboration is part of the initiative.

Going forward, CTC will continue collaboration with Liquid AI to conduct PoC for utilizing LNN, and technical verification for services including data analysis and saving power to run large-scale AI systems.