Imagine a brain-like chip that understands sensor data instantly without needing to translate it into computer code first. Instead of sending huge amounts of data to a distant cloud server, the chip processes information right where it's collected. This makes devices react much faster and use far less battery power.
Current AI sensors waste massive amounts of energy and memory by converting analog signals to digital data. This creates bottlenecks in bandwidth and latency, making real-time cloud processing impossible for critical applications like self-driving cars.
A sensing-to-action computing architecture using VO2 phase-change materials and Mo/HfO2 RRAM devices to process analog data without conversion.
If you are an autonomous vehicle manufacturer dealing with delays in real-time decision making — this project developed a sensing-to-action computing architecture that allows energy efficiency improvements of 100x-1000x.
If you are a wearable medical device developer dealing with high power consumption and privacy risks of cloud processing — this project developed analog neuromorphic computing that processes data locally without analog-to-digital conversion.
If you are a robotics company dealing with bandwidth limitations in factory environments — this project developed an AI-on-Chip solution that enables prompt action from sensors with 100x-1000x better energy efficiency.
Based on available project data, specific pricing or cost-per-unit information is not provided.
The project focuses on the fabrication of low-power devices and architecture models; however, industrial scaling details are not explicitly mentioned in the summary.
Based on available project data, there is no specific information regarding the licensing terms or patent strategy.
It interfaces seamlessly with sensors to process analog data directly, removing the need for energy-heavy analog-to-digital conversion.
The project period runs from 2023-01-01 to 2026-06-30, suggesting the technology will be in development until mid-2026.
The consortium is highly balanced for technology transfer, consisting of 4 partners across 4 countries (NL, CH, DE, HU). With an industry ratio of 50% (2 industrial partners and 2 universities), the project ensures that the academic research into VO2 and HfO2 materials is aligned with commercial application needs.
Contact the Technical University of Eindhoven (TU/e) regarding the PHASTRAC project.
Contact SciTransfer to explore licensing opportunities for neuromorphic AI-on-Chip technology.
