Imagine your phone trying to run a heavy 3D game or a complex security camera feed — it just can't keep up. RAPID built a system that lets your phone secretly hand off the hard work to a powerful computer in the cloud, get the answer back, and you never notice the difference. Think of it like calling a friend who's great at math to solve a problem for you while you keep chatting. The system automatically finds the nearest available "helper" machine and decides in real time whether to compute locally or offload remotely.
Companies building applications for smartphones, tablets, robots, and other embedded devices hit a hard wall: the device doesn't have enough processing power, memory, or storage to run demanding tasks like real-time video analysis, 3D rendering, or AI inference. Today, the only option is to either accept poor performance on the device or rebuild the entire application for expensive high-end servers. There's no simple way to split the work between a cheap device and a powerful cloud machine automatically.
The project delivered a compute offloading platform with a task-based programming model, a runtime system that automatically decides whether to execute locally or remotely, and scheduling algorithms for managing multiple applications on shared cloud accelerators. The concrete output includes the first RAPID-based public acceleration service running on a real cloud infrastructure (SILO), plus 11 deliverables covering the full technology stack.
If you are a mobile gaming studio dealing with the limits of smartphone hardware — this project developed a compute offloading platform that lets resource-hungry games and AR experiences run smoothly on low-power devices by transparently shipping heavy tasks to cloud GPUs. The system was validated with a public cloud service across 8 consortium partners in 5 countries.
If you are an industrial automation company struggling with limited processing power on robotic controllers and embedded sensors — this project built a runtime system that automatically offloads compute-intensive vision and control tasks from edge devices to nearby accelerators. It supports heterogeneous hardware including virtual CPUs and GPUs, meaning your robots get smarter without needing expensive onboard hardware.
If you are a security technology provider running AI-based video analysis on resource-constrained cameras and edge devices — this project created a secure offloading service that moves heavy processing to cloud accelerators while keeping data protected. The 8-partner consortium built scheduling algorithms and admission control policies designed to handle multiple applications on shared infrastructure reliably.
The project received EUR 2,023,800 in EU funding across 8 partners over 3 years. Licensing or integration costs would depend on negotiations with the consortium. The project explicitly aimed to make the first public acceleration cloud service commercially exploitable, suggesting a service-based pricing model.
The architecture was designed for scalability — it supports multiple virtual CPUs and GPUs and uses scheduling algorithms and admission control policies to manage many simultaneous users. A public cloud service was demonstrated as a deliverable, though scaling to full commercial load would require further validation.
IP is shared among 8 consortium partners across 5 countries (Greece, Spain, Italy, Turkey, UK). The project explicitly targeted commercial exploitation of the public cloud service. Licensing terms would need to be negotiated with the coordinator, IDRYMA TECHNOLOGIAS KAI EREVNAS in Greece.
Unlike generic cloud services, RAPID provides automatic task offloading at the application level — the device itself decides what to send to the cloud based on real-time conditions. It also allows any device to act as both a client and an accelerator, creating a peer-to-peer acceleration network rather than relying on centralized data centers.
Based on available project data, the system targets applications in gaming, computer vision, security, robotics, and aerospace. The first RAPID-based public service was deployed on the public cloud infrastructure of consortium partner SILO as a demonstrated deliverable.
Security was a core design requirement — the project title itself includes 'Secure Multi-level.' The system implements secure offloading with registration mechanisms that control which devices can connect to which accelerators. Enterprise-grade security validation would depend on specific deployment requirements.
The RAPID consortium brings together 8 partners from 5 countries (Greece, Spain, Italy, Turkey, UK) with a balanced 50% industry ratio — 4 industry players alongside 3 universities and 1 research organization. This mix is strong for technology transfer: the academic side built the core algorithms while industry partners grounded the work in real deployment. Having 1 SME in the consortium suggests some startup-oriented commercialization interest. The coordinator is a Greek research foundation (IDRYMA TECHNOLOGIAS KAI EREVNAS), which is typical for EU research projects but means a business buyer would need to navigate academic tech-transfer processes.
Contact IDRYMA TECHNOLOGIAS KAI EREVNAS (FORTH) in Greece — a major Greek research foundation. SciTransfer can help identify the right person and facilitate an introduction.
Want to explore how RAPID's compute offloading technology could work for your devices? SciTransfer can connect you directly with the research team and help assess fit for your use case.
