Enabling AI-Driven Research at RCCL with AWS

The Robotics, Automatic Control, and Cyber-Physical Systems Laboratory (RCCL), based at the National and Kapodistrian University of Athens, focuses on advancing robotics, cyber-physical systems, and IoT research. By merging robotics and computational sciences, RCCL leads projects that address both theoretical challenges and practical applications in automated control and sensor-based systems. Through this AWS migration, RCCL has taken a significant step toward ensuring that its data and computational requirements align with the growing complexity of its research initiatives, enabling greater agility and collaboration across its research activities

RCCL had an established on-premises infrastructure that supported theirresearch efforts in robotics, automatic control, and cyber-physical systems. Astheir work evolved, however, the team identified new requirements that wentbeyond what the current setup could offer — particularly in areas involving realtime data processing, large-scale machine learning workloads, and the use ofgenerative AI models.

The laboratory aimed to improve access to high-performance computingenvironments, reduce the time needed to prepare research environments, andintroduce greater flexibility for scaling experiments. They also wanted to expandtheir ability to process and analyze data from IoT-connected robotic systems andenable researchers to collaborate more efficiently through shared environments.

To support this direction, RCCL sought to modernize their research platform byadopting a secure, scalable cloud-based solution tailored to their growing technical and scientific needs.

PCG designed and implemented a AWS-based cloud platform tailored to RCCL’s research needs. The solution brought together advanced computing, machinelearning, IoT integration, and analytics capabilities under a unified, scalable architecture.

At the foundation, a secure Amazon VPC was set up across multiple availabilityzones to ensure high availability and fault tolerance. GPU-enabled EC2 instanceswere deployed for MATLAB environments, with Auto Scaling groups dynamicallyadjusting the compute capacity based on workload demand.

To support AI and machine learning research, Amazon SageMaker Studio was integrated, providing 10 dedicated workspaces for data scientists to build, train, and deploy models. Amazon Bedrock was leveraged to give researchers access to leading foundation models for natural language processing and AI experimentation. Amazon OpenSearch Service Serverless powered semantic vector searches and large-scale text analytics for the research projects.

Real-time IoT data ingestion and processing was enabled through AWS IoT Core, connected to on-premises PLC S7 devices. Events from IoT sensors were processed using AWS Lambda functions, with centralized data storage in Amazon S3, creating a secure, scalable data lake. Analytics dashboards were built with Amazon QuickSight, allowing researchers to visualize and explore their data in real time.

The environment also included a Landing Zone managed by AWS Control Tower, setting up organizational governance, centralized security (AWS GuardDuty, AWS CloudTrail), and efficient user access via AWS Identity Center (SSO).

Monitoring, security, and disaster recovery were incorporated from the start, using Amazon CloudWatch, backup solutions, and cross-region replication practices where needed.

AI/ML Services in Focus:

• Amazon SageMaker Studio – Model building, training, deployment in isolated workspaces
• Amazon Bedrock – Access to foundational models for experimentation
• Amazon OpenSearch Serverless – Semantic vector search for research papers and sensor data
• AWS Lambda + IoT Core – Real-time data triggers for robotic feedback loops

The AWS-based platform fundamentally transformed RCCL’s research operations. Researchers now access high-performance compute environments in minutes, not hours or days. Machine learning projects, including generative AI experimentation, scaled without the need for heavy infrastructure investments.

Real-time IoT data streams and analytics enable faster experimentation cycles, helping researchers move from data collection to insights much more quickly. By shifting from static on-premises infrastructure to a flexible, cloud-native model, RCCL improved operational efficiency, reduced system management overhead, and significantly expanded the laboratory’s research capabilities.

The architecture was designed to scale with future growth, ensuring that as RCCL's research programs evolve, the underlying platform can grow alongside them.

Key outcomes from the AWS AI solution included:

• Faster research cycles through reduced environment setup and model training times, achieved with GPU-enabled SageMaker workspaces
• Generative AI experimentation enabled via Amazon Bedrock, supporting advanced exploration in robotics and automation
• Enhanced data discovery using Amazon OpenSearch Serverless for semantic search and large-scale text analysis
• Real-time IoT integration allowing continuous ingestion and processing of sensor data from robotic systems
• Scalable machine learning workloads without infrastructure bottlenecks, enabling more ambitious and iterative research
• Reduced operational overhead, freeing researchers to focus on AI model development and scientific outcomes

Public Cloud Group (PCG) supports companies in their digital transformation through the use of public cloud solutions.

With a product portfolio designed to accompany organisations of all sizes in their cloud journey and competence that is a synonym for highly qualified staff that clients and partners like to work with, PCG is positioned as a reliable and trustworthy partner for the hyperscalers, relevant and with repeatedly validated competence and credibility.

We have the highest partnership status with the three relevant hyperscalers: Amazon Web Services (AWS), Google, and Microsoft. As experienced providers, we advise our customers independently with cloud implementation, application development, and managed services.