HIVE

High Interoperability Virtual Environment for Humanoid Robots - Edge-Native Physical AI for industrial humanoid robots. Transforming RoBee into a cognitively autonomous, cloud-independent industrial assistant.

Overview —

About the project and BeeOS

The HIVE project addresses the primary bottleneck in industrial robotics: the inability to adapt to real-world variability without long reprogramming times or cloud latency and security risks.

Through HIVE, Oversonic develops and validates BeeOS (formerly known as the HIVE platform), an edge-native Physical AI operating framework embedded directly on-board the RoBee humanoid platform. Powered by BeeOS, RoBee delivers real-time perception, mission-level autonomy, and full data sovereignty.

Key highlights

  • < 20 ms Latency: Real-time on-edge AI processing (0% cloud dependency).
  • Mission-Based Autonomy: Guided by Vision-Language Models (VLM/VLA) to execute complex tasks without rigid scripting.
  • 100% Data Sovereignty: Fully compliant with strict industrial IT and cybersecurity rules.
  • Human-Centric Safety: Dynamic force feedback and collision avoidance for safe human-robot collaboration.

Industrial pilot

Semiconductor manufacturing use case

The HIVE project validates BeeOS in a real-world industrial environment at a leading semiconductor manufacturing plant within the Burn-In Board Cleaning process.

Key tasks executed by RoBee

  • Retrieval & Handling: Automated extraction of electronic boards from trolleys and machine door operations.
  • Process Workflow: Board insertion, flipping, oven drying transfer, and unloading.
  • Sorting: Dynamic sorting under variable object positioning.

Expected ROI & Impact

  • +200% to +300% increase in multi-task robot utilization.
  • 70-90% reduction in task reprogramming time.
  • €100k-€300k/year savings per robot by minimizing downtime.

Results & KPIs

BeeOS bridges high-level AI reasoning with low-level deterministic control, surpassing current State-of-the-Art (SoTA) standards.

  • Control Latency: < 20 ms (vs. 100-500 ms in Cloud AI)
  • Task Success Rate: > 85% under positional uncertainty (vs. 60-70% SoTA)
  • Safety Stops Reduction: > 40% via real-time force/torque feedback
  • Cloud Dependency: 0% (100% On-Premise Execution)

Events & Workshops

European Robotics Forum (ERF) 2027

  • Workshop: Physical AI: Advancing Edge Intelligence for Industrial Humanoids
  • Location: Birmingham, UK | Date: 2-4 March 2027
  • Focus: Interactive session on edge-native AI deployment and bimanual manipulation benchmarking.

Stakeholder Roundtable on NAMs

  • Workshop: Validation of New Approach Methodologies (NAMs) and Trustworthy AI in Restricted Environments
  • Location: Brussels, Belgium | Date: February 2027 (To be confirmed)
  • Focus: Policy and industry discussion on AI safety, ISO standards, and industrial deployment.
Latest updates —

News & Publications

  • September 21st, 2026 - Project Launch & EIC Portfolio Kick-Off Meeting: The HIVE project has officially launched! Oversonic Robotics participated in the official EIC Advanced Innovation Challenge Portfolio Kick-Off Meeting in Brussels. The event brought together EIC Programme Management and European project teams within the Physical AI thematic portfolio to align strategic goals, foster synergies, and set the stage for edge-native Physical AI development.
  • September 1st, 2026 - Project Kick-Off: The HIVE project has officially launched under the EIC Advanced Innovation Challenge. Oversonic Robotics is initiating the development and testing setup for the edge-native Physical AI framework.

Project information

  • Beneficiary/Coordinator: Oversonic Robotics S.r.l. Società Benefit (Italy)
  • Funding Program: Funded by the European Innovation Council (EIC) under the HORIZON-EIC-2026-AIC-01 call.
  • Project number: 101327261

Robotics for humans

© Oversonic Robotics Srl Benefit Company
VAT 11470500965
Privacy policy Cookie policy Corporate Governance Cookie preferences Vulnerability disclosure