260915 KOMSA AI어선안전설계데이터플랫폼 3화 진수식 송출용

Key technological concepts & product features (AI fishing boat design platform)

Problem addressed

• Designing a fishing boat can take up to 2–3 months for skilled designers to estimate size/weight and produce drawings based on existing blueprints and operational characteristics.
• The hardest part is calculating performance (e.g., safe speed).
• Verification is difficult because specialized software requires significant computation time.

Solution

• An AI-based fishing vessel design system (referred to as “KIVE” / “KOMSA AI”) that enables:
  • Data-driven design
  • Performance calculation using numbers instead of intuition/experience

Core functions (two main tasks)

1. Real-time performance estimation

   • Example outputs include speed and insurance/safety-related metrics.
   • Uses user-selected modules as performance objects.

2. Automatic design derivation

   • Generates a vessel design that satisfies user requirements, such as:
     • Performance priorities
     • Additional installation requirements

User input & output behavior

• Users provide:
  • Priority criteria (e.g., speed, fuel efficiency, safety)
  • Other vessel installation requirements
• The system then:
  • Selects three vessel design types/options that fit the conditions
  • Presents an optimal solution
  • Communicates safety of the chosen design to fishermen
  • Provides design firms with best-fit vessel configurations

---

Training data & modeling approach

Large-scale training data requirement

• The AI relies on extensive fishing vessel performance data to build a predictive model.

How data is collected

• A research team visits real fishing sites (streams/sea) to gather information on:
  • Vessels
  • Operational needs
• A module development team builds modules such as:
  • Upper hull structure
  • Equipment

Scale of data described

• Approximately 640,000 (units/cost referenced as “won”) worth of engineering data.
• Includes 156 modules of wave conditions for “East and West siblings” (likely regional variants).
• Includes vessel-specific operating conditions.
• Includes performance records for combinations of modules.

AI modeling capability

• Instead of manually building prediction models one dataset at a time (which requires AI experts), the platform supports:
  • Automatic generation of a performance prediction model from new performance data
• The system claims immediate calculation of speed and safety during design.

---

Performance claims & efficiency benefits

• Accuracy: Reported ~95% accuracy vs. expert manual results.
• Efficiency: Performance prediction occurs simultaneously with design, reducing unnecessary steps and improving workflow efficiency.
• Verification speed: Learned data and automated algorithms avoid slow expert verification cycles.

---

Usability / “barrier to entry” focus

• The video emphasizes that even excellent technology must be easy to use.
• The platform lowers adoption barriers by offering easy, accurate operation for users anywhere.
• Outputs are tailored to what fishermen value most: speed, safety, cost.

---

Positioning / philosophy

• The system does not replace experience; it supports engineering figures that are difficult to verify “with the naked eye.”
• The goal is to bridge the technology gap and connect traditional fishing environments to a “smart future” and safer seas.

---

Main speakers / sources (as indicated in the subtitles)

• KIVE / KOMSA AI fishing vessel design system (product/organization being described)
• Research team (collecting data and developing modules)
• Module development agency/team (developing modules such as hull structures/equipment)