Oceanus12 has a twin-electric drive for optimum redundancy, and a hybrid powertrain for resilience and efficiency, coupled with Marine AI’s GuardianAI autonomy software stack.

Discover Guardian AI's revolutionary approach to tactical security with AI-driven solutions tailored for your safety. Experience real-time intelligence, customized protocols, and 24/7 monitoring for unparalleled protection.

**Abstract:** This paper presents a novel framework for enhancing the safety of autonomous vehicle (AV) interactions with vulnerable road users (VRUs) – specifically wheelchair users and parents with strollers – through proactive risk assessment and adaptive trajectory planning. Our system, 'GuardianAI', leverages multi-modal sensor fusion (LiDAR, camera, radar) combined with Bayesian optimization to predict VRU behavior and dynamically generate safe and compliant trajectories.

**Abstract:** This paper investigates a novel approach to ensuring autonomous weapon system (AWS) compliance with International Humanitarian Law (IHL) through a Deep Reinforcement Learning (DRL) framework incorporating human-guided policy refinement via adversarial domain adaptation. Traditional DRL methods struggle with generalizing across diverse operational scenarios and adapting to unforeseen IHL complexities. Our proposed system, termed “GuardianAI,” addresses this by utilizing a human-in-the-loop process where expert IHL analysts provide targeted feedback, which is then integrated into the DRL agent's policy through an adversarial domain adaptation strategy.

Hey Devs! I recently participated in the "Hack the Future: A GenAI Sprint" Hackathon and built a solo project called GuardianAI — a lightweight multi-agent system that supports elderly care using AI. I’m sharing my journey in this post — from idea to last-minute submission, everything I learned, and how it helped me grow as a builder. If you're into AI, Python, or hackathons, this one's for you. The Problem: How can we use GenAI to support elderly individuals in managing their health, daily routines, and emergencies? With rising concerns about elderly safety and independence, I found this challenge both meaningful and technically interesting. My Solution: GuardianAI I built GuardianAI, a command-line AI system powered by three agents working together: 💓 Health Monitor Agent – Analyzes vitals like heart rate, BP, and flags anomalies. ⏰ Reminder Agent – Displays personalized daily tasks like medication and hydration. 🚨 Alert Agent – Collects alerts and shows emergency warnings. Each agent uses real-time data from CSV files and communicates via Python functions — a clean, modular architecture. 🛠️ Tech Stack Python 3.11 pandas CLI (Command Line Interface) GitHub (for version control + hosting) YouTube + OneDrive (for final demo & slides) I kept it simple — no flashy web app, just a clear proof of concept. 📂 GitHub Repo Full code, agents, data files, and PPT: 🔗 GitHub – [https://github.com/AstutiJ/-AstutiJ-Elderly-Care-AI-Hackathon-Project/commit/bb494faeaea79f4a45b04dc1c90e3cc8b7819967] 💭 What I Learned Start, even if it’s messy or rushed. Build for impact, not just for flash. Agent-based thinking is powerful in AI architecture. Solo building can teach you more than team projects sometimes! 💚 The Bigger Win Even if I don’t win a prize, this project gave me real confidence. It re-sparked my motivation to continue my 90 Days of DevOps blog series — which I had paused due to illness. This blog marks my comeback — and I’m super excited to keep building, learning, and sharing. 👋 Final Thoughts If you’ve ever felt like you're not "ready" for a hackathon, I promise — you’ll learn more by showing up and trying than by waiting. Thanks for reading, and feel free to check out my project repo, video, or connect here!