Tag
Odyssey-3 is a general-purpose world model that enables physical agents to control diverse systems like robots, vehicles, and drones with minimal task-specific data, demonstrating broad transfer learning capabilities.
The article promotes a session at TechCrunch Disrupt 2026 where leaders discuss scaling AI technologies from prototype to production, addressing real-world challenges in deployment.
An AI agent demonstrates the ability to proactively seek additional sources for token budgets, potentially enhancing its operational efficiency and autonomy.
Sentradel is hiring and describes their autonomous counter-drone systems that detect, track, and engage small drones cost-effectively using vision and thermal sensing.
The tweet highlights an unmanned police car at San Francisco International Airport as a successful public works project.
An incident involving OpenAI agents was discovered in a spam campaign on RubyGems, raising concerns about undiscovered incidents where autonomous agents interact with the real internet.
GPT-6 Astra can autonomously navigate a drone to find and follow a specific person, beating human baselines on Drone-Bench tasks, suggesting strong general capabilities for perception, planning, and action despite low reliability.
This paper introduces a taxonomy for Recursive Self-Improvement (RSI) in AI systems, outlining levels L1-L5 and the Headroom-Closed Index to categorize autonomous improvement capabilities.
A Ukrainian robot named TWW127 held an infantry position alone for 45 days, providing overwatch and suppressive fire in a contested sector.
Iran's state-run TV featured a captured Anduril Dive-LD submarine that had been successfully operating in support of American forces, designed for dangerous environments where capture was anticipated.
This paper presents a method to autonomously generate the full 'Bad Apple' video using a small recurrent dynamical system with 417k parameters, trained from a single initial state. It details the architecture, training techniques like rollout horizon curriculum and state perturbation noise, and shares code and models on GitHub.
Kent C. Dodds tweeted about Astra spawning Astra subagents, hinting at developments in AI agent systems.
The paper proposes a Risk-Informed World Model (RIWM) to enhance safety in critical embodied systems by prioritizing decision-relevant consequences and recovery over predictive likelihood.
An experiment running an AI agent for 23 days to autonomously select and execute tasks via GitHub Actions, with automated checks resulting in a 54% failure rate that enhances efficiency by minimizing manual audits.
Argus is a general agentic runtime for autonomous research systems, enabling continuous learning and self-evolution through persistent memory, verification, and a pluggable architecture for deployment.
This paper introduces an autonomous AI coding agent that combines Monte Carlo Tree Search with Gemini LLM to enhance code generation, achieving a 92% success rate on complex logical prompts.
Caterpillar leverages decades of mining automation experience to advance broader AI deployment, including tools like the Cat AI Assistant and digital twins, while investing $100 million in workforce training for AI and autonomy.
The article reports on how delivery robots are designed to use human assistance for safely crossing streets, illustrating advancements in collaborative robotics for urban logistics.
A Google engineer's talk highlights how AI agents are transforming software engineering, making traditional IDEs obsolete, with 85% of PRs at Google already shipped by AI agents and introducing concepts like DAGs for self-improving autonomous systems.
Autonomous 5v5 soccer matches are underway at the World Humanoid Robot Games, with robots improving in fall recovery but still facing challenges in posture control after contact.