At the second World Robot Games, the autonomous robot team Yinhe Tongyong, powered by their AstraBrain world action model, achieved a 100% win rate across all three challenging scenario competitions: home, catering, and supermarket. Unlike most competitors who opted for remote operation due to the high risks of full autonomy, Yinhe Tongyong's robots completed tasks without human intervention, demonstrating advanced capabilities in handling soft objects, recovering from unexpected interruptions, and precise tool manipulation. Their success in these complex, real-world scenarios, including a 30-minute autonomous operation in the home setting and sweeping all medals in the supermarket category, highlights the robustness and adaptability of their embodied AI. AI
IMPACT Demonstrates significant advancements in embodied AI, potentially accelerating the adoption of fully autonomous robots in complex real-world environments.
RANK_REASON The cluster reports on a significant achievement in embodied AI at a robotics competition, highlighting the capabilities of a specific AI model and team. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
