US-based Figure AI ne apne humanoid robots ko **30** alag-alag gharon mein test kiya hai. Inka naya Helix 2.5 neural network bina kisi special training ke kaam kar raha hai aur **56%** success rate dikhaya hai.
Robot ab ghar ka kaam karenge?
Figure AI ne California mein ek experiment kiya jismein unke robots ne bina kisi pre-training ke gharon mein towel fold karna, kamre saaf karna aur bed banana jaise kaam kiye. Ye technology 'generalizable robotics' par based hai, jiska matlab hai ki robot bina kisi manual coding ke nayi jagahon par kaam seekh sakta hai.
Helix 2.5 aur Index ka kamaal
Iske peeche asli dum 'Helix 2.5' neural network aur 'Index' dataset ka hai. Traditional robots ko har naye task ke liye alag se program karna padta tha, lekin ye AI naye environment mein khud ko dhalne ki koshish kar raha hai. Test mein inhone 56% success rate hasil kiya hai, jo ki kaafi impressive hai.
Lekin... kya ye safe hai?
Haalaanki 56% sunne mein sahi lagta hai, lekin iska matlab ye bhi hai ki 44% cases mein robot fail hua ya use insani madad ki zaroorat padi. Iska matlab saaf hai ki commercial market mein aane ke liye abhi inhein bohot refinement ki zaroorat hai.
Investors ke liye kya matlab?
Ye sirf ek experiment nahi hai, balki automation sector mein ek bada shift hai. Tech giants aur VCs ab aisi machines mein paisa laga rahe hain jo multi-tasking kar sakein, na ki sirf ek factory ke kaam ke liye bane hon. Agle kuch saalon mein battery life, safety protocols aur processing speed par focus badhega. Filhaal ye technology experimental phase mein hai, isliye long-term growth ka wait karna hi samajhdari hogi.