机器狗浇花、机器人越野：这比赛比综艺还好看

Core Insights - The article discusses a recent robotics competition held at The Chinese University of Hong Kong, focusing on the capabilities of robots in real-world scenarios, termed as the "Real World Extreme Challenge" [5][7] - The competition featured 13 student teams from global universities, aiming to test robots' autonomy and adaptability in dynamic environments [5][7] - The event highlighted the current limitations of robotic algorithms, particularly in real-time reasoning and decision-making, with the best models achieving only 55%-60% accuracy in 3D scene understanding compared to human performance of 90.06% [7] Competition Overview - The competition included four tasks: garbage sorting, watering plants, crossing a suspension bridge, and orienteering in a mountainous area, each designed to assess specific robotic skills [9][42] - Teams had 30 minutes to complete each task, with points awarded for successful completion, particularly favoring fully autonomous operations without remote control [9][10] Task Details - In the garbage sorting task, robots struggled with object recognition and manipulation, demonstrating the challenges of real-world applications [10][12] - The watering task required robots to mimic human actions, such as picking up a watering can and accurately watering specific plants, which proved difficult for many teams [18][20] - The suspension bridge task tested robots' balance and adaptability, with many teams employing creative solutions to navigate gaps between planks [32][36] Technical Insights - The competition aimed to expose the weaknesses of current robotic technologies, emphasizing the need for advancements in embodied intelligence and generalization algorithms [43][44] - Experts noted that the event's design was not merely for entertainment but to drive technological progress by presenting real-world challenges that robots must overcome [44] - The ultimate goal is to achieve a level of robotic performance that can reliably handle basic tasks, reaching 20%-30% of human operational capability within the next 3-5 years [44]