Robots doing chores is nothing new, with Tesla having put forward its cool Optimus robot and DeepMind revealing its latest achievements in robotics. Recently, a robot called Mobile Aloha caught the attention of many at the beginning of 2024, with short videos of it cooking at someone's home going viral in January.
Developed by a team of researchers at Stanford University in the US, the robot can handle everything for a dish, whether it is chopping（切） vegetables or cracking eggs.
It is also good at various household tasks like watering plants, petting cats, cleaning the floor and doing laundry. It even knows to shake the pillow after putting on a pillowcase （枕套）. An Internet user joked under one video that as long as this thing doesn't try to kill him while he is asleep, he is in real need of it.
However, a following video posted by one of the researchers, Tony Zhao, and showing Mobile Aloha's failures proved that the idea of having a robot servant may just be wishful thinking. In the video, Mobile Aloha randomly smashed（打碎） glasses and plates, collided（碰撞） with cabinets（橱柜） and even burned a pot.
It turns out that Mobile Aloha is not a complete self-learning system that can independently navigate new environments. It relies on demonstrations by human operators in its surroundings, meaning that the robot needs to learn from human behavior before completing each task. Also, according to the team, the robot achieves a 95 percent success rate in wiping red wine stains（污渍）, 80 percent in pushing chairs, and a me re 40 percent in frying shrimp（虾）. In short, it's far from perfect.
The behavioral problems of AI robots have been bothering scientists for decades. Although AI robots do pretty well in things requiring high-level reasoning like math, they perform worse than a one-year-old child when it comes to simple tasks demanding abilities of perception, reflexes（反射） and mobility, among others.
As the team observed, the interaction between the arm and the base of Mobile Aloha would get quite complex if more flexibility is required in a task. Even a slight deviation（偏差） in the base settings might lead to significant drift in the arms' motions, resulting in failure to complete the task.
