Pepper is not dead, Sun Zhengyi’s robot dream is difficult to realize | Business Observation

OFweekrobotNet News SoftBank humanoid emotional robot Pepper lost money and became insolvent. I didn’t expect it so quickly, and I’m not surprised.

I was in Tokyo when it was released, and I really had confidence in it at the time. Masayoshi Son’s strong vision that day, as well as his personal enthusiasm, are very infectious.

At that time, the author believed that Sun Zhengyi, Guo Taiming, and Ma Yun were expected to rewrite the history of global Robots. Because, from the perspective of market factors, SoftBank’s ambitions for robots, its ability to integrate products, and its own ICT infrastructure, Foxconn’s ODM capabilities, and Alibaba’s marketing capabilities and scene construction capabilities can theoretically lead to a new industry.

But it now appears that expectations are too high and there are many doubts.

Pepper is not dead, Sun Zhengyi’s robot dream is difficult to realize | Business Observation

I once laughed at Pepper not to mention that as an emotional robot, even the most common speech recognition and image recognition, its overall response is relatively poor. However, looking at it in combination with the trends in the AI ​​field in the past year or so, it can be felt that its shortcomings are its advanced positioning and lack of support from core competitive elements.

Emotional robots are different from what other Japanese companies are good atautomation equipmentscheme under the concept. But after half a year of hype, its actual application is still simple conversations and some shopping guide scenarios, and the world’s first function with an “emotional engine” that Sun Zhengyi said is overshadowed.

Of course, Pepper can easily pace, but he is no match for Boston Dynamics. On the whole, in terms of the scene, it is not only different, but also obviously lacking in technology. Although it claims to have an emotional engine, it lacks a lot in terms of natural language and corpus support. It is even far less able to chat and understand your emotions than Baidu Dubi and Microsoft Xiaobing. Behind this, what SoftBank lacks is not technology, but the continuous accumulation of the industry for many years. Some parts can’t be done quickly by pure acquisition, and there is a lot of complex deep learning training behind it. And Pepper lacks this long-term accumulation.

As for the artificial care that Sun Zhengyi advocates to adapt to the elderly society, it is even more short-sighted. In this part, SoftBank has not opened up many public service networks at all, let alone connected to countless offline life service scenarios, and it lacks the ability to be O2O.

On the whole, Pepper can only be regarded as the previous generationindustryThe so-called “emotional engine” is the result of the primary stage of AI, which is a stand-alone mobile dialogue robot opposed to robots. Compared with the current series of products, it is outdated.

Looking at it now, the value of the birth of AlphaGo is by no means playing chess or subverting the field of human art. It is really a wake-up call and reform at the bottom of AI. When the machine can learn and train independently, the entire mode of thinking has changed. Compared to it, Pepper’s so-called emotional engine should be just the AI ​​consciousness conscious on the eve of the birth of Alpha Dog, but it has not achieved its ability at all.

But to simply deny SoftBank’s efforts from this, it would definitely underestimate Sun Zhengyi. The author believes that Pepper is only the first step in the layout of Sun Zhengyi in the terminal field, which belongs to the experimental stage. Just like Google’s experimental hardware terminal, Pepper’s value lies in its independent involvement, training teams, and exploration paths without a reference solution.

It is even conjectured that SoftBank will not build an independent robot terminal on a large scale in the future, but build an ecosystem around this field.

1 2 Next page > page

Looking back at why Sun Zhengyi strongly acquired ARM and established a $100 billion fund, this should be the underlying ecological layout. If it is all involved in the terminal independently, the ARM ecosystem may become a disaster.

Not long ago, Sun Zhengyi presided over the acquisition of Google’s Boston Dynamics and the Japanese biped robot Schaft. In 2016, SoftBank introduced IBM’s Watson AI system to increase Pepper’s intelligence.

It can be seen that it is continuously integrating multiple underlying technologies to build the world’s largest robotics platform. Of course, the scope is enlarged, and it is also an AI platform. I believe it will be laid out around more elements of AI. In addition to technology, there will also be data, connectivity, and application scenarios.

Brands hit by AI may die, but SoftBank should make a massive comeback. What Sun Zhengyi looks for is always indomitable. In 2010, at the new 30th anniversary meeting of SoftBank Group, he solicited future transformation opportunities from 20,000 employees, and the most advice he received was to develop robots. In the “New 30-Year Vision” released in June of that year, he emphasized the future blueprint of “coexistence with intelligent robots”.

There is also SoftBank’s sense of crisis behind this. Over the past 35 years, the giant has gone through the core stages of IT industry investment and operation, broadband operation, fixed network operation, and mobile operation. Up to now, not only has global IT investment been huge, but operator business has also spread all over the world. However, with the in-depth integration of global telecommunications and the internet, especially the boom in the Internet of Things and cloud computing services, SoftBank needs an entrance that can integrate its own huge re