China unveils groundbreaking training dataset for embodied AI robots

⁠⁠⁠⁠⁠⁠⁠Chinese robotic developers unveiled the country's first training dataset for embodied intelligent robots in recent days, advancing the robots' abilities to learn from various environments and perform complex tasks.

The training dataset was jointly released by the National-Local Co-built Humanoid Robot Innovation Center and the School of Computer Science at Peking University.

At Beijing's innovation center, robotic engineers use a hands-on approach to train embodied AI robots, employing remote control devices and motion-capture suits to teach them human-like movements and actions through interactive sessions.

The related data is then collected in real-time and used for further improvement and research.

Additionally, the center has built simulated environments that mimic real-world settings, such as retail stores, factories, and homes, to provide the robots with realistic training experiences. 

The dataset contains data from 279 different task scenarios, involving different types of robots, like single-arm, two-arm, and humanoid robots. The dataset also records various robotic skills that can be applied widely. 

It is the first universal dataset for embodied intelligent robots in China, and has been validated through model testing.

"The innovation center keeps collecting data on complex, long-term tasks. So far, we have hundreds of thousands of high-quality data records, mostly from real robot testing, along with some data from simulations and remote operations. In the future, the center plans to gradually release the dataset, covering more robots and the tasks they perform," said Wei Jiaxing, a staff member at the innovation center. 

The quality of the dataset and its scale of application are key to training smarter robots, according to experts. Along with the dataset, evaluation standards have also been released, providing clearer guidelines on the requirements and quality of the robotic data.

China unveils groundbreaking training dataset for embodied AI robots

Worries about an economic recession are mounting among U.S. small businesses amid growing uncertainties around the impact of President Donald Trump's tariff policy, according to media reports.

Optimism about the U.S. economy is rapidly fading among consumers, corporate boards, and the Wall Street amid growing uncertainty fueled by tariffs, CNBC reported on Friday.

A CNBC survey showed that 70 percent of small business owners rated the current economic environment as either fair or poor, said the report.

Meanwhile, 70 percent of small business owners believe the country is heading for a recession, according to the report.

Additionally, 51 percent of respondents said they have already felt or expect to feel the negative impact of trade policy changes within the next 12 months, the report showed.

The report also noted that the ongoing uncertainty surrounding trade policies is adding financial strain to small business owners, both for their businesses and their wallets.

Federal Reserve Governor Michael Barr has also expressed the concern that Trump's tariff policy is likely to push inflation higher, slow economic growth and raise unemployment later this year, according to a Reuters report on Friday.

Barr has also warned that higher tariffs could disrupt global supply chains and create persistent upward pressure on inflation. Some suppliers, particularly the small business, may not be able to adopt quickly enough to the new situation and could fail, the report said.

He also cautioned that the combination of rising inflation and growing unemployment may put the Federal Reserve in a difficult position, according to the report.

Worries of recession mounting among US small businesses amid uncertainties over Trump's tariffs