Google is using deep learning to make warehouse robots smarter


According to a new report by the New York TimesGoogle is getting back into the robotics field. However, as opposed to the Silicon Valley giant’s earlier efforts, its new initiative doesn’t involve building new robots.

Instead, the company is focused on developing machine learning solutions to improve warehouse workflows.

The new program is called Robotics at Google and it’s described as being much more practical than the company’s ill-fated Replicant initiative.

In the early 2010s, the corporation brought together a slew of data scientists and robotics experts to create human-like machines. However, the program’s failure to produce commercial products led to its scrapping.

Robotics at Google is a much more grounded initiative.

The new lab is using deep learning programming to teach robots to recognize, sort, and pack different size objects. Another Robotics at Google project involves teaching bots to autonomously operate tools and equipment.

The Future of Logistics in Robotic

Google officials didn’t outline a specific roadmap for their new deep learning programs. But its most obvious application is in warehouse logistics.

Indeed, market analysts are predicting warehouse robotics will be a major growth sector in the near future.

Research firm MarketandReports recently stated the field will experience a 27 percent compound annual growth rate over the next five years thanks to the integration of industrial Internet of Things (IoT) technology.

Indeed, online retailer Amazon is in the process of overhauling its logistics operations with cutting-edge IoT-enabled robots.

In September 2018, the Seattle-based corporation opened an 855,000 square foot logistics park in Staten Island, New York. The facility pairs human workers with machine learning-empowered robotics to maximize efficiency.

To integrate bots into its fulfillment centers, Amazon first spent $775 million to buy its own robotics company. If Robotics at Google can provide the e-commerce giant with high functioning autonomous robots, it could secure a similar payout.

After all, as experts have pointed out, companies like Amazon have a great interest in using technological solutions to get their warehouse payroll as close to zero as possible.

However, Google’s dreams of a thriving robotics division and Amazon’s hopes for fully autonomous warehouses are dependent on near seamless robotics workflow. Currently, that goal is still a ways off.

Autonomous Robots Can’t Replace Human Workers…Yet

Late last year, one of Amazon’s robotics-enhanced facilities made headlines when one of its automated machines punctured a container of bear mace and sent dozens of employees to intensive care.

To the company’s credit, it quickly moved to bolster the safety of its human workforce. In January, Amazon, workers were equipped with “robot vests” that allow their machine counterparts to be more aware of human presence.

The fact that Amazon’s robots sometimes inexplicably destroy merchandise and harm coworkers is a stark reminder that full logistics automation still isn’t viable. Moreover, Google’s new packing robots are only 85 percent accurate. For a company that ships out 5 billion packages a year, a 15 percent error rate isn’t acceptable.

Still, with Google dedicating resources to developing deep learning logistics solutions and IoT-optimizing 5G mobile internet launching nationwide this year, one thing is certain:

The term “human warehouse worker” will be an anachronism sooner rather than later.