The Birth Of The Cloud-Enabled Hybrid-AI Robotics Platform

SHARE

I continue to think that Robots are the next big technology trend.  The COVID-19 event seems to make this even more likely because of the need to keep people apart from each other to prevent the virus’ spread. 

This week Qualcomm put on a presentation on this topic with their focus being on 5G connecting these robots to centralized AI resources.  This presentation was designed to showcase their Robotics RB5 platform but is also showcased a different way to create robots so that they are more reliable, smarter, and less expensive to build. 

Let’s talk about some of the exciting things they shared.

Centralized AI

Most of the robots that currently work around us have their processing capability.  While this does make the robot able to function off-line, the trade-offs between processing power, size, battery limitations, and inflexible technology result in minimal AI capabilities. If you can centralize this intelligence and offer it more on an on-demand basis, the cloud service the robot is connected to can providing the robot based on need.

For instance, if you have a security robot, you don’t need much intelligence for it to navigate a known route. But if it sees an obstacle or identifies a person, it then needs to decide what it sees and then makes relevant decisions. Then the ability to connect to a centralized resource that can provide the focused processing resources and decision trees that the robot will need to make those decisions.

This requirement for both on-device and cloud intelligence means the solution needs a secure, reliable, low latency, high bandwidth wireless network to make this work. 5G was positioned against this need, given its primary advantages reliability, high-speed, and low-latency. Qualcomm showed a video of a robot at CES balancing a stick on its hand using 5G to connect it to a centralized AI resource and the latency was low enough and the speed fast enough so that the robot was quickly able to find and maintain the stick balance even if the operator interfered with the stick.

By using this wireless connectivity, the cost of the robots can be reduced as a result of centralized AI resources lowering the need for processing power in the robot.  One company to watch, in terms of creative robotic implementation, is Amazon, who appears to be working on a very advanced robotic store concept that could be perfect for a post-COVID-19 world.

In its final form, you’d go on the web and select what you wanted, then either have it delivered to your home by a robot or you’d drive to the store, and a robot would bring what you ordered to you.  In effect, their local stores would be automated Amazon warehouses operated by robots.  Secondary advantages, over convenience, would be lower shrinkage (employee theft), elimination of shoplifting, and a blend of online shopping and near-immediate fulfillment.

Qualcomm’s RB5

Now we can’t do all the processing in the cloud, and the Qualcomm RB5 platform is designed to provide enough processing capability to minimize the amount of data that needs to be sent to this centralized resource. RB5 supports multiple cameras; the robot can use its inherent intelligence to decide which video feed to use and what images need to be sent to the centralized resource. 

The platform is a modular design that allows the implementation of multiple configurations that enhance its sensor suite, provide dedicated multi-camera support, multiple motor control, and wireless support (5G, LTE, CBRS). Depending on the need, the product is packaged in several different ways, including System-on-module, chip-on-board, POP, and non-POP packages. 

While they are talking robots, this intelligence unit could be applied to other needs where you need a hybrid device/cloud solution. 

5G The Move To Centralized Intelligence

What Qualcomm is arguing has implications outside of robotics. If you think about the number of business and consumer devices needing intelligence, this use of centralized AI could both make those devices far smarter and far more long-lasting.  In a fast-moving market like robotics, the useful life of the device is often defined by its intelligence.  If the cloud supplies the extra intelligence, the robot needs upgrades that can be applied virtually without requiring the replacement of the robot except when the physical limitations of that robot obsolesce it.

For manufacturing robots or cleaning robots that could extend their potential useful life substantially given the physical capabilities of these devices are relatively mature and haven’t changed for some time.

The Birth Of Intelligent Hybrid AI Solution

This concept of having essential intelligence on the device, coupled with a low-latency cloud resource for more difficult decisions, has broad potential that goes beyond robotics. Anything that needs a range of intelligence could make use of this model.

One thing that Qualcomm hinted at but didn’t explore in this presentation at length was the concept of shared intelligence.  Shared intelligence would be where the robots could loan processing resources to their peers as needed to further lower centralized resource loading and, in an emergency, to continue to function if those central resources failed.

This robotic advance suggests that we are going to be up to our armpits in robots in a few years.  Thinking through the approach, you want to take now could help prevent a lot of wastewater.  A centralized technology provider like Qualcomm can better assure common standards and weather the disruptive changes that are still in this market’s future. Smaller, more focused firms that are vertically integrated are more likely to fail because they won’t have the ability to drive standards or whether the wave of coming advancements in this space.

These hybrid-robots will disrupt several industries. This trend will provide opportunities for those that come up to speed on the technology early and exposures for those that put this off until too late.  You have time, just not an unlimited amount of it.  This COVID-19 event is putting a ton of interest in this area, and that interest should result in what will look like amazingly fast disruptive changes. 



NewsletterDATAMATION DAILY NEWSLETTER

SUBSCRIBE TO OUR IT MANAGEMENT NEWSLETTER