Collaborative robots may be the key to human-AI collaboration

Adding artificial intelligence (AI) capabilities is a great opportunity to change the way a system behaves and is controlled. What excites me is the new opportunities it opens for (robots) and humans to work together – leveraging the capabilities of each.

The emergence of collaborative robots or “cobots”, designed to interact with humans in shared spaces or work safely in close quarters, is an important step in this direction. Collaborative robots contrast with traditional industrial robots, which work safely and autonomously by isolating themselves from human contact. Currently, collaborative robots only account for a small portion of the overall robotics market. However, we (and a number of recognized analysts) believe that this sector will grow rapidly over the next five years, with applications in manufacturing, healthcare and retail. The Covid-19 pandemic is accelerating this trend, providing more acceptance for increased digital transformation and automation.

Collaborative robots require closer control to implement complex decisions in real time in a collaborative environment. This is an area where there is a lot of focus on artificial intelligence to improve the experience of these types of machines. Such decisions must be made by robots as edge devices to achieve speed and latency to cope with the ever-increasing data from more () and the consequences of making the wrong decision. However, more pioneering manufacturing plants are starting to rethink processes to utilize humans and robots more effectively.

Edge computing will have a significant impact on the development of AI. Currently, AI training generates a large amount of data that is almost entirely implemented and stored in the cloud. Putting computing at the edge changes the processing and finds patterns locally. I believe this can make training models simpler and more efficient. I have seen automakers greatly improve their quality control processes with edge inference because they are able to find any defects in real time before the product is commercialized.

In the first phase of deployment, “edge computing” will typically be standard server blades deployed in facilities. This will evolve to optimized for purpose, deployed in units such as robots. Typically, the core of these systems will utilize very powerful processors to run many different workloads simultaneously. Imagine the traditional highly virtualized cloud computing environment into hardware that fits in the palm of your hand… and ensure that applications are isolated from each other, that one application cannot affect another, and that those functions that must always run in a responsive, deterministic manner, always do so.

From an AI perspective, the "eyes" of these systems have to be computers. There is too much data to process in too short a time to consider any other type of approach. But that doesn't mean we should rely solely on AI to protect and ensure systems continue to work when necessary. I can't see AI being introduced into the mission critical equipment of these systems, even though human lives are at stake.

Until we see a future without human collaboration, we will see AI-enabled systems continue to roll out in a more controlled manner. As the accuracy of models and datasets for specific functions improves, these functions will be AI-enabled and upgraded in the field with safe and reliable delivery via containers.

Artificial intelligence has been widely covered recently, especially as the pandemic has prompted businesses in certain industries to accelerate their digital transformation efforts. In a post-Covid-19 world, I believe the collaborative robotics market will grow rapidly as it is the only way to keep humans separate. So far, robots have been deployed on factory floors and assigned specific tasks. Collaborative robots that work collaboratively and flexibly with humans are an exciting vision whose promise relies on the safe and reliable implementation of artificial intelligence in these systems.