Despite rapid advances in artificial intelligence, humanoid robots still face major hurdles in mobility, communication and energy efficiency.
With all the attention, headlines, discussion, expectations, hopes, fears and, yes, hype surrounding artificial intelligence (AI), there is one area that appears far from ready for prime time: humanoid robotics.
Some may argue that robotics is not really AI at all. That perspective comes from traditional automation, which has been used for decades to load and unload equipment, stock inventory in warehouses or move parts along assembly lines. These systems relied on straightforward programming tailored to specific tasks. In the 1990s, many printed circuit board companies had fleets of loaders and unloaders. More recently, warehouse automation was expected to revolutionize logistics. In both cases, productivity improved significantly, and the automation introduced was indeed transformative.
Today, however, both the technology and the expectations have changed.
Traditional automation has advanced dramatically. Improved sensors can now detect pressure, temperature, weight and location with a level of accuracy that would have been difficult to imagine just a decade ago. Yet even with these advances, most factory automation remains tied to a specific machine or location and relies on fixed power sources or charging stations.
At the same time, humanoid robotics has made impressive progress. For many people, the benchmark remains C-3PO from Star Wars – a robot that could walk, talk, reason and even display something resembling human emotions. Nearly 50 years after that character first appeared on screen, it still shapes expectations of what a humanoid robot should be.
The challenge is that functioning like a human being is extraordinarily difficult.
One way to understand the problem is to look at how humans develop. A baby begins by learning to see and hear. Next comes basic body control: clapping, crawling, standing and eventually walking. Communication follows, along with the ability to process and learn from the constant stream of information coming from the surrounding environment. Over time, humans learn to understand situations, interpret context and respond appropriately. Finally, they become capable of functioning productively for hours at a time without needing a nap.
AI development has followed almost the opposite path. Machines first gained the ability to store and process information. Communication capabilities came next. Physical movement and body control have proven to be among the last and most difficult challenges. In many ways, AI-driven humanoid robotics is developing in reverse order from human development.
That is where the real difficulty lies.
Figure 1. Humanoid robots continue to improve in mobility and interaction, but significant challenges remain in movement, communication and energy efficiency.
For machines, movement remains one of the biggest obstacles. A robot may be able to run faster than a human being, but often that is because running is the only task it was specifically designed to perform. Human movement, by comparison, involves balance, dexterity, adaptability and the ability to perform countless motions under changing conditions. Replicating that flexibility is a far greater challenge.
Communication presents another hurdle. AI systems can exchange information with other machines quite effectively. Communicating naturally with people is another matter. Humanoid robots can increasingly handle factual exchanges, but they still struggle with the nuances of human interaction. People respond not only to words but also to tone, context and emotion. That level of communication remains a work in progress.
The greatest challenge, however, may be energy.
Human beings are remarkably efficient. With three meals a day, a person can typically work eight to 12 hours and, when necessary, much longer. Humanoid robots operate under very different constraints. They require significant power, and the more functions they perform, the farther they travel and the more complex their calculations become, the greater their energy demands.
Military leaders have long understood that an army is only as effective as its supply lines. The same principle applies to humanoid robotics. Their effectiveness is limited by how much energy they can carry and how long that energy lasts. Battery technology, more than software, may ultimately be the factor that determines how quickly humanoid robots become practical.
For workers on the factory floor who move materials, install equipment or perform tasks requiring mobility and adaptability, there is little reason to fear replacement by humanoid robots anytime soon. The limitations of movement, communication and energy consumption continue to outweigh the advantages these systems can currently provide.
Given the challenges involved, along with the costs of developing mechanical solutions compared with software-based ones, humanoid robotics appears likely to remain a work in progress for many years to come.
Peter Bigelow has more than 30 years’ experience as a PCB executive, most recently as president of FTG Circuits Haverhill; This email address is being protected from spambots. You need JavaScript enabled to view it..
