Of Thomas Black*

For an industry that is just starting, there is great publicity around humanoid robots.

You can thank Ilon Musk and his bold statements about Tesla’s Optimus robot. Morgan Stanley also threw oil on the fire of publicity with her prediction for almost 1 billion anthropoids in operation by 2050 as part of a $ 5 trillion market. And then there are the “glittering” videos of the internet showing robots in human form to make bumps, jumps, and other spectacular accomplishments.

The reality is that most people overestimate what robots can do at this point of their development. It is also true that humanoid robots have evolved rapidly over the last decade, since expensive laboratory experiments in universities and specialized companies such as Boston Dynamics. Now, these mobile arms are mainly developed as pilot projects in warehouses, factories, and even in hospitals.

With these early developments of humanoid robots, obstacles for safety, power supply and mechanical learning are much clearer now to achieve the goal of manufacturing mobile robot. that can perform multiple tasks next to people and at a reasonable cost. These robots, which include models that have wheels instead of legs, represent the latest evolutionary step for robots that started more than six decades ago as clumsy, hydraulic moving machines that were screwed to the floor and isolated by workers.

The importance of developing an American domestic human -like industry that includes assembly and complete supply chain – engines, sensors, chips, cameras, batteries, etc. – It is not possible to overestimate. Like the shipbuilding industry, even drones, these products will be an integral part of the private economy with overlapping uses for defense. General robots will allow economies to grow even when the human population reaches a maximum threshold so it will begin to decline. This population decline, of course, is already happening in developed countries.

Skeptics focus on the limitations of these new machines, which are currently performing simple tasks, such as the transfer of objects, their container collection or the placement of plastic boxes in transport films. It is often late, energy -efficient and it takes a long time to train. History teaches us that technological improvements will overcome these obstacles. History also shows that the adoption of robots will last longer than optimists.

To give a sense of how much this industry is still in its infancy, the link to promote automation held its second annual conference dedicated to humanoid robots last week. The crowd was larger than that of the first concentration, with many companies displayed to see if these machines would operate in their factories or their construction sites. The answer: Not yet.

A dose of optimism in the morning by start -ups robot companies was compensated for skepticism in the afternoon, including Brad Porter, who led Amazon’s robotics organization before establishing its own company, Cobot. Amazon’s analysis of industrial robots found only 40 cases Use for humanoids that could not be made by other types of robots, Porter said. One of these uses is a wheeled robot that carries autonomous trolleys loaded with boxes for the AP Moller-Maersk A/S shipping giant. Porter’s newly formed company designed a robot, Proxie, for this work.

Although cases of use are limited now, more will emerge as robots improve their performance. Agility Robotics has developed its robot, Digit, in various warehouses where she can receive boxes and walk to place them in a transport belt. The areas where robots work, however, are closed to people. There are concerns about safety robots with legs, mainly because they crumble on the ground when the current is interrupted. Dipods of robots consume energy for “dynamic stability”, which keeps them in balance when standing or moving. Battery power is an important restriction on humanoid robots and the need for energy will only increase as they will become more capable of performing additional tasks. The robots must have a computational level of artificial intelligence because the delay in the signal by the cloud servers is very high, said Amy Gael, head of the Robotics Edge Computing Ecosystem in Nvidia, during the one -day conference.

One of the solutions is the connection of robots to a power source. This only works if they do not have to move a lot. Another method is to change the robot batteries when their power is reduced to a certain level. This requires additional investments for additional power systems, a space where batteries are switched on and time to replace them. One solution is to place the robots on wheels instead of legs, eliminating the need for power consumption for their stability. The disadvantage is that the wheelbase should be increasingly if the robot is going to lift heavy objects with his hands. It is also much harder for a robot to browse the human world with wheels.

Wheels are the way for moxi, the robot manufactured by Diligent Robotics and can carry medical supplies, medicines, laboratory samples and other objects directly to nurses. The robot can navigate to the hospital alone – taking the elevator, opening doors that require a signal or button to operate and maneuvering people and objects. About 100 Moxi robots have been developed in more than 25 hospitals. This saves time and walking for nurses.

Moxi, for the time being, has one case: the delivery of things to the hospital staff. The proliferation of anthropoids will not even intellectually approach Morgan Stanley’s forecasts if the robots cannot do multiple tasks – and learn them quickly. This is where robot training techniques need to be improved by simulation, Gael said. There is a lack of data for robot training. Robot training through work is expensive and time consuming. If the robot fails, it could cause damage to itself or objects around it. Generally fundamental models depicting real world situations will help reduce the cost of simulation and accelerate education.

Jeff Cardenna, co -founder and chief executive of the newly established Robotics Company, said he was afraid to go to automation conferences where the participants would laugh at the idea of ​​a man -made general use robot and that he would receive a bump. Since then, the common perception has changed along with the enormous progress made due to a challenge of the Advanced Defense Research Service (DARPA) from 2012 to 2015, which has given the industry.

APPTRONIK is working with Google Deepmind to address security issues and has worked with the newly established Jabil company for both the construction and use of Apollo robot. Cardenas said that next year he would still be pilot projects and early orders, with the industry developing hundreds of humanoid robots. “I think 2027 is the year you will begin to see an early scale.”

The era of humanoid robots is going to start, and the US must control their own destiny.

* Thomas Black is a columnist for Bloomberg Opinion and writes about industry and transport sectors. He was a journalist for Bloomberg News that covered logistics, manufacturing and civil aviation.