© ROOT-NATION.com - Use of content is permitted with a backlink.
Thanks to the introduction of an intelligent system for adapting the speed of simulated learning, known as SAIL, the emergence of multifunctional robots for everyday tasks, including cleaning, may become a reality in the near future. Scientists at the Georgia Institute of Technology have made significant steps toward machines replacing human resources not only in extreme conditions such as lunar exploration or deep-sea mining, but also in intensive mass production of cars.
It is about delicate manipulations with fine motor skills and hidden complexity that were previously considered unattainable for coordination and dexterity of mechanisms. These operations include arranging cups, folding laundry, packaging food, or placing fruit on plates – the types of work traditionally performed by staff in healthcare, childcare, elderly care, and restaurant settings. For business owners looking to minimize labor costs to increase profits, this is encouraging news. However, people whose lives and livelihoods are directly dependent on such activities, or residents of cities whose tax base is based on the income of such workers, may see the prospect of replacing people with machines from a completely different perspective.
It is worth taking a closer look at the essence of this technical achievement. In a recently published paper, researchers Nadun Ranawaka Arachige, Zhenyang Chen, and their colleagues describe a way to improve robots, allowing them to perform operations in everyday life and commerce with human precision, but at a much higher speed. Project co-author Shreyas Kousik notes that the group’s main goal is to create a general-purpose robot that can reproduce any manipulation available to human hands. The pace of work becomes a critical factor outside the laboratory, which led to the emergence of SAIL, an AI-based system for rapid adaptation of learning through imitation.
Combining the principles of robotics, mechanical engineering, and machine learning, SAIL is based on an algorithm that maintains smooth and continuous trajectories during acceleration, uses high-precision position tracking, automatically adjusts speed depending on the complexity of the task, and plans actions based on real-world delays. Experimental data shows that in twelve virtual and two physical scenarios, SAIL-enabled manipulators were up to four times faster in simulations and up to 3.2 times faster in reality compared to the usual demonstration speed.
Although developers have previously equipped robots with sensors and cameras for offline learning by copying human behavior, these methods had a significant limitation: they depended on the pace at which a person demonstrated a task. This, in turn, limited the throughput and performance required for full industrial automation. SAIL successfully overcomes this barrier. Previously, speeding up human tasks was difficult for machines, as even minor changes in the environment or physical nuances of working at high speeds led to errors and property damage. Koussik explains that the problem was in tying the robot to training data, so any deviation in the environment caused a failure.
For example, during the whiteboard wipe test, the marker base on the rack began to vibrate when the robot moved too fast. Humans are able to instantly adjust to such vibrations, while robots have not had the same flexibility until now. Understanding when speed is an advantage and when it is a risk is fundamental. The researcher’s colleagues emphasize that sometimes slowing down is the right decision. The goal is not just to mechanically accelerate, but to give machines intelligence so that they understand the line between performance and accident risk. To achieve this, SAIL modules coordinate acceleration beyond the initial data, while providing motion clarity, hardware latency control, and speed adaptability. While SAIL is not yet a universal solution to all robotics problems, it is a significant step toward the full integration of machines into human activity.
This progress brings us back to the discussion of large-scale job losses due to automation. According to the McKinsey Global Institute, by 2030, 400 to 800 million jobs may disappear as a result of the proliferation of robots and artificial intelligence. Experts from Robozaps clarify that this will force approximately 375 million people, which is about 14% of the global workforce, to completely change their field of activity. In the United States alone, automation could consume up to 30% of working time by the end of this decade.
Despite the claims of some optimists that robots do not pose a threat and can become the foundation for the introduction of a universal basic income, other analysts point to the extreme difficulty of realizing such a utopia. It also raises the question of the intentions of those who own the resources: whether they really want to create such a society. If not, it raises the question of who will be able to generate millions of new jobs to prevent a global economic crisis. The Economic Institute for Policy notes that the elimination of 100 positions in retail trade entails the loss of another 122 jobs due to the fall in purchasing power of the laid-off workers. In the manufacturing sector, the situation is even more critical: the loss of a hundred jobs indirectly destroys another 744 vacancies. In the end, machines don’t have to look like action movie characters to bring down the social order – they just need to learn how to fold your towels efficiently.
Read also:
- End of the “Flesh and Bone” Era: 9 Areas Where Robots Will Outperform Humans as Soon as Tomorrow
- Iron Legion: Ukraine will send 25 thousand ground robots to the front