Officials from the National Science and Applied Research Australia (CSIRO) shared their views on what robots of the future might look like. They will definitely not look like the most famous robots from the cinema.
In their paper, published in the journal Nature Machine Intelligence, the researchers say the concept of robots of the future will be inspired by evolution, allowing for truly amazing and efficient designs. A concept known as Multi-Level Evolution (MLE) argues that today's robots exist in unstructured, complex environments because they are not specialized enough, but must emulate a variety of animals that have adapted well to their environment.
“Evolution doesn't care what something looks like. She looks for a much wider design space and offers effective solutions that would not be immediately obvious to a human designer. Animals such as stingrays and kangaroos may look very unusual to the human eye, but they are ideal in terms of living in their environment,”says lead author Dr. David Howard.
The publication argues that with the help of current advanced technology, it is already possible to design a wide variety of robots, from the smallest size to those that can carry out their missions in extremely difficult conditions. Algorithms based on natural evolution will be able to automatically design robots by combining different materials, components, sensors and behaviors. According to scientists, this may become a reality in 20 years.
An artistic depiction of a turtle-like robot designed to work in the extreme conditions of Antarctica. As a turtle, it will be strong and resilient in extreme conditions / CSIRO.
Advanced computer simulations can then quickly test prototypes in simulated "real world" scenarios to determine which one works best. The end result, according to the authors of the work, should be simple, small, highly integrated, highly specialized and highly profitable robots, precisely designed for specific tasks, specific environments and terrain. Machines that are able to independently adapt and automatically improve their performance.
One good example is a robot designed for basic environmental monitoring in extreme conditions. The MLE's approach to designing this machine will depend entirely on terrain, climate and other factors. For example, a robot designed to work in the Sahara Desert will have to use materials that can withstand heat, sand and dust, run on solar energy, be able to move along sand dunes and use ultraviolet light to self-destruct so as not to pollute the environment.
Artistic depiction of a robot capable of working in the Amazon forest, inspired by crawling lizards and geckos / CSIRO.
Promotional video:
The dense vegetation of the Amazon is a completely different problem. A robot designed for this environment must be able to crawl around trees and over fallen logs and draw energy from jungle plant materials. In both cases, the MLE will automatically select the right materials and components for a high-performance robot design based on how well it performs a given task. This technique, according to the authors of the work, is much more practical than modern approaches, which require engineers to develop just one robot.
Dmitry Mazalevsky