CSIRO researchers look at the future of robots



[ad_1]

Artistic impression of marine, coastal or river amphibious robots. I would travel in the water like an eel, but I would have my legs to descend and climb. Photo: CSIRO.

Researchers from the Australian National Science Agency CSIRO offered insights into how robots might look.

In an article published in The nature of machine intelligenceThe CSIRO (AIM FSP) Scientific Platform for Active Integrated Affairs suggests that robots will soon be able to take their engineering tips from evolution.

This concept, known as Multi-Level Evolution (MLE), claims that current robots fight in unstructured, complex environments because they are not specialized enough and should imitate the incredibly varied adaptive animals they have survived in their environment.

In the beginning of January, the main author, dr. David Howard, said that development is not interested in how it looks.

READ: The new UNSW course allows people to work in the robotics and the AI ​​world

"Looking for a much wider design space and offering effective solutions that would not be clear to the public designer.

"An animal, such as a manta ray or a kangaroo, may look unusual for human eyes, but it is perfectly calibrated for its environment," said Howard.

The document argues that in just 20 years of state-of-the-art technologies, such as high-performance detection and characterization, advanced production and artificial intelligence can allow robots to form at a molecular level to carry out their mission in extremely difficult circumstances.

Algorithms based on natural evolution would automatically form robots by combining various materials, components, sensors, and behaviors.

Advanced, computer-based modeling could then quickly test prototypes in simulated, "real" scenarios, to decide which works are best.

The end result would be simple, small, highly integrated, highly specialized and highly cost-effective precision of robots tailored to their tasks, environment and field. They adjust themselves and automatically improve their performance.

An example is a robot designed for basic environmental monitoring in extreme environments.

He should cross heavy terrain, gather data and eventually completely decompose in order not to pollute the environment.

The MLE approach to robot design would be entirely dependent on terrain, climate and other factors.

The robot, designed to work in the Sahara Desert, should use materials that can survive the punishment of heat, sand and dust. It can be powered by solar energy, moves along sand dunes and uses sharp UV light as a trigger, which can eventually be decomposed.

The thick, low-lying vegetation of the Amazon would be a completely different challenge.

The robot, designed for this environment, could crawl around trees and fallen logs, feed on biomass, such as a plant material that covers the ground in the jungle and decomposed with moisture.

In both cases, MLE will automatically select the appropriate materials and components in a high-performance robot design based on how well a robot performs a given task.

A constantly more flexible process than current approaches that require engineers to design only one robot.

[ad_2]

Source link