CarderPlanet
Professional
How does artificial intelligence learn effective navigation from insects?
With the increasing use of artificial intelligence in everyday life, even traditional agriculture is not left out.
Ecorobotix, a two-meter-long GPS-enabled robot powered by solar panels, is able to destroy weeds with up to 95% accuracy without creating waste. At the same time, the Energid and Universal Robots systems use robots with multiple cameras and flexible arms to harvest citrus fruits.
By scanning crop geometry, River LettuceBot optimizes growth and minimizes the use of pesticides. PrecisionHawk also offers the use of drones for remote monitoring and analytics.
However, among all these wonders of modern engineering, one of the developments stands out qualitatively from the rest. For example, scientists from the Universities of Edinburgh and Sheffield are investigating the issue of visual navigation in dense vegetation, relying on how ants do it.
In a paper published in Science Robotics, the researchers emphasize that they were looking for "low-power but efficient onboard solutions" for their robotic navigation research.
"We drew inspiration from insects such as ants, which are able to learn and follow routes in difficult natural environments using relatively limited sensory and neural systems," the researchers said.
"In our study, we present an example of this approach by implementing a network for memorizing visual routes on neuromorphic equipment that directly draws on the latest advances in insect neuroscience," the researchers added.
As a result, the authors managed to develop an artificial neural network that helps robots overcome difficult routes in conditions of high vegetation density based on image analysis and a kind of memorization of the optimal path of movement, as insects do.
The researchers tested their neural model on complex routes through uneven, muddy, overgrown fields and achieved positive results. They believe that their research shows the prospects of using such systems in agriculture, forestry, as well as in environmental monitoring.
With the increasing use of artificial intelligence in everyday life, even traditional agriculture is not left out.
Ecorobotix, a two-meter-long GPS-enabled robot powered by solar panels, is able to destroy weeds with up to 95% accuracy without creating waste. At the same time, the Energid and Universal Robots systems use robots with multiple cameras and flexible arms to harvest citrus fruits.
By scanning crop geometry, River LettuceBot optimizes growth and minimizes the use of pesticides. PrecisionHawk also offers the use of drones for remote monitoring and analytics.
However, among all these wonders of modern engineering, one of the developments stands out qualitatively from the rest. For example, scientists from the Universities of Edinburgh and Sheffield are investigating the issue of visual navigation in dense vegetation, relying on how ants do it.
In a paper published in Science Robotics, the researchers emphasize that they were looking for "low-power but efficient onboard solutions" for their robotic navigation research.
"We drew inspiration from insects such as ants, which are able to learn and follow routes in difficult natural environments using relatively limited sensory and neural systems," the researchers said.
"In our study, we present an example of this approach by implementing a network for memorizing visual routes on neuromorphic equipment that directly draws on the latest advances in insect neuroscience," the researchers added.
As a result, the authors managed to develop an artificial neural network that helps robots overcome difficult routes in conditions of high vegetation density based on image analysis and a kind of memorization of the optimal path of movement, as insects do.
The researchers tested their neural model on complex routes through uneven, muddy, overgrown fields and achieved positive results. They believe that their research shows the prospects of using such systems in agriculture, forestry, as well as in environmental monitoring.
