Scientists from the universities of Lincoln and Newcastle have replicated the way locusts use visual input to keep from flying into things to develop a computer simulated model that could be used in advanced collision avoidance systems for vehicles and in other applications. The simulation has already been used to help a robot autonomously navigate a path using visual input.
The technology could potentially be used to develop accurate vehicle collision sensors, surveillance systems, and to improve video game programming, the researchers said.
Why locusts? The insects process information using electrical and chemical signals while in flight, providing a fast, accurate warning system to help avoid collisions.
Prof. Shigang Yue of the University of Lincoln and Dr. Claire Rind of Newcastle University developed a visually stimulated motor control (VSMC) that converts visual information into motor commands. The system was based around the way the locust brain detects approaching objects.
“We created a system inspired by the locusts’ motion sensitive interneuron — the lobula giant movement detector (LGMD),” said Yue. “This system was then used in a robot to enable it to explore paths or interact with objects, effectively using visual input only.”
According to the research abstract:
The VSMC consists of a pair of LGMD visual neural networks and a simple motor command generator. Each LGMD processes images covering part of the wide field of view and extracts relevant visual cues. The outputs from the two LGMDs are compared and interpreted into executable motor commands directly. These motor commands are then executed by the robot’s wheel control system in real-time to generate corresponded motion adjustment accordingly.
In other words, the robot was able to avoid colliding with objects without the use of radar or infrared detection systems, which require a signficant amount of processing power.
Their findings were published in February in the International Journal of Advanced Mechatronic Systems.
Source: University of Lincoln