This post is also available in: עברית (Hebrew)
As the name implies, work on Artificial Intelligence is essentially an attempt to create a synthetic mind. While there are many ways of achieving the goal, some of the most promising approaches are attempts to digitally replicate the workings of a biological mind. One group of researchers has now made a significant breakthrough and got a worm’s mind to control a robot.
Caenorhabditis elegans, the worm in question, has 302 neurons that have been meticulously mapped in what is known as a connectome – a comprehensive map of neural connections in a brain. It has been studied in laboratories the world over, and now its nervous system has found a new home: inside the body of a Lego Mindstorms EV3 robot.
Researchers from the OpenWorm project reverse engineered the worm’s brain, its neural network and operations, and applied them to the robot’s sensors and software, with regular UDP packets simulating the worm’s neural activity. For example, the robot’s sonar sensor substitutes for the worm’s nose. If the robot approaches within 20 cm of an object, the appropriate sensory neurons are activated just as they would if the worm was itself approaching some obstacle.
Consequently, the robot acts in a manner similar to the worm itself. The robot stops moving of its “nose” is stimulated, but touching its food sensor makes it move forward. “What we found is that rather than just random, crazy movements by the robot, it actually responded to its environment in the same manner as the biological worm,” said Timothy Busbice, one of the founders of the OpenWorm project.
While the simulation is not perfect – the programme has some simplifications and lowering of the thresholds that trigger “neuron firing” – the impressive thing is that there were no behavioural instructions programmed in. The worm’s behaviour simply emerges from the brain map implementation.