The brain has what are called grid and place cells. Grid cells aid in constructing the concept of three dimensional space through a type of grid. This allows humans to monitor where they are in the world. On the other hand, place cells are involved in recognizing specific physical places, such as a room. These two cells collaborate to ultimately provide our natural navigation system. A team of scientists from the Agency for Science, Technology and Research of Singapore has used this concept and applied it to an existing neural network used by an autonomous robot for navigation.

The Results

Although the initial results are still lacking compared to standard neural networks, they imply that the approach might lead to robots that are able to adapt to changing environments through a more robust navigation system. The virtual cells did successfully mimic the natural versions, suggesting that more advancement may lead to a whole new type of navigation system.

Share This Article