Radhika Nagpal tells a great story about her engineering inspiration. On a snorkeling trip in the Bahamas she saw a massive school of yellow and black fish, and she felt that their collective mind was making decisions about speed and movement. Now a Professor of Computer Science at Harvard’s Wyss Institute for Biologically Inspired Engineering, Nagpal works on studying self-organizing collective systems. Her TED Talk What intelligent machines can learn from a school of fish is a fascinating look at what robots can do when their behavior uses nature as a model.
Nagpal says that our perceptions of artificial intelligence often come from movies and television, but talking robots and computer programs are too human-centric. Schools of fish and flocks of birds move in groups and their intelligence can often feel like ‘a property of the group itself.’ Group movement in nature comes without a leader or external control, and the interactions between units are based on a series of rules and decisions.
Her team at Harvard started by building a swarm of 1,000 tiny robots with the ability to move, communicate with other robots, and measure distance. Programming tells the robots the rules of engagement between their neighboring bots, and shifting the rules can cause the robots to exhibit different behaviors.
Nagpal shows a great example of her robots learning two rules, one for motion and one for patterns, to move from a large amorphous blog shape to a straight(ish) line. Several demonstrations show two dimensional patterns before moving to robots that can climb over objects and other robots to erect a three dimensional tower.
This is a fascinating talk full of great demonstrations of what Radhika is already doing with artificial intelligence but also future possibilities. Robots that pollinate crops, monitor coral reefs, or build flood barriers are possible using this technology.