Autonomous robots behave on their own, without any kind of remote control from a human. Here Robot Madeleine, a four-flippered aquatic robot, operates autonomously in a pool. She uses an array of sensors to learn about the world: sonar, depth sensor, altimeter, compass, and accelerometer. She is programmed to explore and, at the same time, avoid obstacles, like walls. Image by John Long.
Evolvabots known as Tadros (short for Tadpole Robot) play the game of life. The job of prey robot (on the left) is to eat and not be eaten. It's food is the light hanging above the surface of the water. It's predator is charging in from the right. The prey robot has a brain modeled after that of a fish. A population of these robots was subjected to selection pressure for enhanced feeding and fleeing. Over ten generations, the population evolved more vertebrae in the tail and faster swimming. Image by John Long.
Tadros are modeled after an early vertebrate fish, Drepanaspis, that lived 400 million years ago. The first vertebrates were all fish, so understanding what evolutionary pressures drove the evolution of these species is key to our understanding of our own evolutionary history. Image by John Long.
Evolutionary Trekkers don't evolve, but they allow us to re-enact the behavior of extinct species. Robot Madeleine was built to test the different ways to swim with four flippers. The plesiosaurs ruled the Mesozoic seas as top-level predators, swimming with all four flippers. If four flippers worked then, why do living flippered species, such as sea lions or sea turtles, use only two of their four flippers for propulsion? Robot Madeleine provided a likely answer: while four flippers help you accelerate, they don't allow you to swim at a faster top speed. And if you try, you'll burn more energy. Evolution is all about trade-offs. Image by John Long.