Hod Lipson builds "self-aware" robots

By Jason L.

Hod Lipson explained that "We have been told for over 40 years that robots will be doing everything for use, they will be cooking, cleaning, shopping and building." He demonstrates a few of his cool robots, which have the ability to learn, understand themselves, and even self-replicate.

Hod is explaining various ways of how we can look at robots in different ways. He compares a beetle to a Swiss Watch, and he says "We can still make the pieces and circuitry, but we can't actually put that together to make something work." They created a robot built up of many pieces, including bars, motors, and neurons. He can create robots that have evolved to what they have rewarded them for, moving forward. They are able to create movements through simulation, and create real robots, that is made up of many brains, each one is rewarded for how fast or how far they can make the machine move. Hod also created "Cube" robots that can swivel and flip, and connect with other cubic robots.

The part that I found the most interesting was when Hod was explaining how they reward the robots. " We added a population of brains, evolving on the machine. It's related to a rodeo show, they all get a ride on the machine, and they get rewarded with how far and fast they move." These machines may not be ready to take over the world, but it is very neat how they can gradually learn and move forward. In addition I also enjoy the cubic robots that he created. They would be aware of their surroundings and attach themselves to each other to create one big robot, or several smaller ones. He showed clips of the simulations that they created of how they wanted the robots to function; they would then program the robot to do such things.

The robots may not be very useful themselves, but they might teach us something about how we can build better robots. Hod says, "We have to get away from this idea of designing the machines manually, but actually let them evolve and learn, like children."