he real worldwith its vast space and many unfamiliar obstacles. To over overcome thisproblem, Brooks programmed this little robots with so-called parallel behaviors. He didnt need the all-encompassing maps others used, just simple leg routines,such as up-down and forward-back motions the operate in parallel. On therobots legs sensors warded of obstacles and triggered these behaviors likereflexes. These machines ran riot in the lab where gradua!te students built new obstacles courses for them every day. The insect likerobots managed to conquer all. Cog represents the basic same principal, but afew steps beyond.Brooks plans to take Cog a long way. Right now , Cog is learning how to see,and relate what he sees to his head motion, to be able to know what motion inthe world, and what is dew to his head movement. They are letting Cog learn byits self. There are also plans to make ears for cog so that he may learn tohear. They have all ready completed a poly styrene head that is complete withmicro-phones and processors. The first step in teaching Cog to hear is to mapsensor coordination between the eyes and ears. With the done, when cog hears asound, he will point his eyes at it, and then if he sees motion, that willreinforce what he just heard with the coordination of the two. Sound will helpcog figure out where to look, but he will also be able to separate sounds, likepeople can do at a noisy cocktail party. These senses will help it to be ableto interact with its environment and learn about its own body. What it seesverses what it is doing.There are also plans to give Cog the ability to be able to reach out and touchsomeone. Right now he has no arms at all. He can lean forwards and sideways atthe hips, and turn its head, but that is all at the moment.Williamson, a graduate student on the Cog team, is building Cogs first arm as apart of his Ph.D. research. Cogs arms will have compliance unlike some otherrobotic arms that ar...
