Page 58 - Modern Robotics Building Versatile Macines
P. 58

38   Modern Robotics


            the Leg Laboratory with him. Raibert served as a professor of electri-
            cal engineering and computer science at MIT until 1995.



            Dynamic Walkers

            In the 1980s, legged locomotion was one of the largest undeveloped
            frontiers in robotics. Progress had been slow. Inventors had experi-
            mented with walking machines as early as the 19th century. These
            machines could only place legs in sequence stiffly and mechanically,
            without regard to obstacles or uneven terrain.
              In the 1970s, computer control began to be applied to creating
            more sophisticated walking machines. For example, in 1977, Robert
            McGhee at Ohio University led the successful development of an
            insect-like hexapod (six-legged) walker that could climb stairs. In
            1980, Japanese researchers built a four-legged machine that used
            more sophisticated computer algorithms to enable it to negotiate
            obstacles.
              This first generation of computerized walkers had a fundamen-
            tal characteristic in common. They were “static crawlers.” Such
            machines balanced by keeping most of their feet on the ground.
            As a leg moved forward, the center of mass remained balanced
            over the stationary legs. While this kept the robot stable at all
            times, it also meant that the robot lacked the agility of a walking
            human. In an article published in 2001 in Science News, Raibert
            recalled that early walking robots were “like tables with moving
            legs.”
              As Raibert had noted earlier in a 1990 article in U.S. News &
            World Report, “Biology already has the solutions—if we can tease
            them out.” Consider how people walk. People have only two legs, so
            they cannot remain balanced in a static way while moving forward.
            Indeed, people sort of fall forward with each step, accelerating. We
            do not fall over because part of us tips backward at the same time,
            compensating and keeping an overall balance.
              Raibert pointed out that there are important advantages to this
            dynamic form of locomotion. For one thing, with no need to keep
            rigidly over our center of mass, we can vary how far apart our feet
            are placed—for example, to avoid rocks while hiking along a path.
   53   54   55   56   57   58   59   60   61   62   63