Robot Learns To Limp

Attention: open in a new window. PDFPrintE-mail

Robot learns to limp, as reported in ScienceNOW and news@nature 16 November 2006 and Science, vol 314, p118, 17 November 2006. Engineers at Cornell University have made a four legged robot that can work out how to walk, and then adjust its pattern of movement if one of its limbs is damaged. The robot is fitted with sensors in its limbs that send information to a computer when it moves its body parts. Using this information the robot builds an information model of its body and works out what should happen when moves its limbs. If one of its limbs is damaged, or a piece is lost, the robot's movements no longer match what the model predicts. The robot can sense this and work out what changes in movements it must make to keep going.

Robot researchers hope that this new method of robot learning can be used in robots sent to explore remote or dangerous places where human engineers will not be available to repair broken robots on site. Ronald Arkin of the Mobile robot Laboratory at Georgia Institute of Technology commented to news@nature that the adaptable robot is a "new development in evolutionary robotics" - described by news@nature as "a field that aims to build robots that learn from their environment without human help, and then teach other robots their skills."

Editorial Comment: When robots are developed that can learn from their environment it will have nothing to do with evolution. The robots, the sensory devices and the mathematical model that interprets the sensory information and sends instructions to the robot's limbs will all be created by intelligent pre existent scientists and engineers. Human movement depends on a similar but much more sophisticated system. Your brain is constantly monitoring sensory information from the body and using it to make adjustments to body movements and is able to see the obvious analogy to a created robot, as well as to choose to not learn from it. (Ref. technology, walking, algorithms)

Evidence News 5 December 2006

q_and_a2
crc_youtube
outdoor_museum_panel
free_audio2