Page 230 - Robots Androids and Animatrons : 12 Incredible Projects You Can Build
P. 230

where significant contributions can be made. Before undertaking
                                       this project it is important to realize that an HMD that supplies
                                       stereo  images  to  the  user  will  be  needed  to  view  stereo  images
                                       transmitted from the T-bot.
                                       The small size of the miniature video cameras is good for stereo-
                                       graphic imaging. It allows two cameras to be positioned side by
                                       side at the same interocular distance (IOD) as that of human eyes.
                                       To  be  more  specific,  the  average  interocular  distance  (pupil  to
                                       pupil) for adult humans is about 63 mm. The camera’s lenses can be
                                       positioned, from center to center, this same distance apart to mimic
                                       the  IOD  humans  use  for  depth  perception.  The  transmitters  for
                                       each camera must be tuned to transmit on different frequencies.
                                       This allows the HMD video receivers to accurately display the right
                                       eye image to the right eye and the left eye image to the left eye.
                                       Seeing in stereo from the Golem provides the operator with depth
                                       perception when he or she is driving. Stereo-vision becomes increas-
                                       ingly more important when depth perception is needed, for instance
                                       when using a robotic arm. Being able to see the manipulator (robot
                                       arm) move along the Z axis in a three-dimensional (3D) coordinate
                                       system (X, Y, and Z) allows for efficient operation.

                                       It  is  difficult  for  an  operator  to  efficiently  use  robotic  arms  via  209
                                       telepresence when the Z dimension (depth) is lost in a monocular
                                       view. Operators are forced to gently bump into objects to approx-
                                       imate the robotic manipulator’s location along the Z axis.

                                       The  same  is  true  when  driving  a  telepresence  car.  One  quickly
                                       loses depth perception, and it becomes difficult to determine how
                                       far ahead of the vehicle something is.
                                       When the stereo system is set up, the operator of the vehicle will see
                                       the T-bot’s environment as a 3D picture. However, the stereo image
                                       transmitted will not contain the very important convergence clues.
                                       Much of our distance reckoning incorporates convergence clues we
                                       get from our eyes. Convergence is the angle our eyes rotate inward
                                       when viewing an object. An object very close to us will cause our
                                       eyes to rotate inward. In contrast, when viewing an object that is
                                       far  away,  our  eyes  look  straight  ahead.  The  brain  automatically
                                       brings this convergence information to bear in calculating distance.
                                       The stereo video cameras are in a fixed position looking straight
                                       ahead. To add convergence clues would require eye tracking engi-
                                       neering.  A  feedback-providing  HMD  would  need  to  constantly
                                       ascertain  the  operator’s  eye  convergence.  The  eye  tracking
                                       (convergence) information would be transmitted to servo motors

                                                       Team LRN                               Telepresence robot
   225   226   227   228   229   230   231   232   233   234   235