Autonomous navigation: Achievements in complex environments

The location of an autonomous robot can be estimated by means of a simultaneous localization and map building (SLAM) technique. SLAM utilizes uncertain information obtained by sensors, while also using the map to localize itself (proprioceptive sensors) with respect to a reference coordinate frame (exteroceptive sensors). Proprioceptive sensors make measurements of the internal state of the vehicle by the use of motor encoders or on-board accelerometers and others while exteroceptive make measurements with the help of transmitter laser lights and perhaps GPS. In environments where GPS is disadvantaged, local sensors like IMUs are a robust solution. However, these sensors will eventually need assistance from external sources. Laser range finders provide 360° coverage of the environment around the vehicle which produce range point clouds for location reference. Another tool is millimeter-wave RADARS (MMWR) which are less affected by dust, fog and ambient lighting conditions, which is hard for laser rangers. MMRW can provide a complete power returns for many points down range. However, autonomous navigation must overcome data association difficulties. The use of multiple-frame multi-dimensional data association algorithms resolve association incompatibilities and ambiguities ore effectively and yield consistent maps.