In a general sense, control can be described as a cyclic sequence of observing, planning, and executing new actions. Similar to the human body, robotic devices are being equipped with a wide variety of sensors for the purpose of acquiring the needed information for planning and replanning. Because of the harsh work environment, construction robots, such as autonomous excavators, depend on hardened and reliable sensory systems, thus limiting the choices for selecting sensors. For control purposes, the sensory data need to be evaluated for decisionmaking purposes. The intelligence for this endeavor has to be based on a thorough understanding of the meaning of the sensory data and high confidence in the appropriateness of future actions. In addition, a special obstacle-avoidance concept is needed to prevent collisions and other types of accidents. The present paper discusses methods for both motion control and path planning for backhoe excavation. Experimental laboratory data are used to verify theoretical findings and to discuss the use of pattern recognition for the purpose of intelligent reasoning about the sensed environment. It is shown how patterns found in collected data could be used for automatic control of an autonomous excavator. © ASCE.
|Number of pages||18|
|Journal||Journal of Aerospace Engineering|
|Publication status||Published - 1 Jan 1993|