First Results in the Autonomous Retrieval of Buried Objects

Abstract

We have developed an autonomous system for the retrieval of buried objects. It is designed to detect, locate and retrieve buried objects. The system is equipped with a surface and a subsurface sensor. The surface sensor is a laser rangefinder used to build the surface model. The subsurface sensor is a ground penetrating radar (GPR) used to detect and locate buried objects. We use an industrial robot equipped with an excavator bucket for automated excavation. Our system uses a “sense and dig cycle” procedure to retrieve buried objects. First, subsurface sensing is used to detect and locate the buried objects. If the object can be reached with one dig, the excavator retrieves it directly. Otherwise a layer of soil above the object is removed and another subsurface scan is made to get a more accurate estimate of the object location. This loop is repeated until the object is retrieved. The paper presents some recent results in sensing and excavation from experiments conducted on our testbed and the specific methods that we use to achieve those results. Such a system has numerous potential applications. They include hazardous waste removal, maintenance of subsurface structure (e.g. gas pipes), construction, and removal of unexploded buried ordnance.