At the moment, industrial robotics uses safety fences to protect humans when both share the same working cell. The aim of this paper is to present a system-architecture for the solution of an efficient risk reduction system which provides real time evaluation of a complex risk function to allow for an efficient cooperation between human and industrial robots. Some earlier approaches deal with foreground detection using a fixed multi camera set up [10] or obstacle detection with risk evaluation [12]. Other approaches focus on danger estimation during interaction with robot motion strategy [11]. The here suggested architecture is based on depth image sequence analysis from a single 3D sensor system. Having additional sources of information (depth measurements and signal amplitudes in real-time), it is possible to estimate features that make it possible to rethink the risk estimation for humans in hazardous environments like industrial robot cells and as a consequence to dispose the safety fences. In the next section, it will be shown in what way 4D (3D spatial and time) image information is a prosperous feature solving the problem of risk reduction when human beings and machines are interacting.
展开▼