Action Planning & General Game Playing for Robots

One of the oldest dreams in the research area of Artificial Intelligence is the design of autonomous robots with problem-solving ability comparable to that of humans. Beginning of the 70's was the Stanford Research Institute Planning System STRIPS (Fikes & Nilsson, 1971). Its formalism for the description of action planning domains is fundamental. STRIPS is motivated by mapping plans to robots, and SHAKEY (Nilsson, 1984) was one of the first robots, who offered its services based on STRIPS. Planning operators divide into preconditions and effects, and provide an inference mechanism wrt. a logical description of start and goal state. Of-the-shelf planners then generate a sequence of actions. STRIPS has been modified over the years into other description languages, such as the Planning Domain Definition Language PDDL (McDermott, 2000, Fox & Long 2003). Parallel to this was the design of the Game Description Language (GDL). Investigations show: many planning problems can be described in GDL, as well as many GDL-specified games have a PDDL-like equivalent. Regardless of the vision for the planning-based robotics, researchers have problems to port "classical" approaches to a service robot, e.g., to perform household work, due to various reasons: the discrepancy of the state of the world from its internal representation; the difference of the work and configuration space (with its many degrees of freedom); dealing with the faulty execution of plans; many plans are "paranoid": all uncertainties are interpreted in a worst-case scenario; the lack of a learning process for continuous improvement of plans; the need for reactive control, i.e., the adaption of plans to the environment; the requirement for planning in real time; the discrepancy between the discrete and continuous representations; the inaccuracy and diversity of different sensors; the inherent incompleteness of information; the fixed execution units in the robot control; the complexity of the inference based on trajectories, etc.