Call for Papers and Demos for the 11th International Rule Challenge

TeleoR is a major extension of Nilsson’s Teleo-Reactive (TR) rule based robotic agent programming language. Programs comprise sequences of Guard ~> Action rules grouped into parameterised procedures. The guards are deductive queries to a set of rapidly changing percept facts in the agent’s Belief Store. The actions are either sets of primitive actions for external robotic resources, to be executed in parallel, or a single call to a procedure, which can be a recursive call. The primitive actions are usually durative and are stopped or modified only when another action is determined. In each invoked procedure the first rule with an inferable guard is fired.

The guard of the first rule of a procedure is the procedure’s goal. The other guards form a sub-goal tree rooted at this guard. Their linked actions are such that when the rule is fired the action will normally eventually result in an earlier rule being fired. If there is always a rule that can be fired, the procedure is a universal action plan for achieving its goal. The goal of a task is the goal of the initially called procedure.

A TR and TeleoR computation is also highly reactive in that on each Belief Store update rule firings are reconsidered, starting with the initially called procedure. If in some called procedure a higher rule can be fired, progress has been made towards the task goal perhaps due to outside help. If instead a lower rule is fired, exogenous events or action failure has resulted in a setback. The lower rule has been fired in order to try to climb back up the procedure’s, and hence the task’s, sub-goal tree. This operational semantics results in both robust and opportunistic behaviour, well suited to robot/robot and robot/human collaboration.

TeleoR extends TR in being typed and higher order, with extra forms of rules that allow finer control over sub-goal achieving task behaviour. Its most important extension is the concept of task atomic procedures. These enable the high level programming of multi-tasking agents that interleave the use of one or more independent robotic resources, without interference, deadlock or task starvation.

TeleoR multi-task programming is illustrated by a program that can be used by an agent controlling two robotic arms in multiple block tower assembly tasks, chosen as a representative of robotic assembly or packing tasks. Both arms need to be shared by the tasks, and the arms execute in parallel whenever possible.