Teleo-Reactive Programs for Agent Control

By Nils J. Nilsson. 1993

A formalism is presented for computing and organizing actions for autonomous agents in dynamic environments. We introduce the notion of teleo-reactive (T-R) programs whose execution entails the construction of circuitry for the continuous computation of the parameters and conditions on which agent action is based. In addition to continuous feedback, T-R programs support parameter binding and recursion. A primary difference between T-R programs and many other circuit-based systems is that the circuitry of T-R programs is more compact; it is constructed at run time and thus does not have to anticipate all the contingencies that might arise over all possible runs. In addition, T-R programs are intuitive and easy to write and are written in a form that is compatible with automatic planning and learning methods. We briefly describe some experimental applications of T-R programs in the control of simulated and actual mobile robots.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Scalable Bayesian Rule Lists (2016, background): “Teleo-reactive programs (Nilsson, 1994) use a decision list structure and could beneﬁt from learning this structure from data.”
Between MDPs and semi-MDPs: A framework for temporal abstraction in reinforcement learning (1999, background): “How can we understand and automate this ability to work flexibly with multiple overlapping time scales? Temporal abstraction has been explored in AI at least since the early 1970’s, primarily within the context of STRIPS-style planning [18,20,21,29,34,37,46,49,51,60, 76].”
HOW MANY SEPARATELY EVOLVED EMOTIONAL BEASTIES LIVE WITHIN US (2002, background): “' Reactive systems can sometimes also be relatively slow, if they use a lengthy sequence of chained internal reactions between sensory input and motor output, for instance in the teleo-reactive systems described in (Nilsson, 1994).”
A Framework for Behavioural Cloning (1995, methodology): “Weassume a circuit semantics (Nilsson, 1994) for the continuous evaluation ofsystem inputs, though as in Nilsson's work in practice our experiments relyon sampling sensor values at some "fast" rate.”
A Survey of Behavior Trees in Robotics and AI (2020, methodology): “For other problems, a so-called implicit sequence design, inspired by the Teleo-Reactive approach [110] might be better, or a backward chained approach as suggested in [22].”
Learning from observation using primitives (2001, background): “Ryan and Pendrith’s architecture combines RL and teleo-operators [19], based on Nilsson’s teleo-reactive formalism [95], and makes use of both domain knowledge and observed information [109].”
Hierarchical control and learning for markov decision processes (1998, background): “The most important features of this approach are: An agent-centered abstraction language The hierarchical use of prior knowledge The optimal re nement of incompletely speci ed behaviors State space reduction These features are realized within the temporal abstraction framework, making strong optimal”
Behavior Trees in Robotics and AI: An Introduction (2017, methodology): “The example continues in [51] with a higher level recursive Teleo-Reactive program, called Amble(loc), designed to add a basic obstacle avoidance behavior Equal(pos,loc)→ Idle (6.”
Hierarchical Approaches (2012, methodology): “In RL-TOP, planning is used at the abstract level to invoke reactive planning operators, extended in time, based on teleo-reactive programs (Nilsson, 1994).”