Reformulating path-planning problems by task-preserving abstraction

doi:10.1016/0921-8890(92)90034-V

Robotics and Autonomous Systems

Volume 9, Issues 1–2, 1992, Pages 105-113

https://doi.org/10.1016/0921-8890(92)90034-V Get rights and content

Abstract

Viewing each path planning problem as unique forces the planner to consider all available maneuvers in each new situation. This is exponental in the number of degrees of freedom of the robot. It is preferable to consider only certain key aspects of each problem, and to generate a plan based only on these aspects. This approach possesses two advantages: (1) it permits reuse of plans in similar situations, which reduces (or even eliminates) the cost of searching among alternative maneuvers. and (2) it reduces the number of aspects of the situation that must be sensed by the system, which reduces the sensory cost. In this paper, it is shown how to classify path planning problems according to their algebraic structure. A technique is given that is applicable to robots whose actions are reversible. Each class of problems is characterized by a particular set of maneuvers that solves all problems in the class.

References (10)

M.A. Arbib et al.
Foundations of system theory: Decomposable systems
Automatica
(1974)
S. Niizuma et al.
A problem-decomposition method using differences or equivalence relations between states
Artificial Intelligence
(1985)
E.D. Sacerdoti
Planning in a hierarchy of abstraction spaces
Artificial Intelligence
(1974)
D.P. Benjamin
Towards an effective theory of representation change. Part I: Semantics
D.P. Benjamin et al.
An introduction to the decomposition of task representations in autonomous systems

There are more references available in the full text version of this article.

Cited by (8)

The geometry of knowledge acquisition through motion and sensing
1992, Robotics and Autonomous Systems
In order to understand knowledge acquisition in autonomous systems, we introduce a representational framework in which knowledge about the world is represented as a point in a knowledge space. This is a homogeneous linear space, with an unusual vector product (the componentwise multiplication of vectors). We demonstrate how motions in the world lead to an induced linear transformation on the knowledge space, and how sensing leads to an induced reduction to a linear subspace. We also investigate how abstraction and reasoning, which are internal transformations of the knowledge, may be represented in the knowledge space.
We treat applications of the framework which demonstrated how it can be combined with Bayesian decision theory to compute optimal goal-directed behavior for both discrete and continuous cases.
Developing a cognitive architecture to be embedded in the physical world
2006, Simulation Interoperability Standards Organization - 15th Conference on Behavior Representation in Modeling and Simulation 2006
Obstacle avoidance using predictive vision based on a dynamic 3D world model
2006, Proceedings of SPIE - The International Society for Optical Engineering
Embodying a cognitive model in a mobile robot
2006, Proceedings of SPIE - The International Society for Optical Engineering
Designing a robot cognitive architecture with concurrency and active perception
2004, AAAI Fall Symposium - Technical Report
On the emergence of intelligent global behaviours from simple local actions
2000, International Journal of Systems Science

View all citing articles on Scopus

View full text

Reformulating path-planning problems by task-preserving abstraction

Abstract

Automatica

Artificial Intelligence

Artificial Intelligence

Towards an effective theory of representation change. Part I: Semantics

An introduction to the decomposition of task representations in autonomous systems