Alex Meystel

Minimum time path planning for a robot

A new algorithm of minimum time motion planning is proposed for robots operating in the obstacle strewn environment. Structure of the topological passageways is analyzed and represented using a model of slalom situations for which a number of rules is determined. Dynamical system of robot is described in a form of sequential machine. This enabled a merger between two kindred algorithms: A* search algorithm, and dynamic programming. An experimental analysis of the simulated mobile robot has confirmed the applicability of results.

Multiresolutional hierarchical decision support systems

Decision support systems gain better performance and higher accuracy by the virtue of building multiresolutional (multigranular, multiscale) representation, and employing multiscale behavior generation subsystem (planning and control). The latter are equipped by devices for unsupervised learning that adjust their functioning to the results of self-identification. We demonstrate that planning and learning are joint processes. The authors intention is to emphasize that the concepts of multiresolutional representation (MR) and multiresolutional decision support (MR-DSS) probably have in common a general significance that crosses the boundaries of particular domains of applications and disciplines. The paper explores this phenomenon. The ubiquity of a principle that somehow persistently delivers benefits to many areas of knowledge and technology seems to be more important than a habit to follow the pigeonhole principle of paper presentation.

A genetic technique for robotic trajectory planning

Abstract There are many multi-stage optimization problems that are not easily solved through any known direct method when the stages are coupled. For instance, the problem of planning a vehicles control sequence to negotiate obstacles and reach a goal in minimum time is investigated. The vehicle has a known mass, and the controlling forces have finite limits. A genetic programming technique is developed that finds admissible control trajectories that tend to minimize the vehicles transit time through the obstacle field. The immediate application is that of a space robot that must rapidly traverse around two or three dimensional structures via application of a rotating thruster or non-rotating on-off thrusters. (An air-bearing floor test-bed for such vehicles is located at the Marshal Space Flight Center in Huntsville, Alabama.) It appears that the developed method is applicable to a general set of optimization problems in which the cost function and the multi-dimensional multi-state system can be any non-linear functions that are continuous in the operating regions. Other applications include: the planning of optimal navigation pathways through a traversability graph, the planning of control input for underwater maneuvering vehicles which have complex control state-space relationships, the planning of control sequences for milling and manufacturing robots, the planning of control and trajectories for automated delivery vehicles, and the optimization of control for racing vehicles and athletic training in slalom sports.

Multiresolutional Pyramidal Knowledge Representation And Algorithmic Basis Of IMAS-2

This paper describes recent research results in the area of autonomous robotics. A multiresolutional (pyramidal) system of world representation is applied to the domain required for Intelligent Mobile Autonomous Systems for outdoor application. Problems of generalization, accuracy, and adequacy of representation are formulated for a system with vision and ultrasonic sensors. A method of nested hierarchical planning is applied for this system of representation. A Pilot with behavioral duality is proven to be efficient for a mobile robot. An execution controller is explored based upon a production system providing near-minimum time operation.

Multiresolutional schemata for unsupervised learning of autonomous robots for 3-D space operation

Abstract This paper describes a novel approach to the development of a learning control system for autonomous space robot (ASR) that presents the ASR as a “baby”—that is, a system with no a priori knowledge of the world in which it operates, but with behavior acquisition techniques that allow it to build this knowledge from the experiences of actions within a particular environment (we will call it an Astro-baby). The learning techniques are rooted in the recursive algorithm for inductive generation of nested schemata molded from processes of early cognitive development in humans. The algorithm extracts data from the environment, and by means of correlation and abduction, it creates schemata that are used for control. This system is robust enough to deal with a contantly changing environment because such changes provoke the creation of new schemata by generalizing from experiences, while still maintaining minimal computational complexity, thanks to the systems multiresolution nature. Experimenting with ASR is especially interesting because the rules of input control do not coincide with human intuitions. Actually, we want to see that the simulated device can learn unexpected schemata from its own experience. Although the traditional approach to autonomous navigation involves off-line path planning with a known world map (such as the potential fields algorithm), in most of the real tasks the environment is not well knowm because of everchanging conditions such as absence of gravity and because of sophisticated, hard-to-predict obstacles like components of the space station. Astro-baby gathers data from its sensors and then, by using a schema-discovery system, it extracts concepts, forms schemata, and creates a quantitative/conceptual semantic network. When the Astro-baby is first dropped into space, it does not have any experiences and its sensors and actuators are sets that do not have any distinction among elements. Then, by trial and error, the ASR learns the function of its actuators and sensors and how to activate them to achieve the goal given by its creator or the sub-goals that it finds. In our simulation, the initial goal is to minimize the distance to a beacon. The results of simulation are positive: Astro-baby displays the ability to learn a number of maneuvers.

Hierarchical Parallel Search for Markov Control with Enhanced Selector

A new method of hierarchical parallel search is proposed for dealing with Markov planning/control processes in systems with uncertain information. It is based on a new concept of analyzing alternative with uncertain cost evaluation. Under definite conditions, instead of making an immediate choice based on expectation of cost at each step of the search, it is recommended to postpone the final decision until information is improved, and the uncertainty is reduced. In addition to elementary alternatives, their combinations are also considered for possible pursuit. ‘The best set’ of rough elementary solutions is to be determined at the upper of two adjacent planning/control levels, then all elementary alternatives of this set as well as their combinations, are being pursued at the lower level with a higher resolution of information, while evaluation of the ‘remaining cost’ for each of the alternatives, is being constantly improved due to the process of evolutionary uncertainty reduction. This bilevel process is easily extendible over the whole hierarchy of the system. The method is working in the graph-search and dynamic programming paradigms. The set of problems to be solved is formulated and some of them are addressed. Various applications are contemplated.

Minimum cost path planning for autonomous robot in the random traversability space

Traversability space is used as the representation in the path planning system of an autonomous robot. A second level of traversability space is introduced to reduce computational complexity and make the problem of control tractable. This generalized level of representation is used to guide search in the original traversability space. This is done with successively smaller envelopes of search and the results are analyzed with respect to percent error from the optLmal path and percent reduction in computational complexity.

Minimum time path planning for robot motion in obstacle strewn environment

A new algorithm of minimum time motion planning is proposed for robots operating in the obstacle strewn environment. Structure of the topological passageways is analyzed and represented using a model of slalom situations for which a number of rules is determined. Dynamical system of robot is described in a form of sequential machine. Thls enabled a merger between two kindred algorithms: A* search algorithm, and dynamic programming. An experimental analysis of the simulated mobile robot has confirmed the applicability of results.

Robot Path Planning

The sections in this article are 1 Planning as a Reaction to Anticipation 2 Planning as a Part of Behavior Generation 3 Classification of Robot Planning Problems 4 Planning of Actions Versus Planning of States 5 Linkage Between Planning and Learning 6 Planning in Architectures of Behavior Generation

Intelligent module for planning/control of master-dependent systems

An intelligent module is proposed in this paper which is capable of performing a joint recursive planning/control operation which propagates through the intelligent module at all planning/control levels simultaneously. Each of the actuators is equipped by an intelligent module, and all of these modules are working independently and concurrently. The model of the world is being constantly updated based upon vision and a multiplicity of other available sensors, and at various resolutions is submitted to each of the intelligent modules as applied to particular properties of the link being controlled by this module. Together these intelligent modules are working as a team of the actuators controllers, and all decisions are constantly negotiated among the members of the team. Each of the resolutional levels within the actuator intelligent modules is using the following set of planning/control tools: learning rule base, learning heuristic search, and situational decision generator which are first applied simultaneously, and then one of them takes over. A neural network is collecting information about the progress and the results of planning/control processes, and is modifying heuristics, as well as enriching the system of rules. The same neural network is used for supplying the provisional analytical model required for the lowest levels of execution control. Provisional analytical models are applied in a simple form which allows for simple real-time controller operation, and the parameters of this provisional model are constantly being updated by the neural network.