Добавил:
Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:
(EOD).Mechatronics.pdf
Скачиваний:
81
Добавлен:
23.08.2013
Размер:
5.07 Mб
Скачать

page 704

Figure 2.7 Multi-Robot WorkSpaces

Manipulator #1

Manipulator #2

Transfer

Point

 

Multi-Robot Work Scheme

(with possible collisions over transfer point)

40.2.8 INTERACTIVE PROGRAMMING

One factor that is now available for all commercial robots is the ability for the human user to plan the path. This may be done with a special graphical program, or by moving the robot and setting points, or by entering sets of points. These methods are proven, and do work, but they are very inefficient, inflexible, and prone to errors. Thus the thrust of this paper will be to ignore the interactive approach, because automated path planning is much more desirable. The interactive approach will almost guarantee that an operator will not find an optimal approach. Finding a good path is not possible without having to stop the robot and lose production time.

40.3 SETUP EVALUATION CRITERIA

The type of environment information required by a path planner is critical to its operation. Some simple path planners may work in a simple collision detection mode. Some path planners require pre-processed information about the environment, and they are much more difficult to deal with.

40.3.1 INFORMATION SOURCE

The sources of information, which a path planner uses, will have a fundamental effect upon how a path is planned. This is a problem, in how the information is collected. Data may be collected before or during the execution of a path. This means data may be presented in the form of an known obstacle (and position), or by a simple indication of contact. These tend to draw distinctions between methods which have Collision Detection (Local or Trajectory Path Planners) and those which have obstacle information (Global Path Planners).

page 705

40.3.1.1 - KNOWLEDGE BASED PLANNING (A PRIORI)

It is much easier to solve a problem if all the information is available at the beginning of the solution. In robotics we may plan paths before their execution, if we have some knowledge of the environment. This is strictly a ’blindly generate’ type of strategy that trusts the knowledge of the environment provided. Planning paths before execution allows efforts to get a shorter path time, more efficient dynamics, and absolute collision avoidance. When working in this mode a priori knowledge (i.e. Known before) is used. Techniques are available to solve a variety of problems, when given the a priori information. Some of the knowledge which we use for a priori path planning may come from vision systems, engineering specifications, or CAD programs.

A Priori Knowledge may be applicable to moving objects, if they have a predictable frequency or motion. A Priori knowledge may not be used for unpredictable or random motion, there is no detection method allowed by its definition.

A Priori Knowledge may be derived from modelling packages or High Level Sensors. These sensors are slow and typically drive a World Modeller in an Off-line programmer. The best example of this type of sensor is the vision systems. This system is the most desired information collector for robotics in the future. This system would provide a complete means of world modelling for the path planner. Another common detection system currently in use is the Range Imaging System, which use stripes of LASER light to determine geometry of objects. One example of this is the use of the system to recognize objects and use the geometrical information to determine how to grasp the object [K.Rao, G.Medioni, H.Liu, G.A.Bekey, 1989]. Some of these sensors require knowledge from the world modeller for object recognition purposes. In general these sensors are slower because of their need to interpret low level data, before providing high level descriptions.

Figure 3.1 A Priori Path Planning

Goal

 

 

Path

 

 

 

Manipulator

 

 

Motion

Planner

 

 

 

Controller

 

 

State

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

World

Modeller

40.3.1.2 - SENSOR BASED PLANNING (A POSTIERI)

Sometimes information is not available when we begin to solve a problem, thus we must solve the problem in stages as the A Postieri information becomes available. Sensor based planning is an indispensable function when environments change with time, are unknown, or there are inaccuracies in the robotic equipment. A Postieri Knowledge may be used to find the next trajectory in a path (by collecting information about the outcome of the previous trajectory) or even

Соседние файлы в предмете Электротехника