planning in the real world time and resources hierarchical task network conditional planning...

32
Planning in the Real World Time and Resources Hierarchical Task Network • Conditional Planning • Execution Monitoring and Replannin • Continuous Planning • MultiAgent Planning • Summary

Upload: alan-wilkinson

Post on 21-Dec-2015

217 views

Category:

Documents


1 download

TRANSCRIPT

Planning in the Real World

• Time and Resources• Hierarchical Task Network• Conditional Planning• Execution Monitoring and Replanning• Continuous Planning• MultiAgent Planning• Summary

Time Constraints

Planning gets complicated in:

• Scheduling Hubble Space Telescope• Operating Factories• Manufacturing• Cooperating Agents

Time Constraints

Goal: Complete jobs within time requirement.

Start:Chassis(C1) ^ Chassis(C2) ^Engine(E1,C1,30) ^ Engine(E2,C2,60) ^Wheels(W1,C1,30) ^ Wheels(W2,C2,15)

Goal: Done(C1) ^ Done(C2)

Time Constraints

First Step: Partial Order.

Start

AddEngine 1 30

AddWheels 1 30

Inspect 1 10

Finish

AddEngine 2 60

AddWheels 2 15

Inspect 2 10

Critical path

Resource Constraints

Resource(k): k units of resource are needed by the action.

The resource is reduced by k during the duration of the action.

Resource Constraints

Example:

Action(AddEngine(e,c,d)Precond: …

Effect: … Resource: EngineHoists(1))

Other examples: Inspectors(2)

Planning in the Real World

• Time and Resources• Hierarchical Task Network• Conditional Planning• Execution Monitoring and Replanning• Continuous Planning• MultiAgent Planning• Summary

Hierarchical Task Network

High Order Goal

Subgoal 1 Subgoal 2 … Subgoal n

Build House

Get Land Construct Pay Builder

Decomposition

• A decomposition should be a correct implementation of the action (plan should be complete and consistent).• A subplan is attained by unifying its variables with the state of the environment.

Hierarchical Task Network HTN is complicated (undecidable). Recursion is a problem. Allows subtask sharing. It is in general more efficient than naïve planning (linear rather than exponential).

A Successful Story

Successful story: O-Plan (Bell and Tate ‘85) It helps develop production plans for Hitachi (350 products, 35 assembly machines, 2000 operations).

It generates a schedule for 30 days with 8-hrshifts. The plan has tens of millions of steps.

Planning in the Real World

• Time and Resources• Hierarchical Task Network• Conditional Planning• Execution Monitoring and Replanning• Continuous Planning• MultiAgent Planning• Summary

Conditional Planning

What to do with incomplete and incorrectinformation?

Assume “bounded indeterminacy”

Solution: Construct a conditional plan with branches to consider all sorts of contingencies (include sensing actions)

Actions

Actions can have disjunctive effects:

Example: Vaccum CleanerAction(Left, Precond: At Right Effect: At Left V AtRight)

Effects

Also add conditional effects:

Example: Vaccum CleanerAction(Suck, Precond: Effect: When (At Left) CleanL When (AtRight) CleanR)

Example

Vacuum cleaner: Sometimes deposits dirt when it moves to a clean destination square. Sometimes deposits dirt if suck is applied to a clean square.

Planning in the Real World

• Time and Resources• Hierarchical Task Network• Conditional Planning• Execution Monitoring and Replanning• Continuous Planning• MultiAgent Planning• Summary

Execution Monitoring

Check to see if all is going according to the plan.

Re-planning agents repair old plans if something goes wrong.

Uses action monitoring to repair and continue with the plan.

Planning in the Real World

• Time and Resources• Hierarchical Task Network• Conditional Planning• Execution Monitoring and Replanning• Continuous Planning• MultiAgent Planning• Summary

Continuous Planning

o The agent persists in the environment indefinitely. o The agent is part of the way through executing a plan.

Example:Blocks-world problem.

Planning in the Real World

• Time and Resources• Hierarchical Task Network• Conditional Planning• Execution Monitoring and Replanning• Continuous Planning• MultiAgent Planning• Summary

MultiAgent Planning

Interaction can be

• Cooperative Joint plan A goal is achieved when each agent performs its assigned actions.

• Competitive

Cooperative

Play Tennis (doubles)

Plan 1: A: [Go(A, [Right,Baseline]), Hit(A,Ball)] B: [NoOp(B), NoOp(B)]

Cooperative

A common solution is to have a convention (constraint on joint plan).Conventions arise in evolutionary processes. Example:

• An ant colony. • Flocking behavior of birds

Besides convention you may have communication.

Planning in the Real World

• Time and Resources• Hierarchical Task Network• Conditional Planning• Execution Monitoring and Re-planning• Continuous Planning• Multi-agent Planning• Summary

Summary Actions need resources (time being an important one) Planning can be divided in a hierarchy of subtasks Conditional plans help under uncertainty Execution Monitoring detects violations Replanning tries to repair plans Continuous planning creates new goals Multiagent planning is needed when different agents interact.