Plan Space Planning

This video lesson is available to enrolled students.

Enroll to watch — ZERO TO HERO

AI Summary

An AI-generated summary of this video lecture.

The lecture introduces Plan Space Planning, also known as Non-Linear Planning, contrasting it with traditional state-based planning. The instructor explains that this approach works with plan structures, allowing the modification or extension of any part of the plan to reach a final solution. Key characteristics include the ability to shift attention midway and solve complex problems like the Sussman anomaly. The lecture transitions to Partial Order Plans, defining them as sets of actions with a partial ordering. An example of getting ready in the morning, specifically putting on shoes, is used to illustrate how independent actions can be ordered flexibly without affecting the final goal.

Chapters

1. 0:00 – 2:00 00:00-02:00

   The instructor introduces Plan Space Planning (Non-Linear Planning) using a slide that highlights its ability to modify or extend any part of a plan. She writes 'Search space of plans' on the board, contrasting this with state-based planning where the planner looks at a state and a goal. The slide text emphasizes that you can 'modify any part of the plan' and 'shift their attention midway,' which allows solving problems like the Sussman anomaly. The instructor notes that unlike linear planning, this approach is not constrained to focus on one sub-goal continuously.

2. 2:00 – 4:59 02:00-04:59

   The lecture moves to Partial Order Plans, defining them as a set of actions with a partial ordering representing a 'before' relation. The instructor writes 'plan -> refine & define totally' and 'plan flexibility' on the slide. She provides an example of 'Getting ready in the morning to go out,' showing a diagram where 'LeftShoeOn' and 'RightShoeOn' are parallel actions between Start and Finish. She explains that whether the left or right shoe is put on first does not matter, illustrating the concept of partial ordering where total ordering is not strictly required for independent actions.

The video progresses from a high-level definition of Plan Space Planning to specific implementations like Partial Order Plans. Initially, the instructor establishes that Plan Space Planning searches the space of plans rather than states, offering flexibility to modify plans and solve anomalies like the Sussman anomaly. This sets the stage for Partial Order Plans, where actions are defined by a partial ordering. The shoe-putting example concretely demonstrates this concept: while actions must happen between start and finish, the specific order of independent actions (left vs. right shoe) is irrelevant, showcasing the efficiency and flexibility of non-linear planning strategies.