RRT Motion Planning: Expansion, Exploration, and Variants

All flashcards

Rapidly exploring random tree (RRT)

A sampling-based planner that grows a tree from an initial state through feasible extensions toward sampled states.

Voronoi region in an RRT

The part of the space closer to one tree vertex than to any other; larger regions are more likely to receive samples.

RRT growth limit

A limit on extension length; if a sample is farther away, the proposed state lies only the permitted distance toward it.

Goal-biased sampling

A strategy that assigns some probability to sampling the goal, encouraging growth toward it while potentially reducing broader exploration.

RRT*

An RRT-based method that considers nearby connections and rewires the tree when it finds a better route.

Quiz questions

1. What is the basic RRT primarily designed to find?Answer: A feasible path through the search space
2. How does LQR-RRT handle systems with complex or underactuated dynamics?Answer: It uses control-oriented local planning
3. How does RRT*FN keep the number of nodes fixed?Answer: It randomly removes a leaf node during an iteration
4. What helps RRV expand around obstacles and through narrow passages?Answer: Dominant eigenvectors around tree nodes
5. What does Real-Time RRT* do to support planning in dynamic settings?Answer: It can move the tree root with the agent while retaining sampled paths