RRT (Rapidly-exploring Random Tree)

What is RRT (Rapidly-exploring Random Tree)?

RRT (Rapidly-exploring Random Tree) is a sampling-based path-planning algorithm that builds a tree of feasible states by randomly sampling the configuration space and growing the tree toward each sample. RRT excels in high-dimensional spaces where grid-based methods like A* fail.

How It Works

Start with a tree containing only the start state. Randomly sample a point in the configuration space. Find the nearest node already in the tree. Extend a short distance from that node toward the sample. If the extension is collision-free, add it as a new node. Repeat until a node is close enough to the goal. RRT is probabilistically complete and easy to implement, but the resulting path is not optimal. RRT* and informed-RRT* variants converge to optimal paths in the limit.

Real-World Example

MoveIt2 ships RRT-Connect as its default arm planner. NASA's Mars rovers use RRT-derived planners. Tesla's autonomous lane-change planner is RRT-based. Indian robotics groups use RRT for 6+ DOF arm planning.

Why It Matters for Robotics

RRT is the workhorse algorithm for motion planning in spaces too big for grid search. Knowing RRT, RRT*, and their variants is mandatory for manipulation and autonomous-driving engineers. Indian PhD interviews routinely test RRT theory.

Try It Yourself

Open /visualizer and load the RRT demo. Draw obstacles in a 2D maze, place start and goal, and step through RRT one sample at a time. Watch the tree grow and the first feasible path emerge.

Quick Quiz

🧠
Quick Quiz
3 questions
1.RRT samples:
A.Only the start state
B.Random points in the configuration space
C.Only the goal
D.Battery levels
2.Compared with A*, RRT is better at:
A.Small 2D grids
B.High-dimensional continuous spaces (e.g., 7-DOF arms)
C.Database lookups
D.Audio compression
3.RRT* differs from RRT by:
A.Using gradient descent
B.Re-wiring the tree to converge toward optimal paths
C.Running slower for no reason
D.Using sound waves

Further Reading

• Path Planning
• Motion Planning
• A Star Algorithm
• Inverse Kinematics

Ask R2 About This

Open the R2 Co-pilot (press ⌘K anywhere on R2BOT) and ask: "Explain RRT (Rapidly-exploring Random Tree) for a Class 9 student in India, with one real-world Indian example." You'll get a tailored, sourced answer in seconds.