RRT and PRM explore how a robot can find a route through a two-dimensional environment.
01 · Sample
Explore before you commit.
RRT grows a tree through sampled points, steering toward new configurations and checking the connecting edges.
02 · Connect
Build a map of what is possible.
PRM constructs a roadmap. A* searches its connections for a route between a start and a goal.
03 · Follow
One route. Many decisions.
The highlighted route shows the idea. My recorded planner output is further down the page.
04 · Evaluate
54 configurations. Better questions.
I ran 27 RRT and 27 PRM parameter configurations to see how each setting changes the route.
ROBOT MOTION PLANNINGBlender illustration of a planning scene Read the case study
Scroll or use arrow keys on the slider
Individual coursework · ROBO 201 / 202 · Fall 2025
Robot Motion Planning
Exploring how a robot finds its way through obstacles.
PythonRRTPRMA* search
At a glance
Role
Implemented RRT, PRM and A* roadmap search
Team
Individual coursework
Courses
ROBO 201 Motion Planning Algorithms for Robotics and ROBO 202 Software Development for Robotics
Built with
Python, on the course’s provided interface
Result
Ran RRT and PRM across 54 configurations (27 each)
01 / Purpose
What I set out to do.
Explore how sampling-based algorithms construct a collision-free route between a start and a goal in a two-dimensional environment.
02 / Approach
How it came together.
The RRT implementation grows a tree through sampled points, steering and checking connections for collisions. PRM constructs a roadmap and uses A* to search for a route. A test harness covers 27 RRT and 27 PRM parameter configurations.
03 / Contribution
What I built.
I implemented the RRT and PRM planners and the A* roadmap search in Python, on top of the course’s provided interface. I also built the route visualizations and the 3D extension, and ran the 54-configuration parameter sweep.
04 / Demonstration
See it working.
Below, my RRT runs live in your browser with the same settings as my coursework. After it, my recorded planner visualizations show the sampled nodes, connections, obstacles and the selected path. The 3D extension plans through free space; adding 3D obstacles is the next step.
Try it · live in your browser
Watch my RRT planner grow a tree
This is my Python RRT ported line for line: uniform sampling, nearest node, a fixed step toward the sample and a collision check on every new edge. Tap the map to move the goal.
Step size
Obstacles
Growing the tree…
Same settings as my coursework: a 10 × 10 world, start at (1, 1), goal at (9, 9), a 0.5 goal radius, up to 5,000 iterations, and the step sizes and obstacle counts from my parameter sweep. Obstacles are random circles and rectangles, kept off the start and goal.
View image Recorded RRT visualization from the coursework repository, with a sampled tree and highlighted path.Husain Altelly, ROBO 201/202 coursework; assignment interface/template credited to the course.View image Recorded PRM visualization showing the roadmap and a highlighted path found using A* search.Husain Altelly, ROBO 201/202 coursework; assignment interface/template credited to the course.
View image Extension: recorded RRT output in 3D free space, without obstacles.Husain Altelly, ROBO 201/202 coursework.
05 / Results
What I achieved.
I ran both planners across 54 parameter configurations (27 RRT, 27 PRM), varying the sampling, connection radius and step size, and plotted the route each one found.
06 / Limitations
Limits and next steps.
Each configuration ran once, without fixed random seeds or identical obstacle maps, so the plots show behavior rather than averages. Seeded, repeated runs would be my next step toward a fair comparison.
07 / Lessons learned
What I’m taking forward.
Algorithm visualizations help explain behavior, while fair performance comparisons also require reproducible environments and repeated measurements.
The Blender scenes and animations are illustrations made for this portfolio, not recorded results. The YouTube videos show the real robots and lab demos, and the AURORA film includes footage from the interactive learning demo.
R. R. Ma, L. U. Odhner, A. M. Dollar, “A Modular, Open-Source 3D Printed Underactuated Hand,” ICRA 2013.
The Yale OpenHand Project is an initiative to advance the design and use of robotic hands designed and built through rapid-prototyping techniques in order to encourage more variation and innovation in mechanical hardware. Please visit the Yale OpenHand site for more details.
