Mohammad Matin MomeniRobotics & data systems
← Selected work

Robotics / Geometry / Webots

From readings
to structure.

A split-and-merge mapping experiment: using geometry to turn a robot’s distance-sensor observations into a more compact description of its surroundings.

Mapping output showing the estimated line segments from the split-and-merge experiment
Output from the project. Sensor observations are processed into line segments.

The question

A distance sensor gives individual observations. A map needs structure. This project explores how to move from one representation to the other, using the robot’s pose and a geometric line-fitting procedure.

The approach

  1. Interpret the sensors. Convert sensor outputs into distances using the lookup table and linear interpolation.
  2. Place the observations. Use the robot’s position and heading to express measurements as points in the environment.
  3. Split. Find the point with the greatest perpendicular distance from a candidate line; split a group when its deviation exceeds a threshold.
  4. Merge. Combine compatible segments to simplify the representation.

The engineering

The Webots controller brings together differential-drive kinematics, compass-based heading estimation, distance-sensor interpretation, and the split-and-merge procedure. NumPy and Matplotlib support the numerical work and inspection of the output.

I’m interested in this kind of problem because each layer matters: the interpretation of a sensor, the coordinate system, the geometry, and the final representation.

Back to all projects ↗