← All projects
03 / Autonomy simulation · 2026

Browser-Based Drone Simulator

An interactive visualization that makes indoor exploration, route planning, collision avoidance, and incremental 3D mapping visible in real time.

Overview

This simulator communicates the systems-level decisions behind autonomous indoor flight without requiring specialized software or hardware. A quadcopter explores a multi-story home, discovers navigable space, and incrementally reconstructs the geometry it observes. The entire experience runs in the visitor’s browser. This is an illustrative simulation, not a validated flight-control or SLAM system.

Navigation

The vehicle uses a graph-based frontier planner that prioritizes unmapped regions instead of following a fixed animation. Known walls influence future route selection, collision volumes keep the airframe clear of architecture and furniture, and the staircase forms the only traversable connection between floors. Once useful coverage is complete, the planner transitions into free exploration.

Mapping System

Simulated sensor rays intersect visible surfaces and feed a shared map representation. The map can be viewed either as a sparse point cloud or as a voxel-based 3D reconstruction, with both views building from the same observations. This makes the relationship between perception, coverage, and navigation immediately legible.

Flight Model

The motion model includes acceleration limits, speed-dependent pitch, turn-induced roll, vertical transitions, propeller-speed variation, and a trailing third-person camera. These cues are intentionally restrained: the goal is not high-fidelity aerodynamics, but believable vehicle behavior that supports the autonomy visualization.