The PinPort online laboratory has introduced a 3D simulation of a city designed for robotics education. The environment features a six-wheeled delivery robot, R-D6, which is controlled by actual Arduino Uno sketches. The simulation runs in a browser, executing code through an ATmega328P emulator.
The system integrates three synchronized components: an AVR emulator running at 16 MHz, the Rapier physics engine with a 2 ms step, and a city logic layer. To ensure deterministic results across different hardware, the simulation avoids real-time dependencies, relying instead on fixed physics steps and a seeded random number generator.
The virtual city consists of a 3x3 grid with 8-meter wide roads and 3-meter sidewalks. It includes 12 vehicles, 20 pedestrians, and 5 traffic lights with a 50-second cycle. Vehicles follow traffic rules, while pedestrians move along sidewalks and cross streets based on traffic light signals.
Navigation is handled by a Dijkstra-based algorithm using a 'carrot' point 2.5 meters ahead of the robot to maintain trajectory. A built-in judge system monitors performance, tracking metrics such as delivery time, traffic violations, and collisions, which are verified via state hash comparisons.
The simulation includes a custom WebGL renderer for 3D visualization and a 2D Canvas view for debugging. Users can test their code against various scenarios, including night conditions, peak hours with 40 pedestrians, and road construction obstacles, with delivery times ranging from 65 to 135 seconds.
