GalaxyRVR
A small robot navigation prototype that combines a view of the whole arena with cameras and sensors on the robot.
Drift Hackathon · 9 November 2025
Arena Navigation
We explored how a small robot could find its way to a target while accounting for obstacles. An overhead camera provides the wider arena view, while the robot's own camera and sensors describe what is nearby. The planner uses those inputs to choose a route and turn it into motor commands.
The hackathon prototype also includes simulation for trying parts of the navigation logic without the physical robot. It brings sensing, route planning, and movement together as an experiment, rather than demonstrating reliable autonomous operation.
Connecting Arena Vision to Motor Control
Locating the Robot Between Camera Updates
Calibration maps camera pixels to world coordinates so the robot, target, and planned route can share a coordinate system. Camera observations locate the robot, while dead reckoning estimates movement between observations using motor commands. A fresh camera update corrects that estimate. The navigation architecture places this position estimate before route planning.
Sensing, Planning, and Control
| Component | Responsibility |
|---|---|
| Overhead camera | Provides arena, target, and robot positions through Python and OpenCV. |
| Onboard camera and sensors | ESP32-CAM, ultrasonic, and infrared inputs describe nearby conditions. |
| Planner | Generates waypoints around obstacles and updates the route. |
| Controller | Turns navigation output into left and right motor commands through serial and Arduino firmware. |
Following and Adjusting the Route
The planner produces waypoints around obstacles, and the controller translates those waypoints into separate left and right motor commands. Sensor feedback then informs the next route adjustment. The overhead view supplies the wider path context, while the onboard camera, forward ultrasonic sensor, and side infrared sensors describe nearby conditions. Simulation provides another way to exercise the navigation logic without commanding the physical robot.
Team
Project Credits
The navigation builds on Drift's hackathon starter code.