Waste-E is an AI-powered robotic system built using CARLA Simulator and ROS2, designed to autonomously drive along sidewalks and collect garbage in defined Areas of Interest (AOIs). It can be monitored and controlled via a connected application.

• CARLA Simulator v0.9.13
• carla-ros-bridge
• ROS 2 Foxy Fitzroy (Ubuntu 20.04 Focal)
• Docker & Docker Compose

To run the CARLA simulator with ROS bridge and all dependencies:

docker compose build
docker compose up

This will launch:

• CARLA server
• ROS2 container with carla-ros-bridge
• RViz2 (optional visualization)
• ROS nodes for spawning the AI robot

Make sure your environment supports GPU rendering (or offscreen mode if on a headless server).

To download every required file

./data_download.sh

• ✅ Pedestrian-level autonomous navigation
• ✅ Sidewalk path planning
• ✅ Onboard RGB camera and perception
• ✅ ROS2 integration for modular control
• ✅ Configurable pickup points via app interface (WIP)

• Garbage detection
• Target zone navigation
• Pick-and-drop mechanism
• Human-safe motion planning

The system will integrate with a mobile/web app to:

• Select garbage zones
• Monitor real-time location
• Trigger missions

Tested on:

• Ubuntu 20.04 (Focal)
• Docker version 24+
• NVIDIA GPU with CUDA support

• Ensure CARLA is run with --RenderOffscreen if using headless servers like Lambda Labs.
• Attach cameras to robots using carla_spawn_objects or Python API if needed.
• Adjust lane detection to follow LaneType.Sidewalk.

MIT License © 2025 Elmond Pattanan