SIMLAN — Simulation for Multi-Camera Robotics

SIMLAN — Simulation for Multi-Camera Robotics

December 31, 2024

SIMLAN (Simulation for Multi-Camera Robotics) is a simulation environment based on Ignition Gazebo and ROS 2 that resembles a Volvo Trucks warehouse. It serves as a playground for rapid prototyping and testing of systems relying on a multi-camera setup for perception, monitoring, localization, or navigation.

Demo

GPSS — Generic Photo-based Sensor System

Inspired by the Volvo GPSS system using ceiling-mounted cameras for warehouse navigation:

GPSS controls (ArUco localization, Nav2 navigation):

RITA — Robotic Arm and Humanoid

Panda arm and humanoid worker integration:

Installation

Features

  • Ignition Gazebo with a library of realistic warehouse assets
  • Real-world environment design inspired by ceiling-mounted camera layouts
  • ROS 2 interfaces (Humble and Jazzy)
  • Simple GPSS (Generic Photo-based Sensor System) navigation
  • Multi-robot localization and navigation using Nav2
  • ArUco marker localization
  • Bird’s-eye view projection and camera stitching
  • Multi-sensor support (LiDAR, RGB, semantic segmentation, depth)
  • Geofencing for safe zones and safe stop on collision
  • Humanoid worker model with deep-learning-based pose capture and replay
  • Panda robotic arm with pick-and-place scenarios

Documentation

Research Funding

Carried out within:

  • SMILE IV — Vinnova, FFI grant 2023-00789
  • EUREKA ITEA4 ArtWork — Vinnova grant 2023-00970

SIMLAN GitHub repository