Warehouse tote-moving robot with ROS 2 and Nav2 in Gazebo

Riverbend picks orders for online retailers. Pickers fill totes in six pick zones and walk them to four pack stations, an average round trip of 140 m. Walking takes about 35% of each picker's shift. Riverbend wants to pilot two autonomous mobile robots (AMRs) to carry totes so pickers can stay in their zones.

Before buying a fleet, the operations director wants a simulation that answers three questions: - Can one robot deliver at least 30 totes an hour on the real layout? - Does it slow down and keep a safe distance in aisles shared with people? - Can the warehouse team change routes, keep-out zones and stations without a programmer?

The robot is a differential-drive base, 0.7 × 0.5 m, carrying up to 25 kg. It has a 2D lidar with a 12 m range and 270° field of view, wheel encoders and an IMU. Speeds are 1.2 m/s maximum in open areas and 0.5 m/s in aisles of 2.4 m where people walk. Pack stations need the robot to stop within 5 cm and 5° of a docking pose.

You will do all of this in simulation: - write a URDF model of the robot with its sensors; - build a warehouse world in Gazebo from the layout in task 1; - map it with SLAM Toolbox; - configure and tune Nav2 for localisation, planning and control; - write a mission node that takes tote requests and drives the robot through stations, using keep-out and speed-limit zones; - run a one-hour test with simulated people and report the numbers.

You need no robot. ROS 2 (Humble or Jazzy) and Gazebo run on Ubuntu or in a Docker container.

Done looks like a repo that launches the whole simulation with one command, a map, a tuned Nav2 configuration, a mission node, and a test report with deliveries per hour, docking accuracy, near-miss count and minimum distance to people.