Survey rover in an oil palm plantation

A rover surveying a virtual oil palm plantation, driven by a ROS 2 control loop that senses obstacles with simulated LiDAR and logs the fruit bunches it passes. The ground is packed particles solved by NDEM, and the wheels leave tracks in it.

Oil palm plantations run in long, repetitive rows — which is exactly the kind of work worth handing to an autonomous robot. This is a rover surveying a virtual one, to see how it behaves before any hardware goes into the field.

The rover is controlled by a ROS 2 node in a closed feedback loop. It uses simulated LiDAR to sense obstacles, steer around them, and scan and record the palm fruit bunches scattered on the ground.

The ground is not just a flat contact surface. It is made up of packed particles solved by VeloxSim's NDEM solver, and a terrain texture follows the particle-based deformation, so the wheel tracks appear as the rover drives through.

The controller is rule-based, and the fruit identification uses labels from the simulation rather than a trained perception model — there is still some way to go before this runs in a real plantation. The intent is surveying, checking where the fruit is, and planning collection, in an industry that is short of labour.

Simulated with NDEM, our real-time granular engine.