Swing Loader Traffic Control

This project has successfully demonstrated (on a 1/7th scale model swing loader and haul truck) the technology to direct drivers to a recommended loading position. It does this in two phases. In the first phase, a scanning range laser on the swing loader is used to generate a digital terrain map of the surrounding area. This is used to define the navigable space around the loader and the loading position closest to the bench (spotting). In the second phase, another scanning range laser is used to continually estimate the pose of the truck with respect to the swing loader. A path is then generated between the current pose and the desired loading position. The steering and speed response required to follow this path is displayed to the driver. If the driver chooses (or is unable) to follow the desired path, the system adapts and recalculates a new path.

The trials conducted at QCAT indicate that whilst the guidance technology is able to assist a driver to more accurately position the truck to the recommended loading position, it takes about twice as long to stabilize.

The difference in time is related to two factors. Firstly, the driver of the scale model truck has 360 degree visibility that enables them to make excellent judgments for steering strategies. Secondly, the stability of the path planning software was affected by errors in the measured orientation of the truck. It is apparent that with a reduction in visibility normally experienced by truck drivers in the mining environment, and improvements in localization that are predicted on full-size trucks, the path planning software has the potential to exceed the performance of the unassisted human operator.

In the original proposal for this project, there were plans to test the laser’s performance at a mine site and to monitor the typical movement of haul trucks and swing loaders. Unfortunately, we were unable to access a mine site, so we have been unable to estimate the precision and suitability of the laser tracking technology on a full scale truck, nor were we able to document the current practice for optimal spotting.  This is an area for future work.