As autonomous driving technology advances, its application to construction machinery is gaining
momentum, offering benefits in productivity, operator fatigue reduction, and industrial safety. However,
machines such as wheel loaders, which feature arti...
As autonomous driving technology advances, its application to construction machinery is gaining
momentum, offering benefits in productivity, operator fatigue reduction, and industrial safety. However,
machines such as wheel loaders, which feature articulated steering and a high center of gravity, differ
structurally from on-road vehicles, making direct application of conventional autonomous driving
methods difficult and requiring explicit consideration of rollover risk. This study addresses these issues
by proposing a nonlinear model predictive control (NMPC)–based control framework that integrates
local path planning and motion control for an articulated wheel loader performing work operations.
The prediction model is formulated as an articulated kinematic model, where position, heading, and
articulation angle are treated as states, and longitudinal velocity and articulation steering rate as control
inputs. Instead of explicitly modeling roll dynamics, a rollover indicator based on the lateral load transfer
ratio (LTR) is approximated from vehicle parameters, and evaluated over the prediction horizon. This
index is then imposed as an inequality constraint so that the predicted LTR does not exceed a predefined
threshold. The cost function is designed in a heading-aligned coordinate frame to independently weight
longitudinal and lateral position errors as well as heading error, while additional constraints account for
velocity and steering limits and circular safety margins for obstacle avoidance.
The software architecture is built on ROS 2, separating a control node (MATLAB/Simulink) and
51
an optimization node (Python with CasADi–IPOPT), which communicate via ROS topics and run in
real time on a low-power ARM-based PC. The proposed framework is validated through Vortex Studio
simulations and full-scale experiments under V-shape loading and obstacle-avoidance scenarios. Results
demonstrate that the NMPC structure maintains path-tracking performance while keeping the LTR below
its threshold, thereby effectively reducing rollover risk during articulated wheel-loader operations.