CBF-Critic-Based Heading-Aware MPPI Navigation for Omnidirectional Mobile Robots

As the deployment of autonomous dual-arm wheeled humanoid robots and heavy-duty logistics robots accelerates, ensuring dynamic stability and collision-free navigation in narrow environments is critical. While 4-wheel independent steering (4WS) platforms offer superior maneuverability, their asymmetric rectangular footprints lead to severe collisions and excessive heading variations when managed by conventional circular-approximated controllers. To address these geometric limitations, this paper proposes a CBF-Critic-Based Heading-Aware MPPI navigation architecture. By modeling the robot’s footprint as an Oriented Bounding Box (OBB), we integrate Control Barrier Functions (CBF) into the Model Predictive Path Integral (MPPI) framework. This effectively approximates collision avoidance as a hard constraint, mitigating cost conflicts among multiple objectives. Furthermore, an online Adaptive Heading Critic dynamically enforces heading alignment in narrow passages using real-time LiDAR, suppressing unnecessary angular jerk. Extensive validations in simulation and on real-world 4WS hardware demonstrate a 100% collision-free success rate in our experimental scenarios. Particularly in the region-locked parallel parking scenario, it was demonstrated that the proposed framework maximizes dynamic stability by reducing the maximum angular jerk by approximately 91.6% compared to the conventional soft-constraint baseline.