Work

Safe and Efficient Mobile Robot Navigation using Sampling-based Optimal Control with Barrier Function Constraints

Year: 2025
Type: article
Abstract: Safe and efficient navigation in constrained environments presents a significant challenge for omnidirectional Mecanum-wheeled mobile robots, especially when operating in dynamic or cluttered spaces. Many existing control approaches struggle to balance real-time performance with formal safety guarantees, creating a gap between theoretical models and practical implementation. To address this, we propose a hybrid Model Predictive Path Integral (MPPI) controller that incorporates Control Barrier Functions (CBFs) to enforce safety constraints during trajectory optimization. The research contribution is a unified framework that combines the flexibility of stochastic samplingbased planning with the rigor of formal safety enforcement, implemented using the GPU-accelerated JAX framework to enable fast, parallel trajectory evaluation at high control frequencies. Our method integrates CBFs through both cost shaping and quadratic programming (QP)-based filtering, allowing unsafe trajectories to be dynamically penalized or rejected while preserving the efficiency and adaptability of MPPI. Extensive simulation studies compare the proposed controller to standard MPPI, CBFaugmented variants, and a conventional Model Predictive Control (MPC) baseline. Results demonstrate that the hybrid controller consistently achieves superior obstacle avoidance and trajectory efficiency. It performs robustly even in densely constrained environments. In particular, across multiple scenarios the proposed controller improves minimum clearance by up to 70% compared to standard MPPI and reduces the number of control steps to the goal by approximately 15% compared to quadratic programming (QP)-based filtering, while maintaining real-time feasibility. These findings validate the effectiveness of combining MPPI with many CBFs, offering a practical, high-performance solution for real-time robotic navigation.
Language: English
Open Access status: diamond
OpenAlex ID: W7160964299
FWCI: 0.443
Cites: 84
Cited by: 2
Related to: 10