Development of an integrated estimation and predictive control framework for safe navigation in mobile robots for industrial environments

The increasing integration of mobile robots in industrial environments has raised critical safety concerns, particularly in shared workspaces with human operators. Effective collision avoidance is essential to prevent accidents and enable smooth navigation in dynamic and unpredictable settings. This paper presents the development of an integrated estimation and predictive control framework for safe navigation in mobile robots for industrial environments. Here, safe navigation refers to improved, accurate motion of the robot (i.e., trajectory tracking, positioning, speed control) and enhanced collision avoidance (i.e., safe navigation around obstacles and humans). The estimation algorithm integrates data from LiDAR, RADAR, and IMU sensors using an Extended Kalman Filter (EKF) to overcome limitations such as LiDAR blind spots and IMU drift. The output of this algorithm is used as input to the MPC, which embeds an obstacle avoidance function directly into its cost function to enable proactive and adaptive collision avoidance. The complete framework is validated through both Gazebo-based simulations and real-world experiments in free-space navigation and navigation through obstacle-rich industrial environments. The results demonstrate that the robot can track reference trajectories while safely avoiding static and dynamic obstacles, including those located within sensor blind zones. The robot consistently maintains safe distances without unnecessary stops, achieving a balance between safety and operational efficiency. The framework offers a cost-effective alternative to high-resolution 360-degree perception systems and is well-suited for deployment in dynamic industrial workspaces.