Optimal process control in multiphase chemical reactors is challenging due to strong spatial inhomogeneities in fluid dynamics, phase distribution, substrate concentration, and micro-environments. In production-scale reactors, optical access is unavailable, and measurements from fixed Eulerian probes are often not representative of the global reactor state. As an alternative, Lagrangian devices travelling with the flow have been proposed as a possible approach for sensing and actuation. To infer transport processes from Lagrangian device measurements, their trajectories must be known with sufficient accuracy. However, such devices would not behave as ideal tracers but be subject to inertial effects that lead to velocities and transport patterns differing from those of the surrounding fluid. We consider simulated devices as inertial particles and use Maxey–Riley–Gatignol equations (MaRGE) to describe their trajectory. In this talk, we present a model-based tracking framework for Lagrangian devices based on MaRGE combined with synethetic measurements from the simulated devices’ IMU using filtering algorithms. A numerical solver for the three-dimensional MaRGE, including the history force, is obtained and MaRGE without history term is used as the prediction model within the filter. We assess tracking performance in two types of flow field, one given analytically, the other obtained by measurements in a lab-scale stirred tank reactor. The results demonstrate that the filter can successfully reconstruct the trajectory of simulated devices in both cases.