Plume Front Tracking in Unknown Environments by Estimation and Control

Oil spill at the Gulf of Mexico a few years ago sets great hazardous to environments. The runtime monitoring and tracking of plume pose challenges to research community. Robot-based approaches have been proposed in previous work to solve this problem in environments with known physical parameters and measurable currents. However, practical implementation of this type of control laws may require additional sensors and the sensing error may significantly impact the convergence. This paper extends previous work by overcoming the challenge to establish a new control law by integrating interactive estimation and control in a unified loop. With a deliberate design, the parameters can be estimated online and the control can be achieved at the same time with provable convergence of the overall system, despite of the interplay of the two parts. An auxiliary system is constructed for efficient estimation of unknown parameters and set projection is incorporated to further improve the transient performance, making the system work well in both slow and fast time-varying environments. Theories of convergence, stability, and transient state are presented to guarantee the performance of plume front tracking. Validation experiments verify the theoretical results and substantiate the efficacy of the proposed scheme for plume tracking in unknown environments.