Near-real-time Water Quality Monitoring in the Raritan River Using Hybrid Vehicular-static Stations

News | Raritan River Initiatives | Water Quality

Near-real-time water-quality monitoring of different variables in the Raritan River is critical to protect the aquatic life and to prevent propagation of the potential pollution in the water. Using only static sensors attached to fixed monitoring stations with predefined configurations is not a real-time and efficient solution for data collection as the phenomenon of interest may occur sporadically and propagate spatially through the water bodies. We have designed a hybrid water quality monitoring system–which is capable of chasing the phenomenon of interest instead of waiting for the pollution to reach the fixed stations–including static nodes as well as both Remotely Operated Vehicles (ROVs) and Autonomous Underwater Vehicles (AUVs). The vehicles are equipped with multiple on-board sensors and are deployed in the Raritan River for gathering scientific data via collaborative monitoring.

Several members of the Cyber-Physical Systems Laboratory (CPS Lab)–in the Department of Electrical and Computer Engineering (ECE), School of Engineering (SoE)–were involved in this project including Professor Dario Pompili (CPS Lab director), Mehdi Rahmati (SoE senior Ph.D. student), Archana Arjula (SoE master student), Mohammad Nadeem (SoE undergraduate student), and Agam Modasiya (SoE undergraduate student). The experiments were conducted in the Raritan River in the Summer and Fall of 2019, with the support of Professor Richard Lathrop from the Department of Ecology, Evolution & Natural Resources.

In this mini-project, we were able to conduct adaptive sampling experiments in the Raritan River using a team of underwater/surface autonomous vehicles (BlueROVs). To achieve this goal, the vehicles explored the environment to learn the policy through experience and make actions by leveraging a Multi-Agent Reinforcement Learning (MARL) framework we developed, which is robust against different types of uncertainty. The AUVs search a region of interest on the surface and the static server optimally coordinates the movements of the AUVs [1]. Future work involves:

  • The development of a hierarchical approach to allow exploration of different depths of the Raritan River bypassing the communication constraints; and
  • The design of an efficient data-compression and data-transmission protocol such that the vehicles can be pre-configured to perform different types of missions on the Raritan River in order to decide quickly where to go, analyze the measured data (or unpredictable change in the pre-configured settings) locally, or report it to the control center using acoustic micro-modems and/or Radio Frequency (RF) signals, for further commands [2].

Our related papers, which were both presented at the 14th ACM International Conference on Underwater Networks and Systems (WUWNet), Atlanta, GA, Oct. 2019.

This work was supported by a 2019 Rutgers Raritan River Consortium (R3C) Mini-grant.  You can learn more about that program here.

For more information, contact Dr. Dario Pompili at or Mehdi Rahmati at