Communications & Networking for Challenging Robotic Environments

Robotics are generally of little use without communications. Collaboration is impossible, and ultimately even a fully autonomous robot must relay data back to human operators to be of value. We work on developing systems that bridge the gap between the autonomy applications and the various physical links that are present in the marine environment (acoustic, satellite, radio, etc.).

Underwater acoustic communications

Subsea digital telemetry is widely performed using acoustic carriers due to the short propagation distance of electromagnetic waves in the ocean. However, acoustic links suffer from low bandwidth and high error rates due to boundary reflections, complex refraction, absorption, and non-negligible Doppler effects. We research and develop software systems that can provide reliable and usable end-to-end data connectivity (typically amongst autonomous vehicles and their human operators) over a variety of existing commercial and research modems. The design of the Goby project allows rapid portability between different physical links, regardless of the interface design or special features (e.g. navigation features such as USBL, LBL) afforded by the hardware.


We participated in the ICEX20 and ICEX16 Arctic exercises in the Beaufort Sea north of Deadhorse, Alaska as part of the team led by the MIT Laboratory for Autonomous Marine Sensing Systems. We ran a Bluefin-21 AUV under the ice to collect acoustic data to understand the changing Arctic environment due to global warming effects. GobySoft led the effort to produce a tracking and communications system in collaboration with components from WHOI, NUWC/Newport, APL/UW, Bluefin and MIT. The 2016 and 2020 posters give a detailed overview of the system, and this paper covers the design of the 2020 system, which relied heavily on the NETSIM and Goby3 projects for success.

Distributed Agile Submarine Hunting (DASH): Multi-hop communications network for marine vehicle and mooring deployments

This DARPA project focused on using deep ocean sonar nodes for rapid detection of submarines overhead. Communications from the deep ocean nodes back to a shore-based operator is complex and highly limited by the throughput afforded by the various links at sea. Using the Goby project, we developed a reliable and effective full-duplex multi-hop network over heterogenous links from the shore operator to the deep sea node (UUV).

DARPA Robotics Challenge (DRC)

The DARPA Robotics Challenge was an international competition focused on developing humanoid robots and autonomy software to intervene in disasters where humans can no longer safely go. The competition, which was motivated by the Fukushima nuclear disaster, also simulates difficult (i.e. low throughput, high latency) networking conditions, such as those that might be found such a tragedy. Using components from the Goby project along with new software, we developed the networking software used by our team in the 2013 Virtual Robotics Challenge (3rd place) and DRC Trials (Homestead, FL; 4th place) as well as the upcoming 2015 DRC Finals.