The U.S. Navy is developing new software programs to make its ocean-going robots more independent.
By the end of the decade, the Navy plans to deploy squadrons of unmanned underwater robots to survey the ocean. But there are a lot of challenges operating underwater, and the robots will require a great deal of autonomy to carry out search and mapping missions.
That’s the goal of the Adaptive Networks for Threat and Intrusion Detection or Termination (ANTIDOTE) program. Funded by the Office of Naval Research (ONR), ANTIDOTE’s team of scientists from the Massachusetts Institute of Technology and the University of Southern California is developing software-based methods for large teams of robots to perform more sophisticated missions autonomously in dynamic, time-critical environments and with limited communications.
A major part of the program focuses on autonomous planning and replanning methods, said Marc Steinberg, ANTIDOTE’s program officer at ONR.
The underlying theory behind ANTIDOTE was for persistent surveillance of dynamic environments, regardless of the vehicle type carrying out the mission, Steinberg said. For example, some successful simulation experiments have been conducted with unmanned air systems.
Undersea vehicles have very limited communications compared with systems that operate above ground, Steinberg said. This drives a need for autonomy because the robot subs can’t rely on a human operator.
Additionally, there are unique challenges in navigation, mobility and sensing underwater. For example, the undersea glider robots in ANTIDOTE's experiments use changes in buoyancy for propulsion, rather than an active device such as a propeller. This enables them to have an extended endurance, but it also requires that gliders move up and down in depth in a saw-tooth-like pattern, which has a big impact on how to do autonomous planning to maximize the value of the scientific data being collected.
“The sea experiments were a great way to examine how some promising theoretical results would work in a real-world situation of practical value to scientists,” Steinberg said. Prior theoretical work had looked at how autonomous vehicles can best perform persistent surveillance in a dynamic environment.
In the sea tests, the new software was used to generate paths for the underwater gliders to collect oceanographic data. The method takes into account both user priorities and ocean currents in determining these paths. The experiments, in southern California and in Monterey Bay, Calif., involved a glider using this new capability and a reference glider that followed a more traditional fixed path.
Results of the experiment showed that the vehicle using the new method executed two to four times as many sampling profiles in areas of high interest when compared against the unmodified reference glider, while maintaining an overall time constraint for the completion of each circuit of the path, ANTIDOTE researchers said in a statement.
Overall, the results validate that the theoretical results can be of value in solving real-world surveillance problems with autonomous systems, Steinberg said.
The ANTIDOTE program is near the end of its third year. After that, Steinberg said that it is up to ONR leadership to decide whether to fund it for an additional two years.
“As a fundamental research program, the main products are new theory and methods," he said. "Some of these are being implemented in software and will be available to other researchers via open architectures for robots.”