The Velocity Obstacle method produces both safe velocities and trajectories that avoid moving obstacles. Currently, the USV path planner only uses the velocity output in its calculations. A further investigation of safe trajectories could enhance obstacle avoidance.
Currently the human involvement in the process of accepting or rejecting the OA route can cause latency. This time delay could result in collisions as new threats appear but the user hasn’t yet released control. In the future, as the deliberative obstacle avoidance component proves more reliable, the communication between the navigator and path planner will be automated (with user oversight of course). The navigator would accept every new path, given that each path is checked for meeting certain safety requirements. This will give the path planner more time to plan more paths, correct any outdated routes caused by noisy radar data or ARPA contacts that have changed their heading, and give a more accurate response to current environmental threats.
SSC San Diego has plans to incorporate additional sensors for both the deliberative and reactive OA components. Additional sensors for the deliberative component will include an Automatic Identification System (AIS) receiver and more detailed bathymetric data. The AIS system receives position, speed and course data broadcasts from other marine vessels with compatible systems. This will provide very accurate data on at least some of the other contacts in the environment. The current nautical charts have only sporadic depth data. The addition of more detailed bathymetric data will complement the DNC data well.
Additional sensors for the reactive OA component will likely include millimeter wave (MMW) radar and ladar systems. SSC San Diego has conducted promising preliminary tests with a scanning ladar system and found that it did not report any return from the water’s surface and only detected obstacles on the water. technologies
An autonomous navigation and OA architecture for USVs has been developed by SSC San Diego that supports both deliberative and reactive OA and accommodates a variety of sensors with both long and short fields of view. This autonomy architecture has been developed under the JAUS framework for future interoperability and portability. Many of the software components of that architecture have been implemented and are being actively tested on the USV.
A quick and capable path planner has been developed to divert the route of a USV away from threats of stationary and moving obstacles in the far-field. On average, the path planner can successfully plan around potential collisions with more than 10 moving obstacles in less than 3 seconds. SSC San Diego will continue to make this path planner more robust to handle a variety of situations and obstacles in the aim of autonomy. The reactive OA software is being transitioned from existing ground vehicle programs and will be tested on the water in the near future.
The combination of a flexible and portable architecture, in depth sensor characterization and analysis, robust planning algorithms, and many hours of on-the-water testing will produce a reliable, transition-ready autonomous navigation and OA capability for use on many USVs throughout the Joint Forces and Department of Homeland Security.