Collaborative autonomy for naval stand-off mine countermeasure - PDF document

UDT 2020 Collaborative autonomy for stand-off MCM operations Unmanned, Remotely Piloted & Autonomous Systems Collaborative autonomy for naval stand-off mine countermeasure operations R. van Vossen 1 , A.L.D. Beckers 2 and J.J.M. van de Sande 3 1 Senior Scientist, TNO, The Hague, The Netherlands, robbert.vanvossen@tno.nl 2 Senior Consultant, TNO, The Hague, The Netherlands 3 Scientist, TNO, The Hague, The Netherlands Abstract — To increase the efficiency of stand-off mine countermeasure (MCM) operations, multiple heterogeneous systems are to be used in parallel. The implementation of an efficient stand-off MCM concept therefore requires that systems are able to cooperate, together with a high level of autonomy to aid the task execution of individual systems. The technology for adaptive search in a complex terrain and autonomous contact reacquisition by multiple systems is developed and demonstrated in this paper, covering self-awareness for sensor and communications management. 1 Introduction In order to be able to investigate collaborative autonomy a testbed has been realised, comprising: Many Navies are moving towards implementing stand-off Multiple autonomous underwater vehicles (AUVs) - mine countermeasures (MCM) concepts in which with different sensor payloads, including side scan heterogeneous collaborative unmanned systems are used sonar (SSS), multi-beam echo sounder (MBES), for conducting the detect-to-engage tasks. Potential optical camera, a software-defined underwater benefits are that (i) new sensor and system concepts can communications modem, and aided inertial be introduced to improve the effectiveness of MCM navigation system; operations, and (ii) there is an opportunity to increase the Relay nodes for underwater communications. - efficiency of MCM operations by using a large number of The individual systems have an open architecture, unmanned systems in parallel. enabling the implementation of an autonomy framework To increase the efficiency of MCM operations, to develop not only functionality to autonomously or scalable solutions are sought, in which multiple systems adaptively execute MCM tasks, but also to pass relevant can be deployed in parallel. In such a scalable stand-off information to the coordinating mechanism and to other MCM concept, individual systems are tailored to conduct systems in the collaboration. Coordination information specific MCM tasks, and cooperation between systems is typically consists of command and control information: needed to complete the entire detect-to engage chain. task initiation, task performance and task completion. The realization of a successful stand-off MCM Collaboration information is about mission data to enable concept with a group of collaborating systems requires other systems to execute their tasks such as mine like echo’s, contacts and contextual information. that: (i) stand-off MCM systems are able to exchange relevant information; 3 Results (ii) stand-off MCM systems can be tasked to conduct the detect-to-engage tasks as a team; Within this framework functionality has been developed (iii) stand-off MCM systems are able to adapt themselves for adaptive mine search and for adaptive multi-vehicle to deal with uncertain/unknown environmental contact reacquisition. conditions and system performance limitations. This paper focuses on collaborative autonomy to aid the 3.1 Adaptive mine search execution of mine-hunting tasks, addressing challenges related to limitations in communication, navigation An adaptive mine search survey is presented in Figure 1. accuracy, and variations in sensor performance. In an The area has a complex bathymetry with sand dunes. As accompanying paper, Van Velsen et al. [1] present a a consequence, it is difficult to realize full coverage with framework for collaborative and adaptive mission a SSS that operates at small grazing angles. In Figure 2 a management for naval stand-off mine hunting operations. SSS image of the concerning area is depicted in which Mission management involves the planning and sand dunes block the view of the sonar. The mission evaluation for the different stand-off MCM systems. example shows that SSS data is interpreted in-mission to obtain coverage information, and subsequently plan 2 Technical approach additional passes using a different orientation. Because full coverage is not realized using two different orientations, the vehicle decides to use a different sensor,

UDT 2020 Unmanned, Remotely Piloted & Autonomous Systems Collaborative autonomy for stand-off MCM operations an MBES, that operates at steeper grazing angles to fill the remaining gaps in the coverage. Fig. 1. Adaptive mine search in an area with sand dunes. In the first pass, shadows of sand dunes result in gaps in the coverage (top left). The gaps are automatically detected and an adaptive new plan is scheduled along a different orientation (top right). Having completed the second pass, there still remain gaps in the coverage (bottom left). A final survey is scheduled to fill the gaps using an MBES rather than a SSS (bottom right). Fig. 2. Side scan sonar image recorded in the mission area. On the starboard side of the AUV sound of the sonar is blocked by a sand dune, leaving a shadow in the image. The area where the shadow resides is considered as not covered. In mission planning, decisions have to be made on the tasking of individual reacquisition and identification 3.2 Adaptive multi-vehicle contact reacquisition systems in comms-limited environments [1]. When the systems are tasked, each system has to optimally plan and In stand-off naval MCM operations, contact reacquisition execute its reacquisition and identification task. It has been is a challenging task due to inherent navigation and shown and demonstrated in [2] that the use of an auxiliary localization uncertainties. Furthermore, sensors that are sensor to aid the navigation solution can greatly improve commonly used for the identification, optical cameras, the efficiency of contact reacquisition and optical usually have a limited swath width, especially in limited identification. underwater visibility conditions. As a consequence, To reduce the reacquisition time of a vehicle, contact reacquisition can become time consuming, landmarks are used to improve the navigation solution. especially when many contacts are to be reacquired. During the search phase landmarks are detected by the

UDT 2020 Unmanned, Remotely Piloted & Autonomous Systems Collaborative autonomy for stand-off MCM operations search vehicle and acoustically transmitted to the shows an example side scan image with the detected identification vehicle. The identification vehicle in this landmarks indicated by the green boxes. way builds up a map of landmarks in the area in which it will need to do reacquisition and identification. Figure 3 Fig. 3. Side scan sonar image recorded in the mission area. Detected landmarks are indicated by the green boxes. The landmarks are broadcast acoustically to other vehicles to use as reference navigation landmarks during reacquisition and identification. During the identification task, the identification vehicle own position uncertainty as well as that of the detected determines the fastest reacquisition plan for each contact contact. Figure 4 shows the result of two search and that must be reacquired. This is either a direct reacquisition identify missions in which detected contacts needed to be in which the contact is systematically covered by the reacquired and identified. In one mission the two-step identification sensor, or a two-step approach in which a approach was used (left); in the other mission the direct pre-reacquisition track is executed first before approach was enforced (right). Depending on the position identification. The pre-reacquisition track is used to find uncertainties, the two-step approach reduces the back objects detected by the search vehicle, and use those identification time significantly. In Figure 5 one of the as a reference to correct its own position with. This identification results for the two-step reacquisition is requires that the identification vehicle is also equipped depicted: a stitched video recording of one of the with an auxiliary sensor to do object detection with. The identification tracks, together with the single camera frame two plans are computed taking into account the vehicle’s catching the dummy object. Fig. 4. Overview of two executed search and identify missions. The left mission shows the AUV tracks executed during a two-step search and identify mission where landmarks were used to improve the relative position accuracy between the vehicle and the contact to identify. The cyan line indicates the pre-reacquisition track. In the right mission a direct identification is enforced in which the contacts are systematically covered with the identification sensor, based on the vehicle’s and contact’s location uncertainties. The difference in traversed distance during the identification legs is clearly observable.