Development
Follow the ongoing development of MMT Sentinel, our first marine safety prototype, including hardware testing, software progress, calibration work, and vessel field evaluations.
July 10, 2026
The first development session focused on establishing serial communication with the ultra-wideband boards, confirming AT-command access, and configuring four devices as vessel reference units and one as a crew tag.
July 23, 2026
Repeated measurements were collected across multiple known distances for each prototype device. The results are being used to calculate board-specific corrections and improve consistency before live vessel testing.
August 12–13, 2026
The prototype's ranging and localization approach was tested for the first time aboard an approximately 40-foot fishing vessel. Testing covered multiple areas of the vessel and a range of operating conditions, confirming that the core tracking concept could function in a real marine environment. This test was focused on feasibility and system behavior rather than precise absolute-accuracy measurement, and it provided the foundation for more rigorous controlled validation.
Early September 2026
Development turned to the real-time localization engine, along with supporting simulation, visualization, and logging tools. This work included preparing a controlled test environment with known reference positions so that system estimates could be compared directly against ground truth ahead of the next validation session.
September 8, 2026
A structured indoor validation session evaluated static positioning, repeatability, dynamic movement, edge behavior, vertical movement, disconnect handling, and fall-like motion, logging more than 15,000 localization cycles over an approximately 70-minute formal session. The system demonstrated strong repeatability and correctly handled degraded conditions. The session also isolated a systematic ranging bias as the primary factor limiting absolute positioning accuracy, directly informing the next calibration phase.
September 11, 2026
Following the September 8 findings, the calibration approach was redesigned to correct the identified ranging bias in software, then confirmed against fresh physical data that was not used to build the correction itself. Held-out validation produced approximately 16 cm mean 3D localization error and approximately 11 cm median error, with roughly 86% of readings within 25 cm. Performance was strongest within the primary monitored area, while extreme positions outside the core anchor geometry remain an area for continued refinement.
Current Development Phase
With the localization and calibration pipeline validated at the current prototype stage, development is shifting toward a more robust wired communications architecture, including Ethernet-based connectivity, designed for vessel-scale deployment. This work will be followed by additional marine testing to evaluate performance aboard larger and more varied vessels.
Current research is focused on wired network architecture for vessel-scale deployment, extending positioning coverage to below-deck and enclosed areas, independent wireless verification, false-alert reduction, and practical crew-device design.
Testing has progressed from controlled bench and ranging measurements through an initial vessel field test and a formal controlled localization validation session. Current work is confirming calibration improvements and preparing for expanded vessel-based trials as the network architecture evolves. Each stage is documented so that hardware behavior and software changes can be compared over time.