Public releases show the Office of Naval Research (ONR) funding at least two efforts aimed at improving robot hand dexterity and finger-level control: one focused on an underwater robotic gripper for explosive ordnance disposal and another on multi-fingered manipulation for shipboard maintenance. Separately, the National Science Foundation has announced up to $52 million for a new dexterity-focused Engineering Research Center. Taken together, these announcements point to growing federal interest in more capable robotic manipulation, though they do not describe a single unified program.
What is verified so far
The clearest technical evidence comes from RE2, LLC, a subsidiary of robotics firm Sarcos. The company reported achieving a technical milestone with its STARFISH gripper, a tactile-feedback robotic hand designed for mine countermeasures and explosive ordnance disposal. STARFISH uses three tactile-sensing fingers and demonstrated fine grip capabilities, including a tweezers-like grasp, during lab testing funded by the Office of Naval Research. Academic collaborators on the project include UCLA’s Veronica Santos and researchers at the University of Washington, both contributing expertise in tactile sensing and control.
The STARFISH milestone matters because explosive ordnance disposal and mine countermeasures are widely understood to be hazardous. In its announcement, RE2 positions the tactile-feedback gripper as a way to support tasks that can put divers and sailors at risk in low-visibility underwater environments. A robotic gripper that can sense pressure and adjust its hold in real time could help reduce human exposure, though the release does not claim operational replacement. The three-finger design is simpler than a five-fingered human hand, and the company’s lab demonstrations of delicate grasps move the concept from theory toward a tool that could eventually be mounted on remotely operated vehicles.
A separate and more recent program targets a different problem: keeping ships running. The USC Viterbi School announced that computer scientist Erdem Biyik received a three-year, $750,000 award through the ONR Young Investigator Program. His project aims to improve multi-fingered robot manipulation by integrating human feedback into real-time learning algorithms. The practical target is robots that can work in tight shipboard spaces, handling maintenance tasks that currently require trained technicians to squeeze into confined compartments and manually adjust valves, cables, and connectors.
Biyik’s approach differs from the STARFISH program in a telling way. Rather than designing a specialized gripper for one mission, the USC computer science effort focuses on the software layer, teaching a multi-fingered robot hand to learn from human demonstrations and corrections as it works. In this framework, sailors or technicians could guide a robot through a complex manipulation and then refine its behavior with feedback when it makes mistakes. Over time, the system would build a library of skills that can be reused across many tasks instead of being locked into a single grip pattern. That flexibility is what the Navy needs for general shipboard use, where a robot might tighten a valve one minute and thread a cable the next.
The broader research landscape reinforces this emphasis on dexterity. The National Science Foundation announced four new Engineering Research Centers, including one focused on dexterous manipulation, as part of a larger multi-center initiative to advance engineering capabilities. Within that group, the NSF selected a new Engineering Research Center called HAND, housed at Northwestern University, to receive up to $52 million for improving robot dexterity. According to Northwestern’s announcement, the HAND center will unite researchers from multiple institutions to develop robots that can manipulate objects with far greater finesse than current industrial systems.
While the HAND center is not a Navy program, its mission overlaps with ONR’s interests. Techniques for sensing contact forces, controlling many joints simultaneously, and planning complex grasps are relevant whether the robot is on a factory floor, in a hospital, or inside a ship’s engine room. The NSF investment therefore provides a parallel stream of basic research that could eventually feed into more applied military projects, even if no formal collaboration has been announced.
What remains uncertain
Several gaps in the public record prevent a full picture of how these programs fit together. No official statement from the Office of Naval Research describes an integration plan between the STARFISH gripper program and Biyik’s software-focused project at USC. The two efforts are funded by the same agency but appear to operate independently, with different technical goals and timelines. Whether ONR intends to merge their outputs into a single platform, or views them as parallel experiments, is not documented in available institutional releases.
The STARFISH milestone announcement, while detailed on the gripper’s capabilities, did not include performance benchmarks that would allow independent comparison with existing military or commercial robotic hands. How much force the fingers can exert, how quickly they adapt grip strength, and how they perform in actual ocean conditions rather than lab tanks all remain unaddressed in the public record. The academic collaborators at UCLA and the University of Washington have not released joint research papers tied to the STARFISH program that would provide peer-reviewed data, at least not in the sources cited here.
Biyik’s ONR Young Investigator award is early-stage by design. The three-year timeline and $750,000 budget are typical for exploratory research rather than deployment-ready engineering. No field trial data exists for the human-feedback learning system, and the USC announcement describes aims rather than results. There is no public indication yet of how the algorithms will handle noisy feedback from non-expert users, or how much training time a robot will need before it can reliably perform maintenance tasks without close supervision.
The broader funding context also carries uncertainty. The HAND center’s up-to-$52-million figure represents a ceiling, not a guaranteed allocation over its lifetime. NSF Engineering Research Centers typically receive funding in phases, contingent on meeting research milestones and demonstrating broader impacts. No public breakdown shows how much of HAND’s work will address Navy-specific applications versus civilian manufacturing, medical devices, or other sectors. Similarly, the NSF announcement of four new centers describes thematic areas, not detailed project lists, leaving open how much of the dexterity research will translate into ruggedized hardware suitable for harsh maritime environments.
How to read the evidence
The strongest evidence in this story comes from three types of primary documents: a company press release from Sarcos/RE2 describing a completed lab milestone, institutional announcements from USC describing a funded research award, and federal communications describing NSF’s new Engineering Research Centers. All of them name specific people, dollar amounts, and technical directions. The Sarcos release anchors claims about STARFISH’s current capabilities, the USC Viterbi story and faculty listing confirm Biyik’s role and funding, and the NSF and Northwestern documents establish the scale and focus of the HAND center.
What they do not provide is independent validation. The STARFISH milestone was announced by the company building it, not by an outside evaluator or the Navy itself. As with many lab-stage demonstrations described in company releases, performance in controlled testing may not translate directly to operational conditions. Salt corrosion, pressure changes, communication latency with remote operators, and the unpredictability of real ordnance all present challenges that a lab test cannot replicate. Readers should weigh the STARFISH achievement as a necessary step, not a finished product.
The USC project sits even earlier on the path from lab to fleet. The Young Investigator Program is designed to cultivate promising ideas rather than fieldable systems. Biyik’s work, as described in the computer science department and school-level materials, is fundamentally about algorithms: how to represent human feedback, how to update a robot’s control policy in real time, and how to keep the system stable as it learns. Those are important questions, but they precede issues like hardening hardware against shock and vibration or integrating with Navy logistics and safety protocols.
The NSF-backed HAND center occupies a middle ground between these two Navy-linked efforts. Its mission is broader, spanning multiple application domains, and its funding is larger and more sustained. That makes it a likely source of foundational advances in tactile sensing, compliant actuation, and learning-based control. Yet the center’s public materials emphasize long-term scientific and societal impacts rather than near-term military deployment. Any Navy benefit will depend on future technology transfer, follow-on contracts, and the willingness of defense agencies to adapt civilian research to their specific needs.
For now, the evidence supports a cautious reading. The Navy, through ONR, is investing in both specialized hardware like STARFISH and adaptable software for multi-fingered hands. NSF, through HAND and other centers, is underwriting the basic science that could make those systems more capable. None of the available sources claim that robots are ready to replace human divers or shipboard technicians at scale. Instead, they point to a converging research agenda: giving machines fingers and control systems that can finally handle the messy, uncertain, and dangerous manual work that has long resisted automation.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
