Where robots can help
Useful non-weapon roles include moving supplies, inspecting ships and facilities, carrying sensors into hazardous spaces, mapping unfamiliar terrain, and supporting search and rescue. The U.S. Army has evaluated autonomous vehicles for logistics resupply during a multinational exercise. A Navy shipyard and research center also developed and tested a climbing robot for submarine hull inspection. These are reported trials and prototypes; neither report establishes broad deployment across the services.
DARPA's completed Subterranean Challenge gives another concrete example. Physical robot teams mapped, navigated, and searched representative tunnel, urban, and cave courses. DARPA framed collapsed-mine, post-earthquake, and cave rescue as possible applications. That result shows what teams demonstrated in a structured competition, not that one system is ready for every underground mission.
A demonstration is a milestone, not a deployment
Ask what was tested, where, for how long, with which operators, and what happened when conditions changed. The Department of Defense technology-readiness scale distinguishes a prototype demonstrated in an operational environment from a completed system qualified under expected conditions, and from a system proven in successful mission operations. Marketing often compresses those steps into the word “proven.” A founder should preserve the distinction and describe evidence at the level actually reached.
What building one involves
Start with one operational job and the people around it: who deploys the robot, supervises it, interprets its data, recovers it, and maintains it. Then choose mobility, sensing, compute, power, communications, and an operator interface for that job. Software must connect perception, localization, planning, control, health monitoring, logging, and a safe way for an operator to pause or recover the system. Existing platforms can shorten mechanical development, but they do not remove integration work.
Field validation should progress from component checks and simulation to repeatable capability tests and representative exercises. Measure the whole system across terrain, visibility, endurance, communications loss, operator workload, recovery, and maintenance. NIST's response-robot program uses responder-defined requirements covering mobility, manipulation, sensors, energy, communications, human-machine interfaces, logistics, and safety, with repeatable performance data intended to support purchasing and deployment decisions.
Where Vesk fits
Simulation evidence for a bounded ROS 2 task.
Vesk can generate ROS 2 Jazzy code and test it in controlled simulation. That evidence is simulation-only: it does not operate physical hardware, certify safety, or provide assurance that a system is ready to deploy. Vesk cannot assist with selecting, designating, or engaging people, including with non-lethal weapons. Non-engagement work such as navigation, reconnaissance, logistics, inspection, safety, and search and rescue can still be considered on its own merits.