By MARDE Inc. · Published
Updated
Prepared by MARDE Inc. as an editorial synthesis of the linked primary sources and a proposed evaluation framework. MARDE’s contribution is the claim-to-evidence questions and hypothetical exercises, not original experimental or clinical findings.
Key answer
Final access is the movement from an aerial delivery point to an agreed receiving location. A useful test specifies the route, payload, operator view, communication limits and recovery plan. MARDE’s current V1 scope is outdoor teleoperation; indoor entry and stair climbing are deferred.
Coordinates are not an access plan
A vehicle can arrive at the right location while its payload remains inaccessible. A fence, threshold, stair, closed door or uncertain receiving point can separate geographic arrival from useful delivery. The final-access problem is the set of tasks required to move from that arrival point to a place where an authorized person can receive or use the payload.
For MARDE, Ground is the proposed component for investigating that continuation. It is intended to be teleoperated first, with a human directing the approach. The October 2026 V1 scope is outdoor teleoperation and supply delivery; indoor entry and stair climbing are deferred. The first useful step is to break broad access language into testable tasks.
Describe the obstacle precisely
A requirement such as handles stairs is incomplete. An evaluation needs dimensions, surface conditions, landing space, payload configuration and operator visibility. A robot that moves successfully on one arrangement may fail on a different height, edge shape or surface. Similar care applies to doorways and narrow passages: clearance alone does not establish that the operator can navigate safely.
Agencies can help by describing recurring barriers in de-identified operational scenarios. The purpose is not to collect patient locations or sensitive building plans. It is to identify representative tasks, exclusion conditions and receiving arrangements. A scenario should state where the aircraft could stop and which additional movement would be necessary.
Observe the operator as well as the machine
Teleoperation makes operator information part of the capability. Cameras may have blind spots, lighting may change and a payload may obscure a view. Testing should record what the operator could see, how commands were issued and when uncertainty required a stop. Otherwise a mechanical demonstration can conceal an unrealistic level of support from nearby observers.
An independent observer may help record events during a controlled exercise, but their assistance must be documented. If someone physically repositions the robot, supplies directions or removes an obstacle, that intervention changes the meaning of the result. Operator proficiency and practice time should also be recorded before comparing configurations.
Treat communications as an environmental constraint
Ground movement can change the communication path after an aerial handoff. A mission that begins with a clear connection may enter a more difficult radio environment. NIST has published a test method addressing response-robot performance under attenuated radio conditions. That is a useful reference for making communication limitations part of a test rather than treating them as an unexpected exception.
For the proposed MARDE workflow, questions include how the operator recognizes stale information, how movement stops and what recovery process is available. This article does not define a validated failsafe design. Those behaviors need engineering decisions and repeatable tests appropriate to the eventual hardware and control system.
Measure task completion and recovery
A useful access test records whether the payload reaches the agreed receiving point, how long that takes and whether interventions occur. Recovery matters too: what happens when the route is blocked, the robot stops or a mission is abandoned? A task that succeeds only when nothing unexpected occurs provides limited evidence for an emergency-response use case.
NIST ground-robot test methods offer a reference for structured evaluation. MARDE would need to select tasks relevant to its proposed workflow and identify which conditions were actually tested. Using a public method as inspiration does not imply certification, endorsement or successful completion of a formal evaluation.
Turn broad access claims into a task record
A task record can describe the starting point, receiving point, route conditions, payload and operator preparation in ordinary language. Add photographs or drawings of the controlled layout when publishing an eventual result, while avoiding sensitive operational locations. Record the configuration identifier and which parts changed since the preceding exercise. These details make it possible to compare attempts and prevent a later reader from assuming that a result applies to every building or terrain type. The record should also state what was intentionally excluded, including occupied rooms, unknown doors or conditions outside the test plan.
A separate interpretation can then explain what the team believes the result suggests. For example, a recurring stop at a threshold might motivate a mechanical change, a camera change or a revised exclusion. The observation alone may not identify which change is best. Treat the explanation as a hypothesis and define another exercise that can distinguish the possibilities. This keeps access engineering connected to evidence instead of letting a successful or unsuccessful video drive a broad product claim. MARDE has not completed the task record described here; it is a proposed reporting approach for future work.
A hypothetical corridor-to-receiver test
Imagine a controlled training area with a marked aerial handoff point, a ground route and a receiving station. Use an inert payload and keep people outside the vehicle path. Before the attempt, document route width, surface, obstacles, lighting and operator location. The target is receipt at the marked station, not entry into an unknown occupied space. This creates a bounded access question that can be repeated.
A first configuration might test a clear route, followed by a configuration with a documented threshold or turn. The operator should receive only the information intended to be available in the proposed system. If someone standing near the robot provides extra directions, record the help. Repeatability depends on controlling these details; otherwise two apparently similar attempts can rely on very different levels of human assistance.
Record task completion, elapsed time, stops, interventions and payload condition. Also record the reason for an unsuccessful attempt: insufficient clearance, loss of observation, inability to move or an operator decision to stop. These categories help determine what to change next. A single completion percentage without its task definition tells a practitioner little about relevance to another scene.
Recovery is part of the access requirement
An exercise should include a blocked receiving point. The operator may need to stop, return or leave the payload at an agreed safe location. Which response is appropriate depends on the use case and system design, so the choice belongs in the protocol. It should not be improvised during a demonstration and later described as a reliable recovery capability.
A useful recovery record identifies who notices the problem, what information they have and who retrieves the system if movement cannot continue. It also identifies whether recovery occupies people needed elsewhere. This helps agencies evaluate staffing consequences instead of considering only vehicle dimensions. A robot that is easy to send but difficult to recover may impose an unacceptable operational burden.
After testing, compare the result with the original access problem. If the receiving station was already reachable by a simpler arrangement, the exercise may establish engineering progress without establishing operational value. If the task addresses a genuine barrier, the next exercise should investigate the remaining limitations. MARDE should communicate both possibilities as part of an honest development record.
Connect access to the handoff
The final receiving point should be defined before testing. It might be a responder-controlled area or another designated location, depending on the intended use case. Reaching a patient is not automatically an appropriate or authorized objective. Clinical responsibility, equipment handling and responder access must remain explicit.
An integrated evaluation should follow the payload from Air through Ground to receipt. It should record its condition after transport and whether a receiver understands the handoff. MARDE's proposed advantage is coordination of these steps; proving each component separately would leave the central integration question unanswered.
MARDE remains at concept/pre-prototype stage. Its proposed outdoor supply-delivery and scene-communication system has not been built or validated; the exercises here are proposals, not completed tests.