NASA’s robot science team could investigate a new target while continuing its original task

NASA’s ASTRA team tested a small robotic expedition in Blacksburg, Virginia, in July 2026. When its drone identified a new area of interest, the fleet could call another robot into the investigation while preserving the original objective. NASA reported the terrestrial demonstration on 15 September.[1]

Planning an investigation rather than a single journey

A scientific expedition contains decisions that a route planner alone cannot settle. Reaching a location is one objective; deciding which location deserves a closer look is another. With different instruments available, the useful question becomes which combination of observations can answer the team’s question most effectively.

The fleet considered robot availability, travel, scientific priorities and acceptable risk as it assigned work. That combination allowed it to respond to a new observation during an ongoing plan.[1]

Consider the general logic of an aerial survey followed by a sample. The wide view can identify a pattern, while a closer measurement can help explain it. The value comes from connecting the observations. Adding more robots becomes scientifically useful when their measurements contribute to a common investigation, rather than simply increasing the number of places visited.

NASA is developing several pieces of autonomous science

A complementary orbital example is MEDOS, the Module for Event Driven Operations on Spacecraft. Developed by Aurora Engineering and demonstrated on the Magnetospheric Multiscale mission, it combines telemetry into physical quantities and estimates which event best explains them. NASA reports that it detected radiation-belt passages missed by a ground prediction and raised a flag to disable high voltage during those periods.[2]

This is an event-recognition problem: a spacecraft needs to recognise that its actual environment has changed. ASTRA adds a different question at the expedition level: given what has just been observed, how should the available team continue its work? Connecting recognition with the choice of a next observation is the conceptual step toward more responsive scientific missions.

Starling’s Distributed Spacecraft Autonomy experiment provides the group-coordination example. Four small satellites used dual-band GPS receivers to study electron density in Earth’s ionosphere. Their changing positions altered which observations would be useful. The satellites shared information across their radio network and coordinated the signals they would prioritise next.[3]

The scientific advantage of multiple viewpoints is that a changing phenomenon can be sampled across space as well as time. A single instrument moving through it may struggle to distinguish a change in location from a change in the phenomenon itself. Coordinated observations can give researchers more of the information needed to separate those possibilities.[3]

The next useful capability is a connected observing strategy

MEDOS shows how immediate sensor evidence can change a spacecraft’s response. Starling shows measurements being selected cooperatively across a group. The ASTRA field demonstration brings that theme to a mixed team whose members can see, map and sample in different ways.[2][3]

Taken together, these demonstrations point toward a more productive relationship between scientists and remote machines. People define the investigation and its priorities; local systems can choose the next useful measurement while the opportunity is present. The goal is a richer scientific sequence: an observation prompts a question, the right instrument investigates, and the resulting evidence informs what happens next.

Report an error or suggest a correction