The research, published in Frontiers in Aerospace Engineering by Adriana Andreeva-Mori of JAXA’s Aviation Integration Innovation Hub, tackles a problem that has become urgent as advanced air mobility concepts mature. VFR flights are notoriously difficult to forecast. A VFR flight plan contains little spatial or temporal detail, and in countries like the United States one is often not even required. Pilots choose their own routes, altitudes and speeds, deviating freely for weather, scenery or traffic. That flexibility is precisely what makes VFR unsuited to congested airspace, and it is why VFR has traditionally been confined to quiet skies, low altitudes and small aircraft. IFR, by contrast, is highly structured: detailed flight plans allow strategic planning and dense operations, but at the cost of pilot flexibility and expensive equipment requirements.

Andreeva-Mori’s insight is that VFR missions are not as random as they appear. Most low-altitude VFR flights flown by rotorcraft are mission-driven: disaster reconnaissance, aerial imaging, firefighting, medical evacuation, media coverage. In disaster reconnaissance, the mission studied here, pilots typically fly along a landmark such as a highway, river or coastline, tracking it visually while a crew in the back of the aircraft assesses damage and searches for evacuees. Interviews with experienced disaster-response pilots revealed that they maintain a roughly constant view angle to the landmark, about 37 degrees, keep a constant altitude above ground to give the rear crew a stable platform, and fly at a target ground speed of about 65 knots with deceleration at turns. Those parameters allow a reconnaissance trajectory to be modeled as a collection of line segments and arcs.

To turn that model into an airspace management tool, the study proposes geofencing each VFR flight within an operation volume, similar in spirit to how small drone operations are handled under unmanned traffic management concepts. Each volume is a four-dimensional shape, defined laterally, vertically and temporally, that encloses the predicted trajectory plus a safety buffer. The buffers account for terrain, landmark curvature and possible in-flight altitude changes. Before departure, predicted trajectories and volumes are loaded into a portable situation-awareness tool, and deviation alerts warn the pilot when the aircraft strays from its assigned volume. The goal is segregation: keeping low-altitude VFR flights safely apart from drones, eVTOL vehicles and other traffic without forcing pilots into full IFR rigidity.

Testing the concept required real flying, not simulation. Two flight-test campaigns were conducted with JAXA’s BK117C-2 research helicopter, a medium-sized twin-engine machine widely used for disaster response. The first campaign comprised nine flights between late January and mid-February 2022; the second comprised four flights in August 2022. Each flight lasted about two hours, taking off from Chofu Aerodrome in western Tokyo, with 70 to 90 minutes available for reconnaissance-modeled segments after the 15-to-20-minute transfers. Three test areas were chosen: a mountainous region near Mount Fuji with a highway landmark and severe elevation fluctuations, a gentler mountainous area on the Boso Peninsula with a longer highway segment, and the Miura Peninsula, where a single flight followed a highway, then a river, then a coastline.

The onboard technology was D-PAS, a fully portable system developed as part of JAXA’s D-NET disaster response research and already deployed on all fire service helicopters in Japan. It consists of a satellite transmission unit, a digital antenna attached to the windshield by suction cups, and a touch-screen display with a built-in map. The display shows the discretized predicted trajectory as a series of dots, the associated operation volume in purple, and the aircraft’s actual position as a red arrow. Telemetry flows to the ground over an Iridium satellite link, so researchers could track conformance in real time. Two alert philosophies were compared: in the first campaign, alerts were issued when the helicopter deviated from the predicted four-dimensional position; in the second, alerts fired when the aircraft left its assigned volume.

The temporal results were among the most encouraging. When no time advisories were given, pilots consistently completed segments earlier than planned, flying slightly faster than the 65-knot reference speed, apparently to improve the efficiency of information gathering. When deviation advisories were provided during the first campaign, temporal predictability improved dramatically, with all flights completing within one minute of the predicted 11-minute segment for the Mount Fuji area. Notably, in the second campaign the veteran pilot maintained temporal conformance even without advisories, once familiar with the reference speed. Pilots reported no significant workload increase when conformance was managed through alerts rather than strict fly-over times at waypoints, and they recommended a temporal buffer corresponding to speed differences of about 5 knots. Because early completion is more common than delay, the buffer should be asymmetric, larger on the early side. One caveat emerged: higher altitudes led to faster flights, since pilots perceive the landmark differently and their field of vision expands, so temporal buffers should grow when altitude changes are likely.

Lateral conformance proved robust to the trajectory model’s central assumption. Across 24 flights over the Mount Fuji highway segment, trajectories clustered tightly around the prediction when the target altitude matched the planned 1,000 feet above ground level, validating the constant view angle of 37 degrees. Pilot differences did appear: one pilot flew about 200 meters closer to the landmark than predicted, corresponding to a steeper view angle of roughly 40 degrees, a difference traced to seat height and posture. Gaussian process regression applied to the first campaign’s data, chosen for its suitability to small datasets, produced a 95 percent confidence interval spanning deviations from about minus 220 to 290 meters when both pilots were considered, supporting a 300-meter lateral buffer. In the second campaign, volumes around the Mount Fuji segment used a 300-meter buffer while the other areas used 500 meters. Even when unplanned altitude increases pushed the nominal trajectory outward, the pilot corrected in response to advisories and stayed within the volume. The largest lateral deviation occurred when unexpected traffic approached head-on, forcing a brief avoidance maneuver, and pilots commented that reserved volumes could prevent exactly such interference while reducing cockpit workload.

The vertical dimension was the hardest. Pilots try to hold constant altitude above ground, but in mountainous terrain the ground itself rises and falls, and over three hills along the Mount Fuji segment the pilot followed the GPS altitude rather than constant AGL, with deviations reaching 500 feet. Even in gentler areas, spikes of 200 to 460 feet appeared. The study concludes that a 100-foot vertical buffer is sufficient only for flat terrain under the current simplistic trajectory model, and that at least 200 feet below and above the target altitude is needed even where elevation fluctuations are modest. Because crewed-uncrewed separation will often happen in the vertical plane first, with drones flying lower, this lowest-predictability dimension poses the most significant challenge for airspace integration. A common altitude reference is also unresolved: the helicopter’s cockpit instruments use barometric and radio altimeters while most small drones rely on GPS altitude alone, a mismatch researchers are addressing through projects like ICARUS.

The findings point toward concrete changes in how VFR flights are planned and shared. Flight plans for volume-constrained operations could include mission type, target landmark, target altitude and speed, and situation-awareness equipage, information that both enables volume design and lets other airspace users avoid assigned volumes. Andreeva-Mori suggests designating airspaces where VFR operations are volume-constrained by default, with users holding predetermined plans receiving priority, creating an incentive for pilots to share their intentions. Implementation challenges remain, including integrating landmark coordinate data into operational systems, developing near-term position prediction alerts to warn pilots before they leave a volume rather than after, and adapting training and certification, which currently do not cover instrument-related skills for VFR pilots. The author proposes starting with disaster response, where D-NET technology is already fielded and lessons can inform future standards. If the approach scales, the long-standing trade-off between the freedom of visual flight and the structure of instrument flight may finally be resolved, opening low-altitude airspace safely to the drones and air taxis waiting to enter it.

Subject of Research: Low-altitude airspace integration

Article Title: Balancing predictability and flexibility through operation volume-constrained visual flight rule operations in low altitude airspaces

Article References: Andreeva-Mori, A. (2024). Balancing predictability and flexibility through operation volume-constrained visual flight rule operations in low altitude airspaces. Frontiers in Aerospace Engineering, 3, Article 1338388. https://doi.org/10.3389/fpace.2024.1338388

Image Credits: AI Generated

DOI: 10.3389/fpace.2024.1338388

Keywords: Balancing, predictability, flexibility, operation, volume-constrained, visual, flight, rule, operations, altitude, airspaces, scientific research