9.1 GNSS Position Hold, Drift & Loss of Positioning
Key Takeaways
- GNSS (GPS and other satellite constellations) gives multirotors position hold so they can hover and resist wind drift without continuous stick work—useful near people but never a substitute for separation, VLOS, or pilot skill.
- Urban canyons, tall buildings, bridges, and reflective surfaces cause multipath and weak satellite geometry; expect degraded hold and sudden mode changes during typical A2 property and street work.
- When satellites are lost, many aircraft drop to ATTI (attitude) or similar non-positioning modes: the craft keeps level but drifts with wind, so the pilot must take active manual control immediately.
- Do not use GNSS alone to guarantee distance from uninvolved persons—legal floors (30 m / 5 m UK2, 50 m legacy) and the no-overflight rule still bind when hold is perfect or degraded.
- Near tall buildings, plan take-off in open sky, watch satellite count and HDOP/quality indicators, keep VLOS, and be ready to climb out, reposition, or land if hold becomes unreliable.
Why GNSS Appears on UAS Flight Performance
The A2 CofC additional theory covers UAS flight performance because Near People operations leave little room for surprise. GNSS (Global Navigation Satellite System—GPS, Galileo, GLONASS, BeiDou, and similar constellations used by the aircraft) is how most modern multirotors know where they are in the horizontal plane and often how they hold altitude more precisely when barometer and vision sensors are combined. When GNSS is healthy, the flight controller can command small motor corrections so the aircraft holds position in a hover and resists gentle wind drift without constant pilot stick input.
That comfort is double-edged. Many pilots fly closer to people than they would in pure manual mode because "the drone is locked." The exam and real A2 flying reject that comfort as a safety case. Position hold is a performance aid, not a legal separation system and not a failsafe against multipath, urban canyons, or sudden satellite loss.
How Multirotors Use GNSS for Position Hold
A typical consumer or prosumer multirotor fuses:
- GNSS for absolute horizontal (and often vertical) position relative to the Earth frame.
- Barometer / IMU for attitude, angular rates, and altitude trends.
- Often vision / downward sensors for fine hold over textured ground at low height.
- Compass (magnetometer) so the aircraft knows which way "forward" is in the horizontal plane (covered in detail in section 9.2).
With a solid satellite fix, the controller compares the current position estimate with the commanded hover point. If wind pushes the aircraft 0.5 m north, the motors tilt slightly south to return. The pilot sees a calm hover. For A2 work this helps you:
- Hold a planned horizontal buffer beside a façade or path without fighting every gust with large stick inputs.
- Execute slow orbits or grid patterns more predictably when low-speed mode is also engaged.
- Reduce pilot workload so you can scan for approaching uninvolved persons and maintain VLOS.
What position hold does not do:
- Authorise BVLOS (still outside Open A2).
- Replace the Flyer ID, A2 CofC, Operator ID, or class-mark rules.
- Guarantee that the aircraft will stay outside a 5 m or 30 m person bubble if the pilot commanded the wrong hover point, if sensors fail, or if wind exceeds performance.
- Double battery life or cancel kinetic energy if the aircraft hits someone.
Exam table — GNSS hold vs pilot duty
| Feature | What it does | What you must still do |
|---|---|---|
| Healthy GNSS hold | Resists drift; stable hover | Keep legal distance; no overflight; VLOS |
| Degraded GNSS | Intermittent corrections; "floating" | Increase buffer; prepare to take control |
| GNSS lost / ATTI | Level flight only; drifts with wind | Active manual control; land or exit if needed |
| "Full bars" on app | Suggests good fix | Still verify sky view; do not trust alone for people separation |
Multipath and Urban Canyons — The A2 Reality
Much Near People work happens where people live and work: streets between buildings, courtyards, estate façades, bridges, multi-storey car parks, and glass-fronted offices. These environments create two classic GNSS problems:
- Urban canyon geometry — tall buildings block satellites low on the horizon. The receiver may see only a few satellites overhead, so position geometry is poor even if the icon shows "GPS."
- Multipath — satellite signals reflect off glass, metal cladding, wet roads, and vehicles. The receiver may lock onto a reflected path longer than the true line of sight, so the computed position jumps or slowly walks away from truth.
Symptoms you must recognise for the exam and for flight:
- Hover point slowly creeps toward a building or path.
- Sudden position jumps of several metres with no stick input.
- RTH (return-to-home) aiming at a home point that was recorded with a bad fix.
- Mode changes, "weak GPS," "positioning unavailable," or automatic switch to ATTI / manual.
- Vision sensors fighting GNSS over contrasting surfaces (water, glass, moving vehicles), producing oscillation.
Exam scenario pattern: pilot launches beside a steel-and-glass tower for marketing shots, relies on the green GPS icon, holds 8 m from a pavement in low-speed mode, then the aircraft drifts into the buffer as multipath worsens. Correct mindset: urban A2 sites are high-risk for GNSS quality; plan open-sky take-off, larger buffers, and abort criteria for position quality—not maximum closeness justified by "GPS is on."
Drift When Satellites Are Lost
When the aircraft loses enough satellites or rejects the solution as unreliable, many designs drop positioning assistance. Marketing names vary (ATTI, Attitude, Manual, GPS-off, P-GPS lost), but the performance idea is stable for A2 theory:
- Attitude / non-positioning mode: the aircraft keeps roughly level using the IMU; it does not auto-correct horizontal drift.
- Wind and residual tilt move the aircraft; without pilot input it will wander.
- Motors do not usually cut (that would be a different failure); they keep flying while the pilot becomes the position controller.
What the pilot should expect and do:
- Expect drift immediately — especially in any wind that would have been invisible under hold.
- Take active stick control — hold position manually relative to a visual ground reference and the people bubble.
- Maintain or increase separation — do not stay at 5 m hoping the fix returns; climb, back away, or land if control workload spikes.
- Keep VLOS — mode changes often coincide with the aircraft moving; eyes and orientation matter more than the map icon.
- Avoid panicking into stick extremes that accelerate toward people; smooth inputs, then land if unstable.
Exam trap: "If GNSS is lost the motors stop and the drone falls." Often false for multirotor attitude modes—motors continue; drift is the issue. Another trap: "The drone will hold position perfectly without pilot input after GNSS loss." False—that is the opposite of attitude mode.
ATTI / Manual Modes and Near People Ops
For A2 candidates, ATTI/manual capability is not optional trivia—it is the recovery layer when the technical GNSS layer fails (Chapter 5 layering: tech can fail; operations and distance must still protect people).
Implications for the three UAS.OPEN.030 evaluations before 5 m work:
- Performance of the unmanned aircraft includes whether GNSS quality supports precise hold near people. Poor satellite geometry is a performance degradation.
- Weather (wind, gusts) multiplies ATTI drift rate; gusty ATTI near a path is a classic no-go.
- Segregation must assume that a sudden mode change could add metres of unplanned travel within seconds.
If you only ever practised in open parks with perfect GPS, declare yourself not yet ready for tight urban A2 until you have practised deliberate control without relying on hold (under safe A3-type conditions during self-practical training).
Maintain VLOS and Be Ready to Take Manual Control
Open Near People operations are VLOS. GNSS maps, first-person camera views, and "follow me" styles of attention do not replace eyes on the aircraft and the surrounding people. When positioning degrades:
- The aircraft may move off the planned hover toward an uninvolved person you are not watching on the screen.
- Orientation (which way the nose points) may confuse stick directions if you were passive under hold.
- RTH or brake functions may behave differently if home point or positioning is corrupt.
VLOS + ready hands is the operational mitigation that pairs with GNSS as a technical aid. If you cannot see the aircraft clearly enough to correct drift, you are already outside the safe envelope for close A2 work—even if the app shows coordinates.
Do Not Rely on GNSS Alone for Separation from People
Statutory and syllabus points that must stay cold:
- UK2/C2: ≥ 30 m horizontal from uninvolved persons in normal mode; ≥ 5 m only with active manufacturer low-speed mode after evaluating weather, performance, and segregation; no intentional overflight.
- Eligible legacy < 2 kg: typically ≥ 50 m person floor under the A2 path taught earlier.
- GNSS hold accuracy of "±0.5 m" in a brochure does not create a private right to sit 0.5 m from a stranger.
- A green GPS icon does not prove the three evaluations passed.
Think of GNSS as helping you execute a separation plan you already chose. It does not define the plan. If hold fails, the plan must still keep people safe through pilot skill and extra distance.
Exam Scenarios Near Tall Buildings
Scenario A — glass office canyon. Pilot records home point between two towers. GNSS multipath shifts the home point. Later RTH flies toward a different courtyard entrance where people walk. Lesson: set home in open sky, verify home on the map against visual landmarks, and treat RTH as a degraded option near people, not automatic safety.
Scenario B — estate agent 5 m low-speed shot. Satellite count drops as the aircraft flies into the building shadow. Mode flips to ATTI; a crosswind drifts toward the pavement. Correct action: manual control, increase horizontal distance or land, do not "wait for GPS to come back" at 5 m.
Scenario C — "GPS will keep me 30 m away." Pilot sets a waypoint path that skirts a park path and looks only at the screen. A position jump puts the aircraft over the path. Correct doctrine: VLOS, human monitoring of the bubble, and no trust of GNSS alone for separation.
Scenario D — bridge deck. Metal structure, vehicles, and limited sky. Compass issues (9.2) and GNSS multipath combine. Best exam answer often: relocate or postpone rather than force a Near People profile.
Pre-Flight and In-Flight GNSS Checklist (A2)
- Launch and wait for a stable high-quality fix in the most open part of the site.
- Note satellite count / quality indicators; do not rush take-off with "searching."
- Confirm home point matches the real take-off spot on the map and by eye.
- For urban façades, plan extra buffer beyond the legal floor where multipath is likely.
- Brief yourself on the ATTI response: hands ready, eyes on aircraft, abort lines clear.
- If quality collapses mid-job, widen distance or land—performance evaluation has failed for 5 m work.
- Never use GNSS degradation as an excuse to break VLOS or overfly people "just to finish."
Memory Hooks for Section 9.1
- GNSS hold = drift resistance aid, not a people-separation licence.
- Urban canyon + multipath = expect jumps and ATTI.
- GNSS lost → level flight, wind drift, active manual control.
- VLOS and ready sticks remain mandatory.
- Tall-building A2: open-sky fix, verify home, extra buffer, abort early.
Master those five and you will correctly answer most GNSS stability and emergency items on the A2 paper.
What is the main benefit of GNSS/GPS position hold for a multirotor during Near People (A2) operations?
If GNSS signal is lost mid-flight and the multirotor switches to attitude (non-positioning) mode, what should the remote pilot expect?
Why are tall buildings and urban canyons a special GNSS risk for A2 property or street work?
A pilot argues that because GNSS hold is accurate to about half a metre, flying 2 m horizontally from uninvolved persons on a UK2 is acceptable without low-speed mode. Which statement is correct?