A bird following a flockmate has a moving opportunity ahead of it. Air rising beside the leading bird’s wing can help support the follower, while air pushed downward creates a different challenge. The useful place depends on the wake and the timing of the wingbeats.[2]
In a 2014 study, Steven Portugal and colleagues recorded the positions and wingbeats of 14 northern bald ibises during migratory flight. Birds flying behind and to one side timed their wing motion to follow the leading bird’s wingtip path through the air. Birds directly behind used a different phase relationship, consistent with avoiding the strongest downwash.[2]
That raises a sensor question for an aircraft: knowing where the leader is and knowing what the local air is doing supply different pieces of information. A formation controller needs a way to connect the two.
A nose gives the model another clue
A Communications Engineering study published on 29 September tested nose and wing pressure taps on a simplified fixed-wing model. Wind-tunnel measurements covered one leader and one follower. An XGBoost model predicted changes in the follower’s lift-to-drag ratio.[1]
Lift-to-drag ratio compares supporting force with resistance to forward motion. The prediction task was to estimate how that ratio changed within the wake. An error in that estimate is therefore an error in aerodynamic-state prediction; fuel consumption is a separate measurement.
What information the prediction received
| Inputs | Prediction error |
|---|---|
| Relative position alone | 0.85Source 1 |
| Position and two nose taps | 0.61Source 1 |
| Position and all twelve taps | 0.53Source 1 |
Using these rounded errors, we calculate that two taps supply 0.24 of the twelve-tap array’s 0.32 improvement: three quarters. Twelve taps still have the lowest error, leaving an engineer to weigh that remaining improvement against the extra hardware.
Repeated measurements of a position were assigned to the same fold, so a repeat could not appear in training while its partner appeared in evaluation.[1]
A clue to read, then a position to choose
Pressure alone performed substantially worse. Relative position supplied the context needed to interpret the local measurement, with distance downstream particularly useful. The experiment used one steady airspeed and did not test closed-loop flight control or the sensory mechanism of a living bird.[1]
A future controller would have two jobs: estimate whether its current patch of air is helpful, then choose and execute a movement. If the goal is to stay in useful air, a richer sensor array would have to earn its extra hardware through a better flight outcome.
Sources & further reading
- Nares-inspired pressure sensing enhances aerodynamic state perception for formation flight Communications Engineering · 29 Sep 2026 · Accessed 18 Oct 2026
- Upwash exploitation and downwash avoidance by flap phasing in ibis formation flight Nature · 15 Jan 2014 · Accessed 18 Oct 2026