01
Flight physics coupled payload, endurance and safety
Mass, rotor disc area and pack energy determine hover power and endurance together. Adding a sensor or recovery system changes the same equation the aircraft depends on to remain airborne.
Selected work
Product design case study
RD-A7
Designing aerial inspection around measurements that can be compared—not photographs that cannot.
RD-A7 is Root Digit’s four-rotor aerial inspection platform for repeatable surveys of fixed assets. Its airframe, positioning, sensing, onboard inference and safety controls are designed from the defect that must be measured and the frequency at which it must be checked.
- Platform
- RD-A7
- Sector
- Industrial inspection and autonomous systems
- Engagement
- Product design, flight systems and sensing architecture
- Period
- Active Root Digit Labs programme
The problem
What the platform needed to solve
Most aerial inspection produces photographs. A useful inspection programme needs evidence that can be compared across time: the same feature at a known position, resolution and viewing geometry, with enough sensor fidelity to distinguish change from capture error.
The design problem was therefore larger than building a stable drone. The platform had to repeat survey lines, hold the required standoff, preserve georeferencing through degraded navigation conditions and return visual and radiometric data whose measurement limits were known before flight.
Why it was difficult
Constraints inside the problem
These operating and governance conditions determined what a credible solution had to achieve.
02
Repeatability required centimetre-level position and heading
Metre-level standalone GNSS cannot place a later survey over an earlier one. Position, heading and time therefore needed a shared precision architecture with a post-processed fallback.
03
Defect resolution had to be known before take-off
Altitude, focal length and pixel pitch determine ground sample distance. A faster flight at greater standoff is not useful if the smallest relevant defect disappears below the sensor’s reliable detection limit.
04
Industrial environments contain correlated failures
Low texture, dust, glare, smoke, GNSS multipath and link loss can occur together. Redundancy had to combine sensing principles that fail differently and preserve a recoverable state.
The solution
What Root Digit built
Root Digit designed RD-A7 as a repeatable measurement platform: a 4.2 kg four-rotor aircraft with integrated multi-band GNSS, network RTK, dual-antenna heading, radar terrain following and visual-inertial navigation for degraded positioning conditions.
A mechanical-shutter visual sensor and radiometric thermal channel provide comparable inspection evidence, while onboard inference can identify candidate defects during flight. Triple inertial sensing, independent recovery, encrypted storage and signed firmware keep navigation, evidence and platform control within one governed design.
How we did it
Engineering the solution
The workstreams below show how the solution was designed, built and controlled.
01
Coupled airframe, power and payload design
Hover power was derived from take-off mass and total disc area, then reconciled with drivetrain loss, usable pack energy, endurance and payload. The same approach sized the parachute recovery target rather than treating it as an accessory.
- 4.2 kg maximum take-off mass
- 0.55 m² rotor disc area
- 266 Wh pack energy
- 1.2 kg usable payload
02
Repeatable positioning and heading
Integrated multi-band GNSS accepts network RTK corrections over LTE, logs raw observables for PPK fallback and uses a 0.55 m dual-antenna baseline to stabilise heading for georeferencing.
- Network RTK via NTRIP
- PPK fallback
- Dual-antenna heading
- Time-aligned sensor records
03
Terrain following and degraded navigation
Radar altitude closes the loop against the surface below the aircraft, while visual-inertial odometry supports motion estimation when GNSS is obstructed or corrupted by multipath.
- Radar altimetry
- Visual-inertial odometry
- Prior surface model
- GNSS-quality monitoring
04
Inspection resolution and sensor evidence
The imaging design works backward from the smallest defect. Ground sample distance, standoff and flight speed define mission limits, while radiometric thermal data preserves temperature values for comparison rather than only a false-colour image.
- 20 MP mechanical-shutter imaging
- 640 × 512 radiometric thermal
- Resolution-aware mission planning
- Comparable defect records
05
Independent safety and evidence controls
Triple inertial sensing with median voting, dual barometers, complementary stereo and radar obstacle sensing, a ballistic recovery system and secure device identity reduce single-point failures across flight and data handling.
- Triple IMU
- Stereo plus 60 GHz radar
- Signed firmware
- Encrypted onboard storage
System design
Architecture and controls
01
Mission-definition path
Required defect size and inspection frequency determine sensor, standoff, route spacing and mission cadence before a flight plan is accepted.
02
Navigation and safety path
RTK, dual-antenna heading, visual-inertial odometry and radar altitude support the flight state; independent limits and recovery controls bound what the autonomy system may do.
03
Evidence path
Georeferenced visual and radiometric observations are retained with position, time and mission context so later surveys can be compared against the same asset features.
Outcomes and evidence
What the current specification establishes
These figures are the current coupled engineering specification for the platform; they are not presented as customer-deployment outcomes.
- ± 2 cm
- horizontal positioning
- 36 min
- endurance at MTOW
- ≈ 1.2 mm
- reliable crack-detection basis
- 1.2 kg
- usable payload
Current specification with RTK fixed.
Derived from the coupled mass, disc-area and pack design.
At 3 m standoff under the specified visual geometry.
For the primary or optional inspection sensor configuration.
Product design conclusion
RD-A7 is designed from the inspection decision backward: define what must be detected, preserve the geometry needed to detect it again, and govern the flight and evidence chain so two surveys can be compared with technical confidence.
Start a conversation