Achieving Aerospace Drilling Tolerances with Robot Arms: Calibration and Accuracy Methods
The problem
Aerospace structural assembly typically requires hole placement accuracy of ±0.25 mm or better. Standard industrial and collaborative robot arms, without additional calibration, deliver absolute accuracy in the range of ±0.5–1.5 mm. This gap has historically limited robotic drilling to applications with looser tolerances, or required expensive metrology equipment — primarily laser trackers — to correct robot positions in real time.
The accuracy shortfall has specific technical sources. Nominal kinematic models embedded in robot controllers carry parameter errors from manufacturing tolerances — link lengths, joint offsets, and coupling ratios that differ from the nominal DH parameters. These errors are compounded by joint compliance, thermal drift, and gravitational deflection, all of which vary with configuration and payload. For a six-axis arm with 28+ kinematic parameters, even small individual errors accumulate to millimeter-level positioning error at the TCP.
Calibration approaches
The traditional approach to achieving aerospace-grade accuracy involves a laser tracker measuring the TCP at dozens to hundreds of positions across the workspace, followed by parameter identification to fit the kinematic model to the measured data. This method works but carries significant practical costs:
Laser tracker systems cost $80,000–$250,000+
Each calibration session requires line-of-sight between the tracker and the target
Recalibration is needed after any disturbance — tool change, cell reconfiguration, thermal cycling
The tracker itself must be calibrated and certified periodically
An alternative approach uses lower-cost measurement methods — accelerometers, touch probes, constraints, or self-motion — to identify kinematic parameters without a laser tracker. Recent research has demonstrated kinematic calibration reducing errors from ~10 mm to 0.2 mm on collaborative platforms (Franka, KUKA, Kinova) using methods that do not require external metrology equipment.
Beyond static calibration
Kinematic calibration corrects the geometric model but does not address dynamic error sources: joint compliance under load, thermal expansion during operation, and backlash or friction effects that vary with speed and direction. For aerospace drilling, where the tool contacts the workpiece and the process generates forces and heat, these dynamic effects can introduce errors that static calibration cannot predict.
Dynamic joint tracking provides a complementary layer — real-time compensation for joint-level errors during operation. Combined with a calibrated kinematic model, this can address a larger fraction of the total positioning error budget.
Performance context
Kinematic calibration alone can improve absolute accuracy by 3–5x, depending on the platform and initial error magnitude
Combined calibration and dynamic tracking can address both geometric and non-geometric error sources
The remaining error budget after software compensation depends on the mechanical quality of the arm — joint stiffness, bearing quality, and structural rigidity set the floor
Integration
Calibration is performed per-arm, capturing the specific manufacturing deviations of that unit
No permanent metrology equipment required in production
Works through the robot's command interface — no controller firmware modification
Recalibration or revalidation is needed after tool changes, cell reconfiguration, or significant thermal cycling

Iago Alves Pereira
Co-Founder and CTO
Share post
Written by
Iago Alves Pereira
See what Reforge can improve on your robot
Bring one representative trajectory or error dataset. We’ll identify the relevant product, required inputs, and a bounded evaluation plan.



