AutoPilot NDT
The Challenge
Name:
Human Operator interface for FA-IUAS
Domain:
Cargo hull & ballast tank
Challenge proposer:
Objectives:
Developing a high-TRL operator interface for the FA-IUAS to unlock the capabilities of fully actuated aerial platforms in autonomous contact-based NDT inspections.
AutoPilotNDT addressed AUTOASSESS Challenge 3: developing an intuitive human operator interface for the fully actuated inspector UAS (FA-IUAS). The core problem is that contact-based NDT inspection of GNSS-denied confined spaces—ballast tanks, cargo holds—still relies on human surveyors in dirty, dangerous conditions, while capable aerial platforms remain hard to use and deploy (initially command-line only, motion-capture dependent, TRL 2). We addressed this by delivering a web-based supervisory interface (mission monitoring, point-cloud inspection-point selection, status feedback) coupled with an onboard autonomy stack (SLAM localization, trajectory planning, state machine, ROS-GenoM bridge), containerized for one-command deployment-raising the platform from TRL 2 to TRL 4.
The Solution
Name:
AutoPilot NDT – AutoPilot NDT: Scalable Autonomous Aerial Inspection with Intuitive Human Interface
Description:
The solution integrates an autonomy stack and a web-based UI to make autonomous contact NDT with an aerial platform (FA-UAS) supervisable rather than piloted. The stack comprises four modules: an FSM mission planner (manual or autonomous), SLAM localisation using IMU/LIDAR for GNSS-denied spaces like ballast tanks, a trajectory planner for energy-efficient exploration, and admittance control for contact inspection via external wrench estimation.
The UI adds state selection (EXPLORE/INSPECT/RETURN), sensor-feedback visualization, and live measurement display. Together they enable autonomous inspection of enclosed environments, cut operator training, shift surveyors from dangerous work to supervisory roles.
Key Results Delivered:
We delivered a two-part, containerized solution that raised the FA-IUAS from TRL 2 to TRL 4. The autonomy stack provides GNSS-denied onboard localization (LiDAR-inertial SLAM, validated against motion-capture ground truth at <0.08 m horizontal RMSE in motion), RRT plus minimum-snap
trajectory planning, a mission state machine, and a ROS-GenoM bridge to the existing flight controller. The web-based operator interface enables mission monitoring, live LiDAR/camera visualization, and inspection-point selection directly on the point cloud. On real hardware we demonstrated SLAM-localized hover, go-to, trajectory execution, and contact-based inspection commanded from the GUI. Everything deploys via one simple command.
Unique Value and Impact:
AutoPilotNDT’s unique value is turning a research-grade fully actuated aerial platform into a deployable, supervisable inspection system. The differentiator isn’t any single algorithm—it’s the integration: onboard GNSS-denied localization, contact-based NDT control, and an intuitive point-cloud-based operator interface, all containerized for one-command deployment. This lets a non-specialist supervise inspections rather than pilot them, transitioning surveyors out of dirty, dangerous confined-space work into safety roles.
The FA-IUAS advanced from TRL 2 to TRL 4, with horizontal localization accuracy well within ballast-tank tolerances, and delivered a reusable open-source stack that de-risks future full-mission demonstrations and commercial exploitation across maritime inspection.
Solution in Action:
An inspection supervisor arrives at a vessel in dry dock. Instead of sending a surveyor into a ballast tank dark, confined, physically punishing—they deploy the FA-IUAS and open the web interface on a tablet. One Docker command brings the full system online.
The operator taps TAKE OFF. The drone lifts and begins localizing itself using onboard LiDAR-inertial SLAM, with no GNSS and no external motion capture, critical inside a sealed steel tank. The live point cloud streams to the tablet, building a 3D view of the space as the platform moves. The operator sees battery, signal, and pose telemetry at a glance.
To inspect a suspect weld, the supervisor simply clicks the point on the point cloud. The autonomy stack plans a smooth trajectory, approaches, and makes controlled contact for the NDT measurement, with live force feedback confirming a clean reading. The operator supervises rather than pilots throughout.
Primary users are maritime inspection and NDT teams; the most accessible entry market is shipyards inspecting ballast tanks during vessel construction, where procurement is more responsive. The solution generalizes to cargo holds, and-because the interface is platform-agnostic to confined-space inspection in other industries via UAV OEM partners.
The Solution Provider
Crobotic Solutions is robotics start-up developing AI enabled robotics solution with advanced autonomy for real-world environments.
Main contribution of the AUTOASSESS programme:
For us, AUTOASSESS’s main value was structural. It provided a credible, funded pathway to move the FA IUAS from research artifact to validated system (TRL 2 → 4) with a demanding real partner, University of Twente, whose platform and constraints kept the work honest. The consortium (DNV, class societies, shipowners, sensor and drone OEMs) gave direct access to the maritime inspection domain, end-user validation, and downstream commercial contacts we couldn’t reach alone. The FSTP structure de-risked deep-tech development that’s hard to fund pre revenue, while the deadlines and reporting cadence forced disciplined scoping.
Testimonial
“AUTOASSESS was one of the most valuable programmes we’ve taken part in as a young deep tech company. The FSTP structure gave us the runway to tackle hard robotics work that’s difficult to fund pre-revenue, while the partnership with University of Twente kept us grounded in real hardware and real constraints. Working alongside DNV, shipowners, and sensor and drone OEMs opened doors to the maritime inspection market we couldn’t have reached alone. Over six months we moved our platform from concept to laboratory-validated system and built lasting relationships. We’d recommend it without hesitation to startups building applied robotics for industrial inspection.”
Filip Zorić, Crobotic Solutions CEO