The wider robot family: inside pipes, on walls, under water and from orbit

In sewer pipe inspection, self-propelled tethered camera crawlers have long been the workhorse. Lowered through a manhole, they drive along the pipe on a cable and film the inside, finding cracks, displaced joints, root intrusion and corrosion. Newer crawlers add laser or sonar profiling to measure pipe shape.

In pipelines, devices called pigs travel inside the pipe with the product flow. Originally for cleaning, intelligent pigs (in-line inspection tools) carry magnetic flux leakage (MFL) or ultrasonic sensors to find wall loss and cracks over tens of kilometres.

Climbing robots cling to steel with magnets or to concrete with suction or negative pressure, and drive across the wall. Unlike a hovering drone, they stay pressed firmly against the surface while moving, which makes them good at scanning (mapping) ultrasonic thickness over whole areas.

• Magnetic crawlers: storage tank shells, ship hulls, steel chimneys and boiler tube walls.
• Suction / negative pressure: concrete walls, dam faces, building facades.
• Cable or gondola-based: suspended from above to scan tall facades.
• Combined with drones: a drone surveys the whole asset and narrows down suspicious areas, then a crawler measures them closely. There are also trials of drones carrying crawlers to the wall.

Dam intakes and submerged faces of dams, port quay walls and ship hulls lie under water. Looking at them used to be a diver’s job. Increasingly, ROVs (remotely operated vehicles) dive on a tether and inspect with cameras and sonar.

A common division of labour: drones above the waterline, ROVs below it, and uncrewed surface vessels (USVs) on the surface itself. See also water, spray & humidity.

In process plants and substations, operators walk set routes every day to read gauges, listen, and check for smells and heat. Examples are emerging of legged robots (e.g. Boston Dynamics Spot, ANYbotics ANYmal) and wheeled robots taking on these routine rounds. Their strength is climbing stairs and steps and using the same walkways as people.

• Strengths: reading analogue gauges, thermal imaging, abnormal sounds, gas-leak detection, fixed-point imaging.
• Versus drones: no flight, so long run times and good in narrow indoor walkways; poor at height.
• Explosion protection: use in explosive atmospheres requires certified explosion-proof versions.
• Combinations: walking robots on the ground, drones for pipe racks and tower tops.

A common framework such as ROS 2 is usually what ties ground robots and drones into one monitoring system. Connecting perception, framing and flight monitoring with ROS 2 (Airffic Aerial) is written for filming, but the integration ideas carry over to inspection.

The most distant inspector is the satellite. Optical satellites photograph wide areas, and synthetic aperture radar (SAR) observes the ground regardless of cloud or darkness. Interferometric SAR (InSAR), comparing repeated SAR images, can sometimes detect slow millimetre-to-centimetre displacement of ground and structures over wide areas.

Choose by three questions: where the asset is (air, water, pipe, wall, ground), what to measure (appearance, shape, thickness, displacement), and how much area, how often.

Inspection is heading towards several kinds of robot sharing the work on one asset. What matters then is not the robot types but whether the data come together in one register. Only when drone photos, crawler thicknesses, ROV sonar and satellite displacement overlay on the same member IDs and coordinates does the whole picture of a structure emerge.

• Common coordinates and member IDs: store every robot’s data at the same “address”.
• Hand-offs: route suspicious spots found by drones automatically to crawlers or people.
• Coordinating movement: when several robots and people work one site, share who is where (remote and BVLOS operations).
• Know the limits: no robot does everything. People still verify and decide in the end.