Grace Chia, BeeX: Can an underwater robot decide like a human?
Underwater inspection barely depends on anything underwater. It depends on the ship that carries the vehicle out, and on the handful of people experienced enough to fly it. Those costs are skyrocketing.
We spoke with Grace Chia, Co-founder and CEO of BeeX, a company that designs, builds and operates a fleet of autonomous underwater vehicles to make underwater infrastructure inspection and maintenance safer, faster and more efficient at scale. It is where she built a company around autonomy, not a bigger machine.
Why a bigger machine was never the answer
The industry's instinct was to build bigger vehicles, more powerful thrusters and deeper depth ratings. Chia calls these good hardware features to work on. Her point is that none of them make the work scale.
The logic behind that instinct is sound on its own terms. If you want to win the next inspection contract, you invest in hardware capability, not software capability. Hardware is visible, specifiable and immediately billable.
Chia's conviction, which she says has not changed since she started, runs the other way. Autonomy is the capability that unlocks the others. Without it, the ability to do work never scales, however capable the vehicle.
What underwater autonomy actually means
Autonomy is a broad term. Chia reduces it to one question: how do you replicate what a human decides, and get the robot to do the same thing?
That is harder than navigation. An experienced pilot arrives at a structure, reads it, and decides what matters, what to look at twice, and what to ignore. Replicating that judgement is the engineering problem underneath autonomous underwater inspection.
It is also why the problem resists a single solution. The decisions differ by structure, by site and by condition, and the vehicle has to reach them without a person on the other end of a tether.
How an autonomous vehicle learns a site
There is no version of this you solve once. Chia describes a cycle that runs for every new location. The hardware and the software pull each other forward. Software capability runs into what the vehicle physically cannot do, the hardware is upgraded to raise that ceiling, and the loop runs again at the next site.
Chia's comparison is to how people learn. When we are very young we know nothing; we accumulate. Her vehicles go through the same process, one structure at a time.
Why the field stayed still for so long
Underwater work went decades without much changing. Chia's explanation is not that nobody tried. It is that the barrier to putting anything new in the water is high enough to keep most people out.
Every deployment needs a vessel, a crew, a window of weather and a customer willing to let an unproven system near their asset. That last one is its own obstacle. In an established company, Chia notes, customers accept that a system will work. In a startup, everything has to be proven before trust is extended.
Her advice to a founder entering a field like this is that it takes longer than you expect. The work is iterative and the gains accumulate in small increments, which is exactly why the field had not been crossed before.
Comments