By David BenDavid, CEO, Rail Vision Ltd.
Sitting in the cab of a switcher locomotive at dawn—its twin Caterpillars rumbling underneath—reminds me why I first joined Rail Vision eight years ago. Back then, as a newly minted CTO, my focus was a narrow but noble one: teach silicon wafers to see as well as human eyes, then teach algorithms to notice what humans miss when fatigue and fog set in. Today, as I return as CEO, the view from that same cab feels broader. I still see steel, ballast, and a screen full of infrared telemetry, but I also see a world wrestling with climate targets, supply-chain bottlenecks, and a workforce asked to do more with less. The transition from chief technologist to chief executive is less a change of altitude than a widening of the lens. In both roles, clarity of vision—literal and strategic—determines where we arrive and whether we arrive on time.
When the board asked me to step back into Rail Vision they asked for applied imagination: keep the trains running, but also make them smarter, safer, and cheaper to run. That mandate mirrors a global truth. Every Class in North America, every operator in Europe, and every mine-to-port operator is being told to deliver more ton-miles with fewer incidents.
I inherit a product line that detects an SUV a mile ahead in pitch darkness, classifies it in 30 milliseconds, and alerts the driver before the braking curve becomes fatal. What I need to add is a system-level mindset that treats those sensors as the first layer of a much larger digital fabric.
Imagine a freight lattice where every wheelset, camera, and switch speaks a common language; where a deviation in axle temperature triggers not just a hot-box detector alarm but an automated slow-order, a re-routing instruction, and a spare crew dispatch, all before the wheel group shells. That’s the orchestration layer we’re building. It’s what turns sensors into savings.
Engineers often ask for a green light to chase perfection. CEOs ask how perfect we need to be for customers to sign. My answer is aspiring to 99% obstacle detection accuracy with a false-positive rate low enough to keep crews from reaching for the circuit breaker. The harder part is translating the savings into language the CFO believes. A single 30-minute classification-yard closure can cost tens of thousands of dollars in cascading delays and millions in direct damage after a yard collision. Remove just one of those events and the sensor suite amortizes itself in months.
But safety metrics alone may not be enough to win procurement. Clients want to see how the data exhaust improves dwell time, network velocity, and ultimately EBITDA. That pushes us into partnerships.
In 2017, I was told: “Rail doesn’t need eyes; it needs more tonnage.” A decade later, every ton that derails ends up on TikTok, and every minute of crew overtime hits operating ratios operators scrutinize to the second decimal place. Vision is no longer a luxury; it’s an operating necessity. As CEO, my goal is to embed that necessity so deeply into rail economics that nobody remembers the industry without it.
Rail is the original network industry. Telegraphs rode its rights-of-way, dieselization followed steam’s path, and broadband fiber will soon share the ballast. Our task is to make sure computer vision rides those rails just as naturally. If I do my job right, a kid boarding a passenger train in 2030 won’t think about the sensors guarding the grade crossing ahead; they’ll just arrive safely, on schedule, in a train that outperforms trucking on both emissions and economics. And the rail workers who maintain that system will go home to their families at the end of every shift. That, ultimately, is why I left the lab bench for the corner office: because technology only matters when it changes lives, not just line items.
