This article first appeared in the Railway-News Magazine Issue 1, 2022.
Rail Vision is revolutionising railway safety and maintenance. Using the latest artificial intelligence (AI), deep learning and sensor technology, this Israel-based technology provider has developed state-of-the-art solutions that lower the risk of accident, unplanned downtime and maintenance costs through advanced obstacle classification technology.
Bursting into the market in 2016, Rail Vision has launched advanced safety systems developed specifically for mainline, shunting yard and light rail environments. These can be engineered to provide a system that meets each customer’s specific needs.
Rail Vision’s CEO Shahar Hania has been with the company from the start. With experience working in optical engineering, systems engineering, electro-optics and deep learning technologies across the communication, defence and train industries, he started Rail Vision as a passion project with his three fellow co-founders.
Their evenings and weekends were spent developing a rail-specific obstacle detection and classification system, while also approaching potential investors. Interest in the technology was high, and it wasn’t long before they had the financing they needed and became the first company to bring such a solution to the industry.
Hania first took on the role of Head of Technology Research and Development (R&D), only later stepping into the position of CEO. His focus was on optimising Rail Vision’s technology, and it was his decision to embrace AI.
“I was familiar with computer vision and knew its capabilities – and limitations. I could see that AI was the only technology that could do what we wanted.”
This is because, unlike computer vision, AI has the ability to imitate intelligent human behaviour and enables machines to process information and make decisions based on logic and reasoning. Taking into account a broader set of factors, it can determine the best-possible outcome based on the input it receives.
AI, paired with electro-optic sensors and deep learning, is what enables Rail Vision’s solutions to detect and classify obstacles such as people, up to 1.5km ahead, while trains and cars can be classified 2km ahead.
Monitoring a predefined area of interest, the system generates real-time visual and audio alerts for both the train operator and the command-and-control centre, and is, to date, the only system that can detect and classify a person at risk in time to avoid a collision.
Use the form opposite to get in touch with Rail Vision Ltd directly to discuss any requirements you might have.