A proactive rail safety device that detects obstacles and track damage just in time to prevent disasters


Canada suffered from 10 main-track and 527 non-main rail track derailments in 2016 alone. 13% of these derailments involved dangerous goods.

Canada carries over 300 billion tonne-kilometres per year, which represents 68% of Canadian freight. Canada is among the heighest in the world of rail tonnage carried per capita and percentage of frieght carried by rail.


February 14, 2015

1.7 Million litres of crude oil released from one cracked track

A CN unit train transporting 100 tank cars loaded with petroleum products derailed. It was travelling at 38 mph, below the 40 mph speed limit in place at the time. Twenty-nine tank cars of petroleum crude oil derailed and 19 of these breached, releasing 1.7 million litres of product. The crude oil ignited, resulting in fires that burned for 5 days.

The investigation found that the derailment occurred when joint bars in the track failed. Pre-existing fatigue cracks in the joint bars at this location had gone unnoticed in previous inspections. Once the fatigue cracks reached a critical size, the combination of the cold temperatures (-31 °C) and repetitive impacts from train wheels passing over the joint caused the joint bars to fail. These defects went undetected as the training, on-the-job mentoring, and supervisory support that an assistant track supervisor received was insufficient.

More Info

With 46,522 km of rail in Canada alone, rail operators don’t have the manpower to monitor our existing tracks.

The Safe unit is a remote IoT device that rides ahead of trains, using its array of onboard sensors to monitor tracks for damage and obstacles.

Upon detection of a hazard, the unit will alert the train within a safe distance for engineers to stop, evaluate the risk, perform applicable repairs, and safely pass over or avoid the problem track

Key Features


The Hack

Due to the size and scale of this project, we’re not building a hardware prototype this weekend. Instead, our hack opts to mock the hardware and focus on the circuit design in VHDL, create the hardware library and the data processing and ML pipeline

Hardware Design Software Design Github