Radar-powered Perception for All-weather Train Operation

NIART collaborated with Alstom Netherlands to successfully demonstrate the highest level of Autonomous Train Operation (ATO), Grade of Automation 4 (GoA4), on a shunting locomotive. This project focused on validating the perception and safety requirements for fully driverless operations within complex rail yard environments, paving the way for increased efficiency and flexibility in shunting maneuvers.
The Problem
The primary challenge in achieving certified ATO GoA4 is ensuring unwavering, all-weather obstacle detection at sufficient range. Traditional sensor systems (like visible-light cameras) are severely compromised by adverse conditions common to rail yards, such as dense fog, heavy rain, snow, or intense sun glare. To meet the stringent Safety Integrity Level (SIL) required for driverless operation, a system was needed that could reliably detect and classify unexpected obstacles (e.g., people, vehicles, or misplaced rail stock) from a distance that guarantees a safe stopping maneuver under any environmental condition.
The Solution
NIART provided an advanced Obstacle Detection System (ODS) rooted in multi-spectral sensor fusion. The solution integrated a high-resolution digital radar—which penetrates adverse weather—with multi-spectral electro-optics (including visible and thermal imaging) for robust object confirmation and classification. This approach utilizes an intelligent fusion engine to combine the data, overcoming the limitations of any single sensor modality. The system was directly integrated with Alstom’s ATO platform. In demonstration, the autonomous locomotive successfully detected and reacted to various unexpected obstacles (including a person, a car, and a rail wagon) from up to 500 meters away, confirming the necessary range and reliability for truly safe GoA4 shunting operations.