A test train ran on a track where silhouettes of people, representations of animals in foam and resin, as well as a dummy rock were displayed. And this was done both day and night in order to vary the lighting conditions. The reason for this? The lineside signalling reading and obstacle detection project (DOS), a major stage in the Autonomous Train project since the end of 2017.
It concerns proposing a lateral signalling reading and obstacle detection system, faculties usually reserved for the driver. “The automation of the observation functions of the driver is a significant technological issue to deal with for the Autonomous Train,” says Tristan Vandeputte, SYSTRA’s Innovation Director.”
“These tests bring a new technological component to the programme. For the time being the train still has an on-board driver, but the aim is that in the long term the train will be able to drive itself in complete safety without on-board human intervention.”
Hugues de Goësbriand, SYSTRA’s DOS programme manager
An on-board technological arsenal
The online test phase made it possible to test the sensors and obstacle display up to 1,000 metres away, regardless of the environmental conditions. In order to ‘perceive’ the track, the test train is equipped with two colour cameras, two infrared thermal cameras for night tests and two lidar sensors, which allow the outdoor environment to be recreated in 3D as a point cloud.
These tests also allow data to be collected in parallel in a situation where the driver gradually brakes until the train comes to a complete halt 150 metres from the obstacle. “Once the tests were completed, we analysed the data,” continues Hugues de Goësbriand. “All the sensors worked. A few adjustments still need to be made, but this is excellent news. This has enabled us to move on to testing the algorithms in the laboratory, with virtual driving tests where we improve their behaviour as we go along.”
The next stage will take place in April 2021, with the delivery of a prototype obstacle detection system for short distances (300 metres) and then, the following year, a second prototype for long distances (1,000 metres), before a 100% autonomous freight train prototype is put into operational service in 2023.
A long-established area of expertise for SYSTRA
“We have been supporting the SNCF group in its Autonomous Train programme since its inception in 2016. This is a flagship project and a precursor in global railway engineering, which enables us to remain at the forefront of the subject for our clients and on the lookout for new technologies or trends for the sector. It’s a logical step for SYSTRA, which has also been a forerunner in CBTC for the metro since the 1980s and is contributing to several autonomous transport projects, both in France and in Singapore.”
Thomas Bruel, Innovation Leader at SYSTRA