Season 2 of the SNCF Accelerator powered by Tektos

Season 2 of the SNCF Accelerator powered by Tektos

In a previous post, we describe some of our experiences with Season 1. The applications are open for Season 2. We would like initial responses from companies by April 15th. Between the 15th of April and the 2nd of May, we will interview candidate companies and provide feedback. We will announce the selected companies by the 3rd of April.

We will start our second season at the end of April. We are looking for 4 startups to work on areas of innovation for SNCF in IoT, sensors, data analysis, converting processes to mobile and managing complex mission critical processes. The problem sets will be described below.

As in the first season, the companies selected will be awarded a contract of 5 000 Euros from SNCF at the beginning of the session to create a Proof of Concept (POC). The teams (up to 3 seats) will be hosted and trained at Tektos in the Palais Brongniart. The goal is to deliver the POC within 6 months and enter contract negotiations for a much more substantial (as in money) B2B rollout.

These are real and definite problems SNCF Reseau wants to solve. They are looking for innovative solutions from startups with which they can create a partnership. Each startup is assisted by Tektos, the Innovation team at SNCF Reseau, an ‘Angel’ from the IT department and a project manager from the client division in SNCF. There are more than enough resources to be successful.

One internal team from SNCF has been selected and will work on the management of the placement of trains in stations (more complex than you think with changes in schedules and safety considerations).

If you are interested in the new projects, please do not hesitate to e-mail either me directly: or the general contact: with your comments and questions. We have an application, but we are OK with discussing it with you beforehand. We can give a very quick indication if you are a strong candidate and explain what exactly we are looking for in candidates. We do not want to waste your time. We will tell you if there is little chance of success before you dedicate too much time to the application process.

What we are looking for in Season 2

We have two clearly defined problem sets and three very open problem sets more concerned with innovation.

Project VUT

One project is Voie Unique Temporaire (VUT) or temporary unique track. This is a temporary condition where trains run in both direction, just temporarily, on the same track. This might happen due to repairs, breakdowns, obstructions or other problems.

There is currently a manual system to handle the situation. However, manual is not always the best. Especially when the staff involved have a phablet and data connections which could be used to automate the management of VUT. In the first instance, just automating a mission critical process will be enough for the Proof of Concept. There are many other ways this could be made more efficient and safer with the inclusion of sensors and intelligence.

Project Incident (REX)

In a system as complex as the tracks, switches and indicators for SNCF, there will be incidents which need to be reported. Very few of these incidents are life-threatening. However, they do need to be documented and uploaded into a database. This history of incidents is required for more advanced diagnostics. This project fits into a more global strategic vision aiming at always improving the security and the robustness of the infrastructures, where more data means a higher confidence in prediction.

Since the staff have smartphones, there should be some way to take a picture, add a few details and then upload the information, along with automatic information like geographical position, time, etc., to a central database.

Photos are key as 80 percent of the incidents are noticeable in photos. Minimal descriptive text is required.

The information about past incidents would be useful to maintenance staff when they are called out to make a repair. The photo would also identify the group responsible for the repair (is it an electrician because the power cord is severed or perhaps the signal group because the track indicator is broken). The project will interface with an internal system called REX and an internal data base of the infrastructures.

The POC should automate the process while later developments can work on obtaining information from the data stored by the mobile application.

Since the goal is to get very busy maintenance staff to report all incidents, the user interface (UI) and user experience (UX) is very important. At present, part of the incidents are properly entered into the database from the paper records. This system should be such a pleasure to use that 100% of the records are entered – automatically by the mobile app.

Possible candidates for Project VUT and Project Incident (REX)

In the first stages of these two projects, they require the ability to create programs which are very reliable and very easy to use. This is the perfect opportunity for a startup which is in the B2C market to use those skills on a B2B project.

I find that it is much easier to succeed with a well-defined B2B project. There is a shorter road to cash flow and lower marketing expenses.

The VUT and Incident projects would be a great project for companies who have excellent technical and design abilities yet have not found traction for their initial product. All companies in the previous season have found clients in related industries with a minimum of configuration and with an excellent reference client.

Project Locomotive Sensing

Maintenance engineers in the technical cabin of the train develop a detailed knowledge of the tracks they use. Following the same routes, over many years, they know where there is a dip in the terrain, a rough track, high winds and other factors they can see and feel. When there is something that jostles the train, the maintenance engineer in the cabin should note the location and what they observed (a shock to the train, trees close to the tracks, etc.). Another problem is the speed of high speed trains. Thirty seconds is a long time when you are traveling at several hundred kilometers per hour. When the maintenance engineer has time to look up and note the distance, it could be several kilometers behind them.

These tasks can easily be done with sensors and some minimal artificial Intelligence or machine learning. In the first instance, a Proof of Concept would simply note the sensor results and the location. Sensors would include accelerometers to measure shock or cameras to determine if vegetation is too close to the tracks. The full range of sensors would be proposed by the startup after spending time with the SNCF team.

This will definitely be a project for companies with Internet of Things experience. Doubly so for companies who are train buffs. This project entails adding independent sensing and capture units to the locomotive cab.

There is both a software and hardware aspect. The sensors, cameras and capture units have to operate at a very high speed and in real time. As the train travels at high speed, it needs to determine if the trees growing near the track need to be trimmed. So, video must be captured at high speed, then analyzed for obstructions and possible obstruction. If problems are found, the location, plus other information such as temperature, must be stored in a log for transfer to a central database.

In the Proof of Concept, it is OK to use light duty sensors. At a later stage, the sensors must be industrial grade and suitable for deployment by SNCF. The proof of concept software can use some post capture processing. The final version must calculate everything in real time.

This is an interesting project in that the goal is to ease the burden of the technician. This requires figuring out the senses and criteria the technician uses to decide if something is out of place and then using sensors and computation to duplicate it.

Project Digitalization of the tracks maintenance procedure

Maintenance of the signal on the outskirts of the tracks is the key priority and activity of SNCF Réseau. It therefore needs to be thorough and efficient. Today, maintenance is done on sheets and papers, and expertise of the agents is everything. The project here is to learn from and build upon the agents’ strong experience in order to facilitate the procedure with more communication and more data from the infrastructures to the agents, and more sharing between the agent and the engineering departments. This facilitation will go through a digitalization of the process – a mobile application – that helps agents following the procedure while focusing on their core expertise : the knowledge of the infrastructure.

SNCF Réseau is also looking for innovation in sensors, communications, Internet of things processing, connectivity and analysis. This is a broad list – and it can be very interesting to companies developing their technology and in need of a testing partner. They are looking for technologies which will improve the supervision of the infrastructure. This is different from the previous project as the sensing is done outside the train.

Some areas under strong consideration are Long Range communications (LORA), Micro-Electro-Mechanical Systems (MEMS), radar and networks of sensors which can communicate amongst themselves and pass messages to a central server. This is not an exhaustive list. If you have something useful, which we did not mention, please let us know.

If you are working with IoT and sensors, you know there is an avalanche of data, usually far from a control point and communications systems, which needs to be processed to be useful information. Do you have a solution to any of these problems using any of the technologies mentioned?

This project has a larger element of experimentation. If you have a novel application of sensors, or new innovative systems, we are open to your proposal.

Project Security in signal centers

Signal centers are highly sensitive areas and they have to be protected, including from mobile waves of the agents’ smartphones. Today, agents need to be very careful and have to let their smartphones outside of the center. Not doing so could have serious consequences. In order to secure this procedure, SNCF Réseau is exploring new ways to facilitate and to automatize the process. We could imagine tools that automatically activate the Airplane mode of the smartphones as the agents approach the infrastructure; or other ways to block the transmission of waves in these areas.

If you are working on infrastructure and data security or/and you have an expert knowledge on hardware and software that addresses this situation, you’d fit perfectly in this project. There should be other solutions to this problematic, that may come from broader domains. You are welcome to propose your ideas.