Safety Device for Railway Track Workers
Introduction:
Our main objective was to make a portable device which will warn the construction workers working on train tracks about approaching trains through a hooter. We were given some constraints to work with, like:
- The device should be placed 2 to 3 kilometers away from the workers.
- The device should not be placed on the railway tracks.
- No sensing should be done on the tracks or on the locomotives.
Vision:
“To build a product which ensures the proper safety of railway track workers by notifying them about approaching train. The same idea can also be implemented at unmanned crossings that could prevent numerous accidents.”
Finding the proper sensors:
After discussing with Prof. M Balakrishnan, we came up with the idea that we will be using two methods to confirm the presence of a train. The first method would sense the train and the second method will confirm the presence of the train and the added benefit will tell us the direction of the train.
Keeping our constraints in mind, we decided upon using audio processing as the first method to detect the presence of the train. For the secondary sensing, we decided to try a few methods and and tried to come up with the most suitable one.So we studied and tried ultrasound, laser, and image sensing methods. At last, we decided to use ultrasound sensors because it was cheap, easily available,reliable and easy to code.
Our mentor Mr. Vikas gave his invaluable advice which helped us in choosing the microprocessor for our device. Raspberry Pi met our needs. We used a short-range USB microphone for recording the train sounds. HC-SR04 ultrasound sensors were used for the second method. For communication over 1–2 kilometers, we decided to use RA-01 LORA Modules modules.
Final Set of Equipment used:
Our Journey:
When we first got the project, we discussed various ideas for obtaining the sound samples of the trains. As the sound data for Indian trains was not available on the internet, we decided to collect our own data. For this, we visited the Sarojini Nagar Railway Station.
On our first trip to the Railway Station, we collected about 15 sound samples using the microphones present in our mobile phones. Upon analyzing those sound samples we figured out that the analysis will depend on which microphone is used, did the microphone face the train while recording and so on. This was our first roadblock. Then we figured we will need to record the sound samples with the same microphone and in the same orientation of the microphone which will be used in our final device.
Now that our collected data was obsolete for the time being we decided to complete the ultrasound part of the project. Taking advice from our professor M. Balakrishnan, we decided to keep both the sensors at some angle so as to completely determine the object as well as the direction of it. It was successfully tested in our hostel rooms and the only thing left was to do the field testing of the device.
The objective of our second trip was to collect as many sound samples as possible with the USB microphone which we purchased to be used in the device. We collected about 50 samples and then analyzed them .The results thus obtained were satisfactory . Another objective of this visit was to field test our ultrasound sensors which was successfully done.
Now that our ultrasound part of the project was complete, we began working on the audio part. For train detection, in addition to the train sound samples, we needed sound samples from traffic, birds chirping etc for our code to work accurately. So collect these samples we made a short trip to IIT Hostel gate and recorded the traffic sounds.
Once our device was set up we had to again test it and resolve any errors we get. So we made another trip to the Railway Station. The device was successfully tested on field with Bluetooth Module(Communication Component)which had to be used for Open House Demonstration. Lora RA-01 Module can be installed instead of Bluetooth Module for long range communication.
The final packaging of our device was done so it becomes portable and compact.
Challenges:
We believe that there is always a scope of improvement in any project and the path to that improvements is never too easy. The device design has to be looked upon as it needs to sustain different weather conditions. It can be made more robust so that number of false alarms can be dragged to nil. The delay associated needs to worked upon as it must be able to detect the presence of very high speed trains and signal the worker as soon as possible.
Future Prospects:
During the course of our work, we identified certain applications where our device or its modified version can be implemented for similar purpose.
Firstly,Unmanned Crossings is one of the major issue faced by the Railway Board of India. Many lives are lost every year due their presence but still 3479 such crossings exist in the country. We propose to install an alarm at these unmanned crossings which gets triggered once our device detects the presence of a train 1–2 km away from the crossing.
Secondly, a lot of wildlife casualties especially elephant deaths, are rampant in remote areas across the nation. Modified version of the device can be used to alert the driver about the presence of elephant on the track or some kind of alarm can be installed at at specified locations to shoo away animals whenever their presence is detected in the vicinity of the track.
Link to Github Repo:
Demonstration at Open House:
Open House is the annual exhibit of an extensive collection of innovative research and product development projects going on at IIT Delhi. We also presented our project at the Open House ’18. Since ours is a train based project, the field based presentation with real train would have suited best, but certainly it was not feasible. Hence, we adopted a dual demonstration idea using a video shot on field and a live demonstration using toy train. The field video was shot at Sarojini Nagar Railway Station, where our project actually did come up from a nascent stage by various test visits there.
The live demonstration involved the use of our actual device (with minor modification in code) so as to detect a toy train. The train sound was played from a speaker and the microphone was placed in front of it, so as to complete the audio detection part.
Open House Demonstration Video:
Open House Visitor Feedback:
Contact Us:
Aditya Abhishek (ee1160435@iitd.ac.in)
Gajera Dev (ee1160431@iitd.ac.in)
Hritik Bansal (ee1160071@iitd.ac.in)
Kunal Narayan (ee1160439@iitd.ac.in)
Shubham (ch1150130@iitd.ac.in)
Under Guidance of
Prof. M. Balakrishnan, IIT Delhi
Mentor: Vikas Upadhyaya