Sundeep Dayalan
India has an unsolved problem of handling the traffic congestion wisely
Key highlights
Won the title on Smart India Hackathon in 2019
Smart India Hackathon is a nationwide initiative to provide students with a platform to solve some of the pressing problems we face in our daily lives, and thus inculcate a culture of product innovation and a mindset of problem-solving. The problem of solving the traffic congestion, enabling smooth ambulance flow through busy roads, detect potholes and map them in a dedicated navigation map to avoid accidents was given by leading car manufacturing company Maruti Suzuki. Students from all over India proposed a solution to the particular problem statement and only a few teams make up to the finale with rigorous filtration and examinations. With our solution, we reached the top 15 contestants in India. Our team has been invited to Banaras Hindu University, Varanasi, India to convert the idea into a product. After multiple rounds of evaluation, our team has officially announced as the Winner. This problem solution is nationally recognized by the Government of India and the Maruti Suzuki organization.
₹ 1,00,000 cash award
Maruti Suzuki along with the Government of India issued a ₹ 1,00,000 (One lakh) cash award for developing a robust and innovative product for the problem statement they provided to us in Smart India hackathon 2019.
Published and telecasted in various news mediums in India
This innovative product solution is recognized and telecasted/published in various News channels and Newspapers in India. Click to view
Patent
This entire problem solution and the product are registered under the Ministry of Commerce ad Industry, Government of India(Application number: 202041024036). Hence, the entire concept of this problem solution is provided below for educational and research purposes.
Jury
Abhijit Bora
Deputy General Manager at Maruti Suzuki India Limited
Nikhil Madan
Senior Manager at Maruti Suzuki India Limited
Our team
Introduction
A Dynamic approach based on traffic Density is much more feasible and efficient. Our approach is to let signals handle themselves based on various real-time factors such as Traffic density, Emergency vehicles, etc. The Traffic analysis module uses an already provided surveillance camera to analyze the parameters by using Artificial Intelligence and optical character recognition through CCTV cameras. Each lane is separated and distinguished into three or more segments according to the scenario. These lines are used to determine the traffic density. Whenever the vehicles hide a part of the line in a lane, we can estimate the Traffic density with respect to a time where the camera frames correspond to time by this manner, we can generate real-time traffic data and implement a time sequence for the signal to clear off the traffic. The scope can be further widened by co-operating this system with INTELLIGENT TRANSPORT SYSTEMS. The System is capable of learning from the processed datasets. The advanced structure provides an embedding of a GPS module in the vehicle which will be able to identify rogue vehicles and can also minimize the theft of vehicles. The module with the GPS sensor makes a good accurate score on the prediction. For the implementation of this system, we use video feed from the CCTV cameras and detect the amount of traffic density in each lane using OpenCV. Instead of Fixed time updating of signals, we used a Dynamic approach based on the traffic density. Whenever an emergency vehicle comes within the range of the signals, it automatically detects it and sets the green signal for the vehicle to pass swiftly and also alerts other vehicles through a mobile notification on the way of the ambulance. A console is developed to detect any potholes, accidents on the roads and marks them on the map for the convenience of the users.
Problem statement
The INTELLIGENT TRANSPORT SYSTEMS (ITS) has already been deployed in various developed countries but the cost of implementation and maintenance of the system is very high. The prioritizing of the signal for the emergency vehicle is still to be addressed in many of the developing countries because the cost of maintenance and implementation becomes high on tracking for every vehicle. In our country, every day 1214 road crashes occur, and most of them are due to traffic violations. This issue needs to be addressed immediately in our country. Potholes are also a serious issue that needs to be addressed in our country where 3597 were killed due to potholes in 2017.
Problem solution
Our proposed solution first addresses the problem of cost-efficient and standard systems. We developed a device called STC (Smart Traffic Controller) which can be integrated with existing traffic signals by which traffic lights will be triggered based on the traffic present in the road. Instead of detecting all the vehicles on the road which deals with an enormous amount of data to process and manage we will simply write words (Slot 1, Slot 2, Slot 3) horizontally on the road, mark them up with slots, we will use OCR along with image processing to scan the letters written horizontally on the road (refer below figure). The presence of vehicles will hide those horizontal wordings based on that we will measure the density of that lane. We separated density into three levels from levels 0-3. Each level will have actions and according to that traffic signals will be triggered with the help of the STC (Smart Traffic Controller) device. The potholes can be detected using the OpenCV and train them with the data then detect the potholes and indicate the user of the potholes so the user can travel carefully. Emergency vehicles (Ambulance, Fire Engines) get tracked down as they are fitted on with our INS (Intelligent Navigation System) module where once the vehicle nears the signal, the signal will be automatically changed to green to allow smooth flow of Emergency vehicles. This will be done dynamically when INS syncs with STC. It allows the traffic to flow through the junctions smoothly and there won’t be any congestion caused due to unnecessary wait times in the junctions. The CCTV camera in roads along with the modules fitted within the vehicles allows reducing traffic violations and other crimes related to it. It also detects incoming emergency vehicles and allows them to pass through the junction without waiting which helps save lives in timely situations. It effectively reduces travel time, travel cost, air pollution, and accident risks.
What we developed
1. Traffic Density Calculation
The density of traffic will be calculated by detecting the horizontal words present on the road using Tesseract OCR and Image Processing. This can be done with the help of CCTV present in the roads(If not, need to be installed). The CCTV data will be sent to server and density will be calculated after processing the data. Then Density value will be sent to STC device and it will trigger traffic lights according to that.
2. Intelligent Navigation System(INS)
We developed a separate navigation system for Emergency vehicles by which smooth flow of emergency vehicles can be achieved. This system provides navigation to the accident location in a single click along with it simultaneously tracks the location of ambulance location in the background. While moving INS dynamically searches for nearby traffic signals where the STC device will be installed in it (refer below image). When an ambulance enters the radius of a traffic signal, the INS and STC will be automatically synced and that lane signal will be turned to green by changing all other signals of the lane to red. All this process will be done independently without any interaction with the ambulance driver. By which no emergency vehicles will have to wait in traffic.
Not only for emergency vehicles, normal vehicle users also will have separate INS (refer below image)for their dashboard navigation where the traffic signals are marked in the navigation map along with the lane traffic. The status of the traffic signals will be dynamically updated in this navigation system. By this public can reroute based on the traffic and it will decrease the waiting time in traffics considerably. Even the nearby emergency vehicles will be shown in INS for global users.
3. Traffic violation and Pothole detection
The vehicles that violate the traffic rules get detected from the camera as if the signal is red it scans for the STOP line (refer below image) and if any vehicle crosses that vehicle is flagged and the OCR takes the license plate number and stores it on to separate child in the cloud as flagged. The potholes can be detected using the dashboard camera of the vehicle using the trained model. In case of the absence of CCTV in that road, if the vehicle user utilizes the INS, then INS will throw user details to the server if the user violates the traffic signal.
Our pothole detection module will collect the data on the presence of potholes on the road. The can be achieved by integrating our pothole detection module into dashboard cameras in the cars. The presence of potholes will be detected, and the accurate location of potholes will be sent to serve dynamically. Then the server will mark the location of potholes in INS maps. So that the global users will be notified in their INS when they enter the radius of potholes (done with the help of geofencing)
4. Extraordinary Event Detection
Detection of extraordinary events such as protests, parades and other important happenings was achieved by scraping the web for news articles and headlines relating to such events. Those headlines and descriptions were then returned in a JSON format. The locations of the events were then obtained from the returned JSON object in our application, and a marker was placed onto the same location on the INS to convey the information that an extraordinary event is currently happening at that location. The scraped information is refreshed regularly to keep up with constant updates. INS will notify the users about the event so that users can reroute instead of waiting in the event locations.
5. Control Interface
This central system or also called a server will have all the datasets about the traffic lights, pothole locations, Users who violated traffic signals, traffic densities of roads, etc. This cloud database will store all the data and communicate with each STC device dynamically. Also, this interface will help traffic police to control the traffic signals remotely and block particular roads in case of special events (ex. VIP convoys). Using this interface, no traffic policeman needs to visit the traffic signal in order to control or troubleshoot it. Everything can be done remotely (Refer below image). This integrates every single module and serves the entire traffic management system. Using this interface Traffic police can control traffic signals remotely and along with that can also view the live status.
6. STC (Smart Traffic Controller)
This is a hardware device that must be installed in existing traffic signals. After that normal signals will get smarter so that traffic signals will operate automatically based on the accurate traffic present in the road. This device will be connected with server wirelessly with the help of Private Network. The traffic density value will be received from server and STC will trigger traffic signals according to that. These traffic signals will automatically fall back to traditional mode (triggers traffic lights based on fixed timings) of operation in case of any server issues (Lack of internet connectivity). Also, this device has the capability to sync with upcoming INS of emergency vehicles.
7. Emergency vehicle prioritization
Ambulance drivers using an INS upon entering a predefined radius from the traffic signals will be automatically synced to the signals they approach so that the signals turn green and provide immediate free passage to the ambulance. This is done by tracking the ambulance using the GPS module present in the console of the vehicle. When an ambulance gets near a traffic junction, the traffic light of the specific lane turns green until the ambulance passes. These are updated to a cloud system, from which the information is relayed back to the users in front of the ambulance alerting them of the incoming ambulance in INS and requesting them to give way for the ambulance. Apart from this, the users are also intimated well in advance about the ambulance’s arrival via a voice alert in their mobile phones and INS.
8. Pothole Detection
The pothole is one of the major issues in our country, pothole took a deadly toll in our country taking almost 10 lives daily with an annual fatality in our country adding up to 3597 in the year 2017. The percentage of the death tolls keep on increasing. Our solution for the pothole is we use an object detection API and a trained model to detect the pothole and map the pothole in the Google maps.
Attracting functionality
It allows the traffic to flow through the junctions smoothly and there won’t be any congestion caused due to unnecessary wait times in junctions. The CCTV camera along with the modules fitted within the vehicles allows reducing traffic violations and other crimes related to it. It also detects incoming emergency vehicles and allows them to pass through the junction without waiting which helps save lives in timely situations. It effectively reduces travel time, travel cost, air pollution, and accident risks. By implementing the whole system no traffic policeman is needed to operate every individual traffic signal. This will effectively decrease the count of traffic policemen standing along the roadside to control the traffic and traffic lights. This system will considerably decrease the accidents occurring due to potholes. Global users can dynamically view the traffic signal and density status will be very useful to decide their travel route prior to the departure. Every emergency vehicle driver can operate the vehicles without any interruption by utilizing INS
System Design and Implementation
Smart Traffic Controller(STC) Installation
This system mainly works with the help of the STC (Smart Traffic Controller) device. This device syncs with every other module to make this system fully functional. This STC device communicates with the server with the help of a private network. STC also has the capability to work independently when it failed to communicate with the server (Offline circumstances). This STC can be installed in any existing traffic lights thus it does not need any prerequisites. STC will trigger the traffic lights according to the traffic present on the road. This can be achieved by detecting the traffic using the OCR concept. CCTV cameras will be placed in the position as shown below figure and it continuously detects the traffic density and sends the data to the server. After the data processing is done it sends the accurate density values to STC and according to the density values, STC triggers the Green lights. All this process will be done simultaneously without any delays.
Camera placement over traffic signals
This is how the CCTV cameras will be installed above traffic lights. These CCTV cameras must be installed at considerable heights in order to get accurate results. The existing surveillance cameras can also be used for this system. The data from CCTV will be stored in the server periodically to allow surveillance backups.
Communication with server Flow
The process begins with the data captured from the CCTV cameras at the junctions where it collects all the information like the density of vehicles, on which lane is the high density, and sends the data to the server where all the data gets processed and analyzed. The data accumulated will calculate the density and send it to the STC. The STC modules receive the value and change the signal correspondingly. STC will also receive several other values from the server like priorities, troubleshoot instructions, etc.
Offline Mode:
Clearly, to make this system completely utilitarian, it needs a web network. Here comes a most concerning issue of communication between INC, STC, and Servers during the absence of internet connectivity and because of this, the system may glitch. To overcome this issue, we utilized Non-cellular Data Transmission and this system will be utilized automatically during offline conditions. Here the communication will happen by means of encoded SMS with the help of the SMPP (Short Message Peer to Peer) protocol. Once the network is established, the system falls back to normal methodology.
Bay Detection concept
Clearly, to make this system completely utilitarian, it needs a web network. Here comes a most concerning issue of communication between INC, STC, and Servers during the absence of internet connectivity and because of this, the system may glitch. To overcome this issue, we utilized Non-cellular Data Transmission and this system will be utilized automatically during offline conditions. Here the communication will happen by means of encoded SMS with the help of the SMPP (Short Message Peer to Peer) protocol. Once the network is established, the system falls back to normal methodology.
LEVEL 0
All the texts (slot 1,2,3) are visible to the camera and the processing unit sends the traffic density value as 0 to the STC. When STC receives traffic density as 0, STC won’t trigger GREEN SIGNAL for that particular lane why because there are no vehicles, and triggering GREEN signal will be waste of time.
LEVEL 1
SLOT 1 text won’t be visible to the camera why because it was hidden by vehicles present there. So, the processing unit sends the traffic density value as 1 to the STC. When STC receives traffic density as 1, STC will trigger GREEN SIGNAL for20 Seconds (Duration can be changed) and it was enough for the vehicles to pass the signal.
LEVEL 2
SLOT 1, SLOT 2 texts won’t be visible to the camera why because it was hidden by vehicles present there. So, the processing unit sends the traffic density value as 2 to the STC. When STC receives traffic density as 2, STC will trigger GREEN SIGNAL for40 Seconds (Duration can be changed) and it was enough for the vehicles to pass the signal.
LEVEL 3
SLOT 1, SLOT 2, SLOT 3 texts won’t be visible to the camera why because it was hidden by vehicles present there. So, the processing unit sends the traffic density value as 3 to the STC. When STC receives a traffic density of 3, STC will trigger GREEN SIGNAL for60 Seconds (Duration can be changed) and it was enough for the vehicles to pass the signal.
Hardware specification
The STC device needs 12V input power to operate. The 240V AC was given as input to the STC device. It converts 240V AC-12V DC to power the microcontroller and other modules. The Network module will be connected to a private network to communicate with the server. Relays are used to trigger the traffic lights. It has 4 channel relay (RED, YELLOW, GREEN, EXTRA- in case of failure of other channels) and these channels will be triggered by microcontroller
Conclusion
By resolving congestion at a base level, we regulate the overall traffic in a much more efficient manner. The proposed system uses already existing traditional traffic handling devices no new additional and costly sensors are required for implementation. The whole module allows the commuters to travel faster by reducing the waiting time, on a larger scale the amount of waiting time is drastically reduced. As the system has direct control over the flow of traffic emergency vehicles such as ambulances and fire engines can be easily routed to the destination with minimal hindrance due to traffic stagnation. User-based features like potholes detection update users with live pothole alerts and instantly maps them on the map to avoid future accidents. As the system works based on estimating the traffic density using bay-based processing this eliminates the system to track each and every vehicle to estimate traffic, thus eliminating the possibility of processing a huge amount of data in real-time. This allows the system to be more cost-effective and functional.
