Lane Detection and Object Collision Avoidance(ADAS)

Project Title

Lane Detection and Object Collision Avoidance(ADAS)

Project Area of Specialization Computer ScienceProject Summary

ADAS is a smartphone based driver assistance system that makes use of rear camera of phone to capture real time video and process it to provide assistance to the driver of vehicle. The application is developed on Android Studio. According to the Pakistan Bureau of Statistics data, 15 persons die daily in car accidents. We can provide some help to assist driver to make his journey safe. Auto companies are offering minimal expense sensor abilities included diminished computational expense; assist both examination foundation and business explicit association with the project to bring the possible future utilization of their work into this area in proceeding with progression in transport framework. There are already some applications in the market, which provide the purpose already, but accuracy is still not achieved. They have limited functionality and less effective in identifying the vehicles in different weather conditions. The proposed solution will aim to improve the accuracy and efficiency in existing systems using image processing techniques and technologies, canceling the need of physical sensors that are expensive too. The proposed system will give effective progress in the field of ADAS. Pakistani car industry has recently started progressing, as far as the integration of technology into vehicles is concerned. Although companies such as Hyundai, MG, and Alsvin are making some headway in this realm, the industry has a long way to go before ADAS can be normalized here. There are several infrastructural issues with Pakistani roads that do not allow autonomous driving technology to function properly. The roads lack proper markings, which inhibits the system’s capability to read them and react accordingly. The local traffic conditions are confusing enough for the system to identify, and motorists tend to violate the road rules inappropriately, which could also become hard for a computer to process. Given these concerns, the system will work on providing solution that will try to achieve some level of accuracy and making journey safe for everyone.

Project Objectives

Vehicle detection and tracking applications are under research for many years now, the main objective of this field of study is to investigate vision based intelligent systems that can be used to extract useful and precise information about roadways and vehicles. In the domain of digital image processing number of techniques are being used to process images in order to extract information. After extracting the data from the images, this data can be used to make decisions. Different operations can also be applied on the extracted data. Monitoring roads and its surrounding by smartphone camera can give us useful advantages for road safety and driver assistance. Although it is an emerging field of study these days but as the growth in the development and research of digital image processing looks likely to continue, this technology will become increasingly more affordable and easier to use. The future seems rich for road safety and driver assistance applications. The main goal of lane detection is to detect the lane and forward vehicle then warn the driver for lane departure or perform required action. First it will suggest keeping vehicle within lane. Then it will detect next vehicle and will determine the distance of next vehicle. The safe distance should be 10m or above, otherwise it should warn the driver. Moreover, while keeping lane track, system should assist steering advice in case car leaves the lane.

Project Implementation Method

The proposed solution is aimed to improve accuracy in existing system that is not efficient enough to meet market goals. Making use of image processing via camera capture will cancel the need to use high cost physical sensors like lidar, radar or ultrasonic proximity sensors. The pre-trained object detection model will be combined with image processing techniques like Hough transform and Canny edge detection to make use of road markers for detecting lane, object (vehicle, person or cat) and measure distance to keep driver in safe distance range. The software development lifecycle used is incremental model. First things go first, modules to be tested on backend one by one. Next, app interface developed separately on android based platform. Once all modules are working well, it’s time to integrate them with the app interfaces to make it user friendly and effective. The app can allow user to sign in or to register new user, making features enable for user.

Benefits of the Project

Car industry has some severe and hard limitations in their development driver help framework applications in the terms of dependability, cost viability and ongoing execution. Auto companies offering minimal expense sensor abilities included diminished computational expense; assist both examination foundation and business explicit associations with the objective to bring the possible future utilization of their work into this area in proceeding with progression in savvy transport frameworks. Pictures or casings extracted from camera module mounted in vehicles give huge data and information about the vehicle’s flow position and general climate which should be investigated to separate important data or object to help continuous development driver help frameworks. If we detect two lanes, we will acquire center of the screen area close to between two lanes so that we keep ourselves in the middle of them ignoring rest of the unconcerned objects like trees, street bench or lights. We will define a threshold to get a binary edge map. Then the image will be defined into blocks and classifying each block as lane mark. Then we will identify lane by using an algorithm to compensate perspective view. The low-high intensity patterns will search for the aimed lane along images. Then for vehicle distance, pixel ratio will be calculated with the screen area. If the forward vehicle covers 10 percent of the screen, then it is safe. If car pixels cover more than 30 percent area of screen that means trouble and alarms should be activated.

Technical Details of Final Deliverable

The final deliverable of “Lane Detection and Object Collision Avoidance " will be an android based application along with documentation manual having all technical details of the application such as Prototyping, Document Analysis, Interface Analysis.

Prototyping is the technique that provides a mock up of the software to be implemented. It renders the partial implementation of the requirements stated by the end-user/consumer, but doesn’t involve actual functionality of the system. It helps the developer to understand what actually he is supposed to develop and provides good understanding to consumer also, as consumer is able to visualize the impression of actual system required and can add/remove features before going towards next level. To develop ADAS app, this technique may be useful to validate the features and making interfaces.  It will be developed in Java along with Firebase database. We will develop a python's API that we will integrate with our android application. The application will be compatible with android API version 21 and above.

Starting with the documentation available about the intended software. Going through research papers and documents can help to make process well organized and refined. It can help the analyst to properly map the requirements with the proposed system features and functionalities. There has been lot of research available to get a sneak peek about autonomous driving assistance system that can lead to reach the goals effectively, resulting in good understanding of the intended system. Interface Analysis is an analysis elicitation technique that helps to identify interfaces between solutions/applications to determine the requirements for ensuring that the components interact with one another effectively. The nature of ADAS app involves switching between different interfaces and platform, that’s why it will benefit to map app efficiently and to increase mobility.

Final Deliverable of the Project Software SystemCore Industry ITOther Industries Others Core Technology Artificial Intelligence(AI)Other Technologies RoboticsSustainable Development Goals Industry, Innovation and InfrastructureRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	50000
Android device	Equipment	1	50000	50000