In this project we propose to build smart vehicle monitoring and assistance systems using cloud computing in vehicular Ad Hoc networks. Increasing number of on road vehicles has become a major source of unintended accidents. Developing intelligent transportation system using cloud based connected vehicular networks can provide better estimate of time of arrival and localization matrix of vehicles in a given range of interest. We propose to build a safety enhancement feature to cater to accidents that occur due to random stoppage of vehicles and random opening of doors. Sensors will be used to detect and estimate events related to stoppage and door opening and create an event token. This event token will be correlated with the time of arrival and localization matrix of vehicles in a given radius of the vehicle from where the token was generated. Based upon this correlation a warning system and door enable disable system will be implemented to avoid collision. Additionally, temperature and humidity sensors will be deployed inside the vehicle to automatically or partially open the windows for maintaining proper air flow and ventilation.

Project Proposal

1. High-level project introduction and performance expectation

One of the lesser addressed concerns in vehicular safety is the aspect of “Dooring “. Dooring occurs when the doors of a vehicle are inadvertently opened either when the vehicle is stationary and the passengers choose to alight or the doors open while the vehicle is moving. Dooring often causes accidents due to lack of warning both to the driver of the vehicle whose doors are opened and the drivers of the vehicles who are in close proximity.

For safe navigation and prevention of Dooring we propose to develop an intelligent tracking system based on cloud computing which will alert drivers of vehicles to this possibility. Sensors mounted on the vehicles will be used to detect abrupt deceleration coupled with probability of opening the doors while simultaneously tracking the trajectories of proximal vehicles.

FPGAs are preferable for developing devices that utilize image processing and various sensor fusions. We need real-time processing for all the inputs that require a high computing power device like the DE10 Nano kit. It consists of an 800MHZ Dual-core – Arm cortex-A9 processor, which is feasible to complete all the required tasks.

A data acquisition system implemented on the FPGA will acquire data from proximity sensors, cameras and accelerometers mounted on the vehicle. Information regarding deceleration will also be obtained from the ECU. After noise filtering data will be uploaded to the cloud using Microsoft AZURE. In the cloud a linear estimation and tracking filter will be used to track the trajectories of proximal vehicles in a window of interest.

A learning algorithm will be used to predict the behavior of door opening when the vehicle under consideration stops. This behavior model will be used to alert the drivers of all the vehicles in the window of interest against the possibility of Dooring. Additionally, humidity and temperature sensors are placed in the vehicle to detect if the temperature exceeds the threshold limit and thereby take the necessary steps to provide care to the passengers by maintaining proper airflow and ventilation.

2. Block Diagram

3. Expected sustainability results, projected resource savings

FPGAs are preferred due to its following characteristics:

• A FPGA can easily be customized to an embedded system.
• Unlike processors like GPU and CPU which require an OS as part of the software and memory management a FPGA doesn’t require the above and replaces it with the hardware which boosts the system performance.
• FPGA can execute in parallel, that provides to take information from all the sensors at the same time and increases the speed in processing the sensor data.
• The FPGA comparatively consumes low power when compared to a processor.
• An FPGA can work on image processing and neural networks at a faster pace compared to a microcontroller.

4. Design Introduction

5. Functional description and implementation

6. Performance metrics, performance to expectation

7. Sustainability results, resource savings achieved

8. Conclusion