Monthly Archives: September 2020

Inside Access: AIS 140 Software Stack Development for Telematics Applications

The Delhi Nirbhaya case in 2012 shook the conscience of the whole nation. Since then, the government of India has been actively taking measures to ensure passenger safety in commercial and public transportation. In 2019, the government mandated the adoption of AIS 140, a key regulation which ensures that all commercial vehicles can be tracked through GPS devices fitted in the vehicles.

AIS 140 was framed by the Automotive Research Association of India (ARAI) and Automotive Industry Standards (AIS) under Intelligent Transport System (ITS) requirements. We have given you an overview of AIS 140 standards in one of our previous articles.

The AIS 140 standard, apart from being applicable to the overall  system environment of a public transport vehicle, is also relevant to its individual components. This standard has been seeing increased adoption in the Indian telematics domain. A few years down the line, every truck or public transport bus in India would be fitted with a GPS device that enables telematics, i.e., vehicle tracking, navigation, SOS features, etc.

The engineers at Embitel’s Innovation Lab have been busy designing a ready-to-deploy AIS 140 compliant software stack solution that would accelerate the development of such a GPS device. Read on to understand the solution in detail.
 

AIS 140 Compliant VLT Device/Data Logger – System Overview

Standard Requirements for an AIS 140 Compliant VLT Device

• A Vehicle Location Tracking (VLT) device should be configured to obtain the position information of vehicles through Global Navigation Satellite System (GNSS).
• The device should transmit data to a government authorized server (Backend Control Server) through a communication network such as GSM or GPRS, as per the specified protocol.
• The device should have the capacity to transfer data to at least 2 different IP addresses.
• The system should be designed to store data in its internal memory (up to 40,000 log messages). In the event of unavailability of the communication network (GPRS), the device should transfer this data through SMS to the preconfigured numbers. The history packets will also be transferred when the network is back, in LIFO sequence.
• The device should have a unique identifier and an embedded SIM.
• The system should support standard minimum input and output, i.e., 2 analogue, 4 digital, and 1 serial communication for interfacing external systems.
• The device should transmit the position of the vehicle, velocity, time, and direction of travel to the server at a configurable frequency.
• The system should have Emergency alert buttons that send SOS signals to the pre-configured IP addresses. These buttons will be spread across the vehicle at accessible locations for the passengers and the driver. So, if the passengers or in-vehicle crew presses the emergency button, an alert will be sent to the server in real-time.
• The device should have a backup battery that enables it to operate for 4 hours continuously.
• The device parameters and firmware should be configurable Over-the-Air (OTA).
• The system should have a 3-axis accelerometer and a 3-axis gyroscope for detecting harsh turns, braking and acceleration.

Apart from the above, there are several hardware requirements such as operating voltage range, sleep mode current specifications, waterproofing, and more. Our software stack solution is developed assuming that all the hardware requirements are met.

Let us now take a look at the software stack solution.
 

Key Features of Our AIS 140 Compliant Software Stack

The AIS 140 software library designed and developed by our engineers brings about significant reduction in automotive ECU development time. Some of the key features of this stack are as follows:

• This software library helps in constructing the standard frame format needed to transfer data to the server.
• It helps in analysing the commands received from the server, parsing them and storing the modified data.
• It also calculates the CRS (32 bit) required for data transmission.
• The stack is flexible enough to allow the customisation of frames. It is possible to easily modify the field position, start character and end character. This flexibility enables the stack to be deployed in multiple projects easily.
• It enables driver behaviour detection. Activities such as harsh turns, braking and acceleration are recorded.

The testing team has been iteratively testing the software and fine-tuning it to be error-free, over the past few months.
 

Challenges Faced in the Development Journey

Throughout the software development life cycle, there have been some challenges with respect to hardware availability. These have been mitigated to a large extent through the dedicated efforts of the team.

It is with great pride that we now introduce this solution to you. Our solution can be used for the software configuration of AIS compliant Vehicle Tracking Systems (VTS) across India in a few weeks’ time, and we couldn’t be more excited!

Kudos to the development and testing teams that worked on this project!

Inside Access to DriveSafe: A Driver Distraction Detection App Powered by Machine Learning

A closer look at the motor vehicle accident statistics around the world shows that driver distraction and fatigue are some of the major risk factors to road safety today. As road traffic injuries and death have been showing an incremental trend over the years, it is high time we address this issue.

The source of driver distraction, in most cases, is within the vehicle itself. Activities that distract the driver include texting or talking on the phone, reaching out to the back seat, playing music on the infotainment system, etc. while driving. With the increase in the use of smartphones and hands-free devices connected to phones, cognitive distraction that affects driving behaviour is also on the rise.

Project DriveSafe – Real-time Driver Distraction Detection App

The Innovation Lab at Embitel has been buzzing with activity as our engineers tried to develop a solution that keeps a check on cognitive distraction, i.e., activities that take a driver’s mind off the road.

Our work on this project was initiated at the beginning of 2020 when we designed a Python-based machine learning algorithm that analyses driver movements, identifies each activity, and sends alerts based on the results.

The next step was to identify the most optimum end-user application that would relay these messages to the driver.

The pandemic lockdown ceased to deter our enthusiasm to see this solution through to the end, and we came up with a great idea!

We decided to develop DriveSafe, an intuitive Android-based app that would use the existing hardware of the user’s mobile app to assess their attentiveness while driving. The app that was designed captures images of the driver at pre-defined intervals to act as input to the machine learning algorithm within.

Key features of DriveSafe:

• The intensive training given to the algorithm ensures that it identifies the driver’s activity with a great degree of accuracy.
• Currently, we have identified 16 activities that the driver could be engaged in. This includes 15 driver distraction activities such as using a mobile phone for calls, texting, taking selfie on a phone, adjusting makeup, talking to passengers, reaching to the back seat, smoking, reaching for the audio system to play music, and more. The focused activity identified is the driver looking at the road while driving.
• The activity is analysed through the hand movements of the driver as well. Hence, the camera should be positioned at a 45-degree angle to the direction in which the driver is facing.
• The driver can pre-configure the app settings so that his/her images are captured at periodic intervals throughout the journey. This interval for photo capture can be configured up to milliseconds.
• The user can also configure the resolution of the images captured. Additionally, he/she can configure whether all images captured need to be retained on the device after processing.
• After analysis of the images, if the algorithm finds that the driver is engaged in a distracting activity, it sends a visual notification on the phone in the form of a text image. It also sends an audio notification to alert the driver.
• The solution can also be customised to integrate a cloud server to which the alert messages can be sent. Integration of this functionality could open up a wide range of business use cases for which this solution can be deployed.

Business Use Cases for DriveSafe

The driver distraction detection app conceived at Embitel’s Innovation Lab can be the solution for a large spectrum of business use cases:

1. Driver assistance – When there is a need to offer some assistance to the driver to help him/her stay attentive on the road, DriveSafe would be the perfect solution. While using this app, if the driver is distracted, they receive a real-time notification that helps them focus on the road and drive safely.
2. Insurance companies – DriveSafe can be beneficial to insurance companies in identifying the primary reason for a road mishap that precedes an insurance claim. If there was an accident, the insurer can easily analyse the driver activity at that point of time to take a judicious decision.
3. Drowsiness detection – Currently, the app performs image analysis using pictures captured from the mobile camera positioned at the side of the driver (45 degrees from the front of the driver’s face). For detecting driver drowsiness, it is necessary to install the camera at the front of the driver. When the camera is at the front, it is possible to analyse the eyes of the driver and detect drowsiness. The DriveSafe solution is scalable to include driver drowsiness detection feature as well. The same algorithm can be used for this purpose; only the model needs to be trained with different data to accomplish this feature.
4. Cab aggregators – The DriveSafe app helps cab aggregators track their drivers’ behaviour for streamlining operations. The driver activities such as smoking, texting, excessive use of music system, etc. can be of interest to such companies for this use case.

Data Security

Currently, all computations and predictions are performed locally on the app. There is no need for internet connectivity for app functioning.

In the future, if a module is integrated to connect to an external server to transfer driver alerts for analysis, the data will be stored locally and transmitted to the server when there is connectivity. Hence, there will be no data loss or threat to security from this architecture. In such a scenario, the security aspects of the solution will also rely completely on the network security.
 

Overcoming Challenges

This project was sparked by a unique “Eureka moment” and fuelled by an undying urge to craft a solution quickly. As with such kind of projects, we had to face some challenges along the way. One that we are particularly proud of overcoming is the sudden change in operations when we adopted a work-from-home policy during the pandemic lockdown.

Today, we are thrilled to announce the successful completion of the development activities of this project, as planned before the lockdown. This is a feather in the cap of the engineering team that worked on this project – and clear evidence of their dedication and passion to break barriers and deliver exciting solutions for the future!