“A Good long ride can clear your mind, restore your faith, and use a lot of fuel”. As of the most recent report by the Department of Transportation, there were 8,410,255 motorcycles registered in the United States by private citizens and commercial organizations in 2011. To put this staggering number into perspective, out of every 36 people you meet in the U.S., one of them probably has a motorcycle. We, the developers of Reverever wanted to fulfill the dreams of millions of motor-bike lovers and people who are fascinated about a relaxing long drive with a holistic user experience that enhances the very concept of long drives.
Contrary to the orthodox belief that, driving safe requires you restrict yourselves and focus seriously on the job, we believe that people can drive safe and enjoy the fun that comes along with a drive.
What it does
We do this, by providing the user, with a pleasant experience of driving in the most beautiful routes of the nation which boasts about 33% of green and fresh land that comes along with breathtaking scenic views ( http://data.worldbank.org/indicator/AG.LND.FRST.ZS ). We strongly believe that passionate motorbikers do deserve the opportunity to relish the Mother Nature. We aid them by crowdsourcing a set of picturesque locations and geotag them to provide them with a calm, serene and sublime routes. And guess what? The drivers don’t have to do anything extra! We take care of capturing breathtaking pictures of the places, whenever the driver stops. We further, characterize the image and provide tags, weather and other extra features of the place. We further pool out the temperature, breeze and bumpiness of the route of travel to provide greater amount of information, well before a planned ride, coz C’mmon, people would love a plethora of options to choose from.
Go ride! Drive safe! And do not miss out on loads of fun it comes along with.
How we built it
We have a state-of-the art multidisciplinary team, that strives hard to learn and innovate with new stuff. We used the Raspberry Pi interfaced with a Camera, that snaps a picture whenever a vehicle stops, processes it using the Microsoft Cognitive Services to provide a detailed description of the place along with a captivating caption. We use the confidence of the information retrieved to provide fruitful outputs in the form of weather conditions, conditions for driving, traffic and various other real-time parameters to the user.
The image captured using the Raspberry Pi Camera is first uploaded into an AWS S3 Bucket. A Lambda function will be triggered with insertion of image into S3 which will query the database to get the specifics of the image. This function obtains the characteristics of the image using the Microsoft Cognitive Services which describes the scenic beauty of the place. In addition to this, in case a Speed Limit Board is encountered, its characteristics are provided too. We are also using MapQuest APIs to obtain the location. On fetching the necessary information, we log them into a MySQL Database running on AWS RDS Service.
Further, As an extension to our product, with a strong societal value, we believed that, the vision and speech, though considered as core senses, are not the only way to experience the wonderful nature and decided to put a smile on their faces. We developed a prototype that provides an immersive experience to visually impaired people to feel the fresh air on their faces and listen about key aspects of the place they are in during a long drive whenever they wish. We did this by using a high performance NVIDIA TX1 GPU, that captures images on demand and provides voice-over description of beautiful and much needed aspects of the place.
Challenges we ran into
Hardware Integration issues. GPU interfacing Issues. AWS SQS, Lambda-RDS integration
We spent a lot of time trying to do a remote access into the DB instance created on RDS Service. This was necessary to create the table and to log in image related data. We deciphered the issue to be in the Security Groups configuration while creating the DB Instance.
We also faced a couple of issues while installing the Python libraries necessary for successful execution of the code in AWS Lambda. Further, the crowdsourcing simulations involved a lot of complications.
Accomplishments that we're proud of
Getting the Image capture performed on Raspberry Pi and passing the image over AWS S3, AWS EC2 and AWS Lambda Services as well as NVIDIA GPU interfacing. Geotagging, with Chat Bot Implementation and the MAP-based WEBGUI. Finding Picturesque Locations. Crowdsourcing implementations
What we learned
We learned a crazy lot from HW as well as SW perspective on the Above topics
What's next for Reverever
We are Planning to expand with improvement in CUDNN Models for improved Machine learning