Somvra. Shadow pattern of Madrid
If you want to contaminate less, Somvra is your solution
Somvra is a project that works on predicting the best parking lots for solar vehicles by analysing shadow behaviour during the day. To do so, factors such as sun path, seasons, time, architectural pattern of the city and weather need to be taken into account.
ZunPower, a startup that is designing a solar carsharing vehicle, proposed a challenge to our team: to create a solar city map to boost solar bicycles efficiency and lower the costs. AI Saturdays, a non-profit organization on a mission to empower to learn Artificial Intelligence, has been a stepping stone for us to create, invent and grow.
To properly comprehend the task we are upon, we have made a thorough analysis evaluating the company’s activities that can be improved with AI. Our assessment finished in identifying three main areas of work:
- To create a shadow pattern
- To optimise solar battery productivity
- To create an algorithm of dynamic prices and reward for parking in sunny lots
The shadow pattern is the essential element for a solar vehicle to be sustainable and self-sufficient, so we focused on it in the first place.
Technical approach
Brainstorming period culminated with the awareness of gathering a particular personalised dataset for our project because none of the existing open datasets could have given a full perspective of the influential variables. Moreover, a deep comprehension of mathematics areas such as trigonometry, geometry and topology were required to elaborate a model for calculating the shadows.
Our first method consisted in using data from land registry to obtain the height and shape of the buildings in the form of coordinates. To scale up the data collection process, as a second option, we have applied satellite measurements.
The idea itself ensues the following steps:
- First of all, we have implemented a mathematical approach to the dataset to obtain a point where the shadow thrown by the buildings ends. With a full dive into cartography and architecture tools, we have managed to lift the buildings and project the angles of a shadow polygon.
- Taking into account that each building is surrounded by many others, we have crossed the shadow areas.
- As a result, we have created a sort of solar google maps bearing in mind the day and time.
According to Spanish legislation, ZunPower vehicle belongs to C1 type of urban vehicle mobility (MVU) and thus can be parked at the pavements. That is why we have merged the shadow map with the pavements map elaborated with the help of the City Council of Madrid. (more detailed information about the parking can be found here)
As for consumer usage, we propose a screen of low consumption integrated into the vehicle that will display a chatbot asking for the desired destination. A trip will be supported through with a road map. Arriving at the destination, the app will demonstrate the trajectory and the sunniest lots to park at.
Somvra is a project created by society and for society.
Having a generic dataset applicable to many fields with minor adjustments, we see different scopes of AI application:
- Random forest to estimate if in a concrete point there is sun or shadow based on the day and time.
- Random forest to create a list of the best parking lots ranked by the sunny hours left.
- Computer vision for accidents, collisions, pedestrian crossings and traffic congestion prediction. These aspects are perfectly completing the final service offered to the customer.
- Computer vision for identifying trees growing on the pavements. This approach is crucial for minimizing the error rate of sun presence.
- Analysis of solar battery charging while it is circulating through the city.
Somvra is a product that adds value to society at large. We aim to contribute to the preservation of the environment by transforming the idea of sustainable self-sufficient transport in a modern city.
You can try out the chatbot demo at www.aicitymaps.com
We are on our way of improving Somvra and would be happy to get your feedback. Contact us by aicitymaps@gmail.com
We will write a second post in a few months, keeping you up to date with the latest advances. Get ready to hear from us.
Katerina Godunova, Miguel Ángel Romero Lluch and Lucía Trillo
