Description:

Zero is an intelligent smart-home solution which automates scheduling of appliances using an efficient electricity usage policy model by taking into account variable rate plan and carbon footprint of the electricity that is being delivered to the home. Schedules for appliances are computed locally on a embedded controller mitigating privacy concerns using according to a consumer-provided usage window depending on the preference of low electricity cost or low carbon footprint cost or a balance between the two. We also came up with an algorithm to aggregate the overall cost of a given demography. (Complete implementation details provided in the report provided in the project page of Github.com)

Inspiration:

Electricity consumption is rapidly increasing because of the rapid economic and industrial progress. Paramount importance to optimize electric energy generation and consumption for both producers and consumers while also taking into account environmental effects. Very tough for the consumers to adopt to appliance scheduling algorithms. Need to automate the scheduling of appliances based on an efficient energy usage policy which is efficient as well as environmental friendly.

What it does:

Efficient utilization of non-renewable resources, cost minimization for individuals and whole demography.

How we built it:

Hardware - Raspberry Pi, WEMO Smart switches Software - Python (Please see the Github page)

Challenges we ran into:

Configuring the smart switches with Raspberry Pi over the wireless connection was very challenging Implemented a python daemon for the Raspberry Pi to control the smart switches Democratic aggregation – Facilitating second order effects Regular resource allocation algorithms don’t take into account second order effects Implemented a machine learning algorithm to order to alleviate these effects Evaluation of the proposed design in terms of savings obtained for typical applications with deferrable load (EV, heating/cooling)

Accomplishments that we're proud of:

The estimated savings were 20% for charging an electric car using the proposed model which is considerable given that it is high load. Also, we estimate 20% - 25% savings for HVAC appliances of a living room in an average sized house in U.S. The system successfully scheduled the appliances according to the requirement. As part of the future work, secondary effects will be more closely considered when optimizing for the cost.

What we learned:

Efficient resource allocation for large demography is non-trivial problem. We started by optimizing the resource usage for a consumer. But soon we realized this would result in alternate peaks in power usage. Then we realized that we also need optimize for given demography such that the distribution of power usage is as close to a uniform distribution as possible.

What's next:

Further improve the algorithm to result in uniform distribution in power usage in a given demography while not strongly penalizing certain individuals.

Built with:

Hardware - Raspberry Pi, WEMO Smart switches Software - Python (Please see the Github page)

Prizes we're going for:

HAVIT RGB Mechanical Keyboard

Intel® Movidius™ Neural Compute Stick

DragonBoard 410c

$100 Amazon Gift Cards

Social Entrepreneurship Award

Grand Prize

Lutron Caseta Wireless Kit

Team Members

Pradeep Ambati, Archan Ray, Sourabh Kulkarni, Sachin Bhat
View on Github