A low-maintenance, solar-powered outdoor recycling unit that instantly identifies and sorts recyclable material using object recognition . The Saru unit allocates CRV (California Refund Value) directly back to the user through their method of choice.
Our recycling unit is complimented by a mobile application for users, allowing for convenient interface with the unit for fast and easy recycling. The app also features personal recycling history, flex rates, and a host of other deals & incentives.
Our clients are able to access an up-to-date device map featuring the location of each unit and with information on the remaining storage capacity of each unit. The service is packed with analytics and notifications, simplifying waste management.
Secure, Autonomous, and Low-Maintenance
With a solar panel capacity of ~110 W, we exceed our maximum energy consumption of all components. This includes our Raspberry Pi, Peripheral Sensors, Stepper Motor, and Touchscreen yielding usage of ~73.2 W assuming peak consumption. The unit's idle load is approximately 10-15 W.
Fast and Reliable Object Recognition
Using our model, we had a response time from Azure of roughly 0.35s. This time is inclusive of the time it takes to upload, process, and send the result back. Our times are very uniform, with the exception of occasional outliers beyond our control (network latency, other resources having priority, etc).
Sensors and Structure Promote Connectivity
By capturing images of each item recycled, we have insights into how often and what people are recycling. Using a combination of weight and height sensors, we can also individually determine the live capacity of each bin. This allows for efficient collection, by letting the maintenance staff know which bins they can skip collecting. By also saving images of rejected items, we can identify the frequency of non-recyclable material.
In Addition To Its Speed, SARU is Perceptive
By using Microsoft Azure’s Custom Vision, we have trained a model that accurately identifies recyclables based on the type of material (e.g. aluminum, glass, and plastic). Upon the user inserting a bottle, a picture of the recyclable will be taken, uploaded to Azure, and will then sort based on the response received.