Detector Case Bottom View
Detector Case Assembly View
Mobile App Interface
Watcher Green: Observing Forests from Within
We have been looking at the forest from the sky: satellites. We know the temperature and sunshine in a forest, but we can never know how a single tree is feeling: Is it doing well? Does it feel dry? Or does it need more nutrition?
Now, Watcher Green offers a solution. Watcher Green is a daring project aiming at the prediction and protection of forests. We deploy countless detectors to the forests, attaching them directly to trees. These detectors regularly reports the status of a single tree to the server, representing the state of trees in its area. Data from all corners aggregate to unveil the secrets of the woods, previously unseen by the eyes from above. With Watcher Green, we will be able to know immediately when a drought or logging is happening, and send our forest rangers on site. We can even predict forest fires with greater accuracy than ever before.
Challenges: A Balance between Sustainability and Scalability
The forests are, simply put, too large. To cover vast landmass of forests with our detectors, detectors are required to be cheap enough for mass production. On the other hand, we need to scale down natural impacts as much as possible, meaning that recovering of older detectors is a must.
The forests also differ. Although solar panel works well to add to the longevity of a detector, some forests are so dense that obtaining enough sunlight is almost impossible. This forces us to use batteries which requires regular recharging. Modularity thus became one of our main focuses, hence our solution.
Solution: Faithful Watchers
Modularity allows us to achieve mass manufacturing, deployment and replacement with ease.
The detectors(on DragonBoard 410c) are assembled from inexpensive parts so that the cost of our prototype was as low as $10. If the product enters mass production, the actual price can be lower still.
To cater for all terrains, we designed each functioning piece to be a module that can be replaced without affecting the integrity of the system. Depending on the sunlight level, a detector can choose between a battery and a solar panel. We adopted I2C protocols so that the software can smart adjust to the modules being used without the need to rewrite software each combination.
In addition, a maintainer can easily pinpoint a dysfunctional detector by its distress beacon and change its damaged part by simply unplugging an older part. The hardware and software work together to help us better understand the forests. Hardware are the senses of Watcher Green. The temperature module, the moisture module, the tree humidity module, and the solar module: all are options to choose from. A study but easy-to-remove outer shell protects the vulnerable insides from the weather, with downward-facing slits allowing for only air.
Humans: talking to Mother Nature via Cloud
ElasticSearch, Kibana and GKE
One of the biggest challenges that IoT devices faces is the scale of dataset. We deployed our ELK on Google K8S Engine which can scale easily when facing different number of users. From there we can further analyze what combination of factors lead to most wildfires. With the raw data from our sensors and accessible public data, we can get to almost anywhere.
Watcher Green is designed individuals like forest rangers, providing them with data to navigate the forest and locate abnormalities more efficiently. It is also an easy-to-use toolkit for family usage. We designed the Watcher Green terminal as a mobile application. In addition to detailed status of single detectors, the mobile app allows for access to Kibana so that users will have an understanding of the situation as well.