Result
Daily health report
Daily health reports are stored in DynamoDB and can be seen from the Web App. In the report, the feeding plan, motion and weight are shown to users. Also, a real-time pet status can be shown to the users. If the pet is moving, the status shows “active” and the active pet icon appears. Otherwise, the status shows “sleep” and the sleeping pet icon appears
Daily health reports are stored in DynamoDB and can be seen from the Web App. In the report, the feeding plan, motion and weight are shown to users. Also, a real-time pet status can be shown to the users. If the pet is moving, the status shows “active” and the active pet icon appears. Otherwise, the status shows “sleep” and the sleeping pet icon appears
Health status alert
The main algorithm to track the health status of a pet is to trace the weight of it. A threshold is set to determine health status. Here we set 3 kg as the threshold for the weight sensor. If the pet is heavier than 3 kg, it is in normal health status. Otherwise, the pet is unhealthy and an alert will be shown on the Web App.
Intelligent feeding plan
According to the daily health report, an algorithm is implemented to generate a nutrition feeding plan. It can automatically provide the food, water or other nutrition for the pet by controlling the servo (control the time and duration it works) at the bottom mouth of the container.
The main algorithm to track the health status of a pet is to trace the weight of it. A threshold is set to determine health status. Here we set 3 kg as the threshold for the weight sensor. If the pet is heavier than 3 kg, it is in normal health status. Otherwise, the pet is unhealthy and an alert will be shown on the Web App.
Intelligent feeding plan
According to the daily health report, an algorithm is implemented to generate a nutrition feeding plan. It can automatically provide the food, water or other nutrition for the pet by controlling the servo (control the time and duration it works) at the bottom mouth of the container.
Discussion and Further Work
The Smart Home Companion system has reached all its designed functions. It can obtain the indoor position, weight and images of a pet, then track and monitor the status of a pet, including weight, movement pattern, and daily activity in order to determine the nutrition feeding portion throughout the day. Irregularities such as unhealthy status of a pet will trigger alerts to the pet owner. All pet status will be store for future checking. AWS plays a significant part in data collection and processing, which can generate intelligent nutrition feeding based on the weight, movement pattern and daily activity level of the pet.
Here are the improvements that should be done in the future:
Indoor Positioning
Bluetooth is not accurate enough to provide an indoor position, which can only achieve one-meter precision. However, Ultra Wideband (UWB) that has a much higher localization precision is a better substitution. UWB is more stable and can achieve ten-centimeter precision. Using UWB can bring a Significant enhancement in localization to our system.
Health status detection
For now, the system can only detect health status by weight. This is not enough to track the accurate health status for pets. Using ML to detect the pet health status with an image should be added to the system, which can detect more health parameters and give out a more detailed health report.
Performance
Google vision kit take a picture and detect the object locally, which cause a delay of 2-4 seconds. We can move the detection step to the cloud and make it faster.
Currently, we use Node.js Plotly API to draw a plot for the website, which takes 2 seconds. We can improve by using AWS lambda functions.
User Interface
Currently we build a web application. We can further improve the UI design, extend the supported platform like Android or IOS, make the UI automatically adjustable for different devices.
Cloud - AWS
Stability and security are very important for IoT systems. For stability, a more reliable algorithm should be implemented in the local devices and AWS, which can manage most kinds of emergencies such as network congestion. For security, an encryption algorithm should be used for data transmission. Since there are risks for sending images of users’ home to AWS, this data should be encrypted for privacy protection.
Here are the improvements that should be done in the future:
Indoor Positioning
Bluetooth is not accurate enough to provide an indoor position, which can only achieve one-meter precision. However, Ultra Wideband (UWB) that has a much higher localization precision is a better substitution. UWB is more stable and can achieve ten-centimeter precision. Using UWB can bring a Significant enhancement in localization to our system.
Health status detection
For now, the system can only detect health status by weight. This is not enough to track the accurate health status for pets. Using ML to detect the pet health status with an image should be added to the system, which can detect more health parameters and give out a more detailed health report.
Performance
Google vision kit take a picture and detect the object locally, which cause a delay of 2-4 seconds. We can move the detection step to the cloud and make it faster.
Currently, we use Node.js Plotly API to draw a plot for the website, which takes 2 seconds. We can improve by using AWS lambda functions.
User Interface
Currently we build a web application. We can further improve the UI design, extend the supported platform like Android or IOS, make the UI automatically adjustable for different devices.
Cloud - AWS
Stability and security are very important for IoT systems. For stability, a more reliable algorithm should be implemented in the local devices and AWS, which can manage most kinds of emergencies such as network congestion. For security, an encryption algorithm should be used for data transmission. Since there are risks for sending images of users’ home to AWS, this data should be encrypted for privacy protection.
Conclusion
As an intelligent IoT system, Smart Home Companion tracks and monitors the status of a pet, including weight, movement pattern, and daily activity in order to determine the nutrition feeding portion throughout the day. Our solution targets the pet owners that live a busy lifestyle or travel regularly that their pets need to be taken care of without involving the risk of pet sitting or pet boarding. Bluetooth has been used for indoor localization, AWS IoT (MQTT) has been used for data transmission and other AWS services such as Lambda, DynamoDB and S3 have been used for data processing. Finally, through intelligent data collection and processing, we were able to achieve intelligent nutrition feeding based on the weight, movement pattern and daily activity level of the pet. Daily health reports can be checked by the pet owner. In the future, UWB will be used for more precise indoor localization. The system performance should be improved and the UI should be more user-friendly and flexible. Furthermore, system reliability and security need to be improved for emergency management and privacy protection.
Acknowledgments
We would like to express our special thanks of gratitude to our professor (Zoran Kostic) as well as our teaching assistants (Peter Wei, Long Jiao and Chengrui Wu) who gave us the golden opportunity to do this wonderful project in the course of IoT - SYS &PHY DATA ANALYTICS, which also helped us in doing a lot of research. We learned many new knowledge and skills during this project and we are really thankful to them.
References
[1] https://github.com/guoshanglin/Smart-Pet-Care
[2] Instructables. (2017, October 15). Cat Tracking Using Bluetooth Indoor Positioning. Retrieved from https://www.instructables.com/id/Cat-Tracking-using-Bluetooth-Indoor-Positioning/
[3] “Quick Start Guide – BC037 IBeacon.” Blue Charm Beacons, 6 Apr. 2019,
bluecharmbeacons.com/quick-start-guide-bc037-ibeacon/.
[4] “[Arduino|module]Hx711电子秤套件(视频).” Agu's Mill 古作坊, aguegu.net/?p=1327.
[2] Instructables. (2017, October 15). Cat Tracking Using Bluetooth Indoor Positioning. Retrieved from https://www.instructables.com/id/Cat-Tracking-using-Bluetooth-Indoor-Positioning/
[3] “Quick Start Guide – BC037 IBeacon.” Blue Charm Beacons, 6 Apr. 2019,
bluecharmbeacons.com/quick-start-guide-bc037-ibeacon/.
[4] “[Arduino|module]Hx711电子秤套件(视频).” Agu's Mill 古作坊, aguegu.net/?p=1327.