![Figure 1: PigNet System Overview 31, 48, 57, 71]. However, these approaches bring severe drawbacks. Image based approaches require constant light and large processing and storage capability, making their deployment impractical in real farm environments. Motion detection has been used to identify whether or not animals are active, but often fails to identify subtler behaviors such as nursing. Wearable sensors can solve these problems, but the social behavior of the animals limits device longevity. PigNet is the first system to use structural vibration to monitor animal behavior. Our approach relies on the idea that animal activity creates unique vibration patterns in the structure of their holding pens. For example, when a pig walks, their footsteps create vibration in the pen structure. When they lie down, their weight changes the natural vibration of the structure. By sensing this vibration, we can infer the different activities of the animals. We use geophone-based sensors attached to the floor of a pig pen to sense the structural vibrations generated by different animal activities. We monitor the farrowing (birth) and pre-weaning period, focusing on activity detection that relates to piglet survival. Each farrowing pen that we monitor contains a single sow (a mother pig) and several piglets. When a sow is ready to give birth, she is moved to an individual farrowing pen where she remains and nurses her piglets for three weeks. First, we detect lying activity of the sow, which can be used to predict onset of farrowing. Studies show that monitoring the farrowing process can halve the mortality rate during this time [27]. Second, we determine when the piglets nurse, a crucial time for the health of the piglets [63]. If we can track piglet nursing, then we can alert the farmers when the piglets are being underfed. Third, PigNet gives a pen-level metric of piglet growth, which can be used to help farmers determine if a pen is progressing normally. Traditional pig growth tracking relies on farmers manually observing piglets, which is costly, time-consuming, and unreliable. In addition, manually handling the piglets to weigh them causes stress and may expose them to health risks [55]. The key focus of our work is on improving the robustness of our system to withstand physical and algorithmic faults that occur in the challenging environment of an operational pig farm. Physical fault tolerance describes the robustness of our hardware system to environmental damage. Over several iterations of our hardware, we improve its robustness to node failure and increase physical node protection. To achieve this robustness, we also focus on simplifying our inexpensive sensor nodes, allowing us to have multiple nodes that we can use as backups in case of failure. Algorithmic fault tolerance describes the robustness of our algorithms to sensing unreliability due to the deployment environment. This can be caused by noise from temporary environmental changes (e.g. vibration from the sow urinating on a sensor), or differences in the structural response because the sow is lying in](https://www.researchgate.net/profile/Yiwen-Dong/publication/351467816/figure/fig1/AS:1023759842557952@1621094775284/PigNet-System-Overview-31-48-57-71-However-these-approaches-bring-severe-drawbacks_Q320.jpg)
May 2021
·
327 Reads
·
26 Citations
What is this page?
This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.
Publications (1)
Citations (1)
... Respiration rates and heart rates are measured by thermal IR and RGB cameras [110]. The 'PigNet' system leverages structural vibration sensing in pig pen floors and exhibits 90% accuracy in behavioral monitoring, addressing video-based limitations requiring constant lighting and welfare concerns with wearable sensors [111]. Moreover, the 'PigV2 system' utilizes ground vibration sensing to monitor pig heart and respiratory rates, achieving average errors of 3.4% and 8.3%, respectively, while ensuring non-intrusive and continuous measurement [112]. ...
- Citing Conference Paper
- Full-text available
May 2021