Refine
Document Type
- Article (2)
- Conference Proceeding (1)
Language
- English (3)
Is part of the Bibliography
- no (3)
Keywords
- machine learning (2)
- Kalman filter (1)
- MOS sensor (1)
- Zustandsmaschine (1)
- activity classification (1)
- activity monitoring (1)
- ammonia monitoring (1)
- electronic nose (1)
- home cage monitoring (1)
- hybrid sensor system (1)
Abstract
Animal welfare in barns is strongly influenced by air quality, with gaseous emissions like ammonia posing significant respiratory health risks. However, current state-of-the-art ammonia monitoring systems are labor-intensive and expensive. Metal Oxide Semiconductor (MOS) sensors offer a promising alternative due to their compatibility with sensor networks, enabling high-resolution ammonia monitoring across spatial and temporal scales. While MOS sensors exhibit high sensitivity to various volatile compounds, temperature-cycled operation is commonly employed to enhance selectivity, effectively creating virtual sensor arrays. This study aims to improve ammonia detection by designing a virtual sensor array through a cyclic data-driven approach, integrating machine learning with solid-state sensor modeling. The results of a two-week dataset with measurements of four different pig barns demonstrate ammonia sensing with a sampling rate of about 2/min and a range of 1–30 ppm. The method is robust and exhibits a 10% increase in normalized RMSE when comparing testing results of an unseen sensor module with results of the training dataset. A filter membrane boosts accuracy and prevents data loss due to contamination, such as flyspecks. Overall, the used MOS sensor BME688 is effective and economical for widespread continuous ammonia monitoring and localization of ammonia sources in pig barns.
Abstract
This work presents a vibration-based non-contact monitoring method to classify the physical activity of a mouse inside a home cage. A novel tuned-beam sensing device is developed to measure low-amplitude activity-induced cage vibrations. The sensing device uses a mechanical beam structure to enhance a six-axis IMU that increases the signal-to-noise ratio (SNR) by 20 to 40 times in a relevant environment. A sophisticated classification algorithm is developed to process vibration sequences with a variable time frame that utilizes multi-level discrete wavelet transformation (MLDWT) to extract time–frequency features and optimize signal properties. The extracted features are classified by a convolutional neural network–long short-term memory (CNN-LSTM) machine learning model to determine the activity class. The ground truth is obtained with a camera-based system using EthoVision XT from Noldus and a custom post-processor. The method is developed on a dataset containing 300 h of vibration measurements with camera-based reference and includes two separate home cages and two individual mice. The method classifies the activity types Resting, Stationary Activity, Walking, Activity in Feeder, and Drinking with an accuracy of 86.81% and an average F1 score of 0.798 using a 9 s time frame. In long-term monitoring, the proposed method reproduces behavioral patterns such as sleep and acclimatization as accurately as the reference method, enabling home cage monitoring in the husbandry environment with a low-cost sensor.
A Robust Interface for Head Motion based Control of a Robot Arm using MARG and Visual Sensors
(2018)
Head-controlled human machine interfaces have gained popularity over the past years, especially in the restoration of the autonomy of severely disabled people, like tetraplegics. These interfaces need to be reliable and robust regarding the environmental conditions to guarantee safety of the user and enable a direct interaction between a human and a machine. This paper presents a hybrid MARG and visual sensor system for head orientation estimation which is in this case used to teleoperate a robotic arm. The system contains a Magnetic Angular Rate Gravity (MARG)-sensor and a Tobii eye tracker 4C. A MARG sensor consists of tri-axis accelerometer, gyroscope as well as a magnetometer which enable a complete measurement of orientation relative to the direction of gravity and magnetic field of the earth. The tri-axis magnetometer is sensitive to external magnetic fields which result in incorrect orientation estimation from the sensor fusion process. In this work the Tobii eye tracker 4C is used to increase head orientation estimation because it also features head tracking even though it is commonly used for eye tracking. This type of visual sensor does not suffer magnetic drift. However, it computes orientation data only, if a user is detectable. Within this work a state machine is presented which enables data fusion of the MARG and visual sensor to improve orientation estimation. The fusion of the orientation data of MARG and visual sensors enables a robust interface, which is immune against external magnetic fields. Therefore, it increases the safety of the human machine interaction.

