In this work, authors address workload computation combining human activity recognition and heart rate measurements to establish a scalable framework for health at work and fitness-related applications. The proposed architecture consists of two wearable sensors: one for motion, and another for heart rate. The system employs machine learning algorithms to determine the activity performed by a user, and takes a concept from ergonomics, the Frimat’s score, to compute the corresponding physical workload from measured heart rate values providing in addition a qualitative description of the workload. A random forest activity classifier is trained and validated with data from nine subjects, achieving an accuracy of 97.5%. Then, tests with 20 subjects show the reliability of the activity classifier, which keeps an accuracy up to 92% during real-time testing. Additionally, a single-subject twenty-day physical workload tracking case study evinces the system capabilities to detect body adaptation to a custom exercise routine. The proposed system enables remote and multi-user workload monitoring, which facilitates the job for experts in ergonomics and workplace health.

Kinetic energy harvesting technologies have been progressively used to power wearable devices and to sense the context through energy generation patterns. However, detecting human activities with signals from kinetic harvesters still needs improvement due to the use of approaches based on handcrafted features and the overfitting to device location or subjects. Hence, in this article, we present a deep learning architecture that leverages the feature extraction capability of the convolutional neural networks and the construction of the temporal sequences of recurrent neural networks to improve existing classification results. To provide sufficient data for the deep learning classifier, we propose three data augmentation methods to increase intraclass variance simulating new users performing the same activities. The proposed architecture outperforms existing approaches of kinetic harvesting-based human activity recognition by 13% of accuracy when the training data are augmented with the proposed methods. Finally, given the dependency of kinetic harvesting signals on device location and subjects, we employ transfer learning to improve the classification performance when the system is exposed to new subjects and locations. Transfer learning helps to increase classification performance by 30% when the device location is changed and 35% when the data come from a new subject.

Kinetic energy harvesting is a promising technology towards the development of alternative battery-charging schemes or even self-powered wearable devices that obtain their power supply from human motion. Although there are many developments with schemes that leverage piezoelectric materials and human motion to power devices especially from footsteps, some other body locations like the wrist still need assessment with piezoelectric generators to evaluate their potential of limitations. In this work, we present the results of logging the energy transference from a wrist-worn piezoelectric harvester to a battery in a wearable device. This system is the continuation of our previous work where we implemented the harvester with a resistive load previously tuned to obtain maximum power and assessed the energy harvested during physical activities. Now, we replace the linear load with a charge controller and a Li-ion battery in the same wearable set-up. These new conditions are not optimal for the piezoelectric generator but present a more realistic environment for the kinetic harvester and allows a more precise study of the feasibility of a self-powered system. Tests show that five minutes of activities that involve arm motion can provide between 1.75 mJ and 2.98 mJ of energy, which can represent between 3.6 seconds and 6.2 seconds of additional battery duration. Hence, these results provide an insight of the limitations and challenges remaining in the piezoelectric-based kinetic harvesting field for wearable devices.

This paper presents a system that reports the harvested energy from a wearable piezoelectric generator to a mobile application in real time. The system is adapted to be worn as a bracelet, harvesting energy from wrist movements during physical activities. A mobile app connects automatically to the harvesting monitor, tracks and stores the harvested voltage and power values, and also can compute the total harvested energy during a customizable time window. Test results showed that some arm movements displayed enough harvested energy to obtain a sample from a heart rate tracker and transmit it to a mobile device. Moreover, the harvested energy from jogging during 30 minutes can supply the tracker up to 37 sampling cycles. The developed system allows real-time harvesting monitoring, which can be used to determine the suitability of kinetic energy harvesters with the advantages of mobile computing.

This paper describes a real-time human activity recognition and heart rate tracking system using ultra-low power wearables in a mobile platform. The mobile application shows both real-time and historical data of activities performed by the user along with the average heart rate for each activity. Random Forest and k-Nearest Neighbors algorithms were used to classify, showing general accuracies of 97.3% and 98.6%, respectively, with only eight features. The presented heart rate monitor is compared with an Apple Watch series 3 and they were demonstrated to display statistically equal values, confirming the reliability of the built monitor. Both wearables have a diameter of 4.5cm and consume currents below 10mA on transmission events and are powered with coin cell batteries, setting feasible characteristics for the need of modern wearable systems.

Occupational hygiene requires evaluation of different risk sources in the workplace. The level of physical workload may create stress, fatigue and injuries. Therefore, activity monitoring provides valuable information for companies in assessing and solving possible hazards in the workplace. The article presents a system using wearable technology to monitor and evaluate physical workload with in situ measurements. The system uses a smartwatch and a mobile application for Android phones. During workload monitoring, the application displays physiologic variables such as heart rate, calories, body temperature, galvanic skin response and number of steps. Additionally, the system computes absolute and relative cardiac cost, and Frimat coefficients. Tests were performed on 10 individuals from a janitor staff (5 men and 5 women), monitoring every task during their most demanding hour. Results agree with the type of activity developed in different intervals, showing light and very light workload for different tasks in all workers.