John Kemp

Augmented Reality Environmental Monitoring Using Wireless Sensor Networks

Environmental monitoring brings many challenges to wireless sensor networks: including the need to collect and process large volumes of data before presenting the information to the user in an easy to understand format. This paper presents SensAR, a prototype augmented reality interface specifically designed for monitoring environmental information. The input of our prototype is sound and temperature data which are located inside a networked environment. Participants can visualise 3D as well as textual representations of environmental information in real-time using a lightweight handheld computer.

Postural activity monitoring for increasing safety in bomb disposal missions

In enclosed suits, such as those worn by explosive ordnance disposal (EOD) experts, evaporative cooling through perspiration is less effective and, particularly in hot environments, uncompensable heat stress (UHS) may occur. Although some suits have cooling systems, their effectiveness during missions is dependent on the operatives posture. In order to properly assess thermal state, temperature-based assessment systems need to take posture into account. This paper builds on previous work for instrumenting EOD suits with regard to temperature monitoring and proposes to also monitor operative posture with MEMS accelerometers. Posture is a key factor in predicting how body temperature will change and is therefore important in providing local or remote warning of the onset of UHS. In this work, the C4.5 decision tree algorithm is used to produce an on-line classifier that can differentiate between nine key postures from current acceleration readings. Additional features that summarize how acceleration is changing over time are used to improve average classification accuracy to around 97.2%. Without such temporal feature extraction, dynamic postures are difficult to classify accurately. Experimental results show that training over a variety of subjects, and in particular, mixing gender, improves results on unseen subjects. The main advantages of the on-line posture classification system described here are that it is accurate, does not require integration of acceleration over time, and is computationally lightweight, allowing it to be easily supported on wearable microprocessors.

Leveraging Knowledge From Physiological Data: On-Body Heat Stress Risk Prediction With Sensor Networks

The paper demonstrates that wearable sensor systems, coupled with real-time on-body processing and actuation, can enhance safety for wearers of heavy protective equipment who are subjected to harsh thermal environments by reducing risk of Uncompensable Heat Stress (UHS). The work focuses on Explosive Ordnance Disposal operatives and shows that predictions of UHS risk can be performed in real-time with sufficient accuracy for real-world use. Furthermore, it is shown that the required sensory input for such algorithms can be obtained with wearable, non-intrusive sensors. Two algorithms, one based on Bayesian nets and another on decision trees, are presented for determining the heat stress risk, considering the mean skin temperature prediction as a proxy. The algorithms are trained on empirical data and have accuracies of 92.1 ± 2.9% and 94.4 ± 2.1%, respectively when tested using leave-one-subject-out cross-validation. In applications such as Explosive Ordnance Disposal operative monitoring, such prediction algorithms can enable autonomous actuation of cooling systems and haptic alerts to minimize casualties.

Increasing Safety of Bomb Disposal Missions: A Body Sensor Network Approach

During manned bomb disposal missions, the combination of the protective suits weight (37 kg), physical activity, high ambient temperatures, and restricted airflow can cause the operatives temperature to rise to dangerous levels during missions, impairing their physical and mental ability. This work proposes to use body sensor networks (BSNs) to increase the safety of operatives in such missions through detailed physiological monitoring, fusion of health information, and remote alerts. Previous trials conducted by the authors have shown no correlation between the suit wearers temperature at any single skin site and their core temperature, nor between single-point temperature variations and subjective thermal sensation. This paper reports on the development of a wearable, wireless, networked sensing system suitable for integration within the suit and deployment in manned missions. A sensor fusion and modeling approach is proposed that estimates the overall thermal sensation of the suit wearer, in real time, based on the multipoint temperature data. Zhangs thermal sensation model was used in this work. Modeling is performed locally to enable cooling system actuation, provide local feedback, and accommodate application specific constraints. Experimentation with the prototype confirms the importance of multisite skin measurement, timely cooling actuation, and monitoring the operatives thermal state. Evaluation of Zhangs model highlights the need for a bespoke model to account for suit and mission specific factors. The deployed BSN has been evaluated through experimental trials using a number of subjects in mission-like conditions and has been shown to be appropriate for the target application.

Wearable posture recognition systems: Factors affecting performance

This paper presents an investigation into the design space for real-time, wearable posture classification systems; specifically, it analyses the impact of various factors/design choices on classification accuracy when using C4.5 decision trees. The factors can be broadly divided into: 1) system factors (such as sensor sampling rate and number of sensors used) and 2) algorithm and training factors (such as quantity of training data and temporal data features used). These factors are analysed in the context of a case study involving postural activity monitoring of Explosive Ordinance Disposal (EOD) operatives. The case study involves classifying a set of eight postures commonly encountered in EOD missions: sitting, walking, crawling, laying (on all sides) and kneeling. Design guidelines and generic lessons for a wider class of applications can be drawn from the work.

Applicability of thermal comfort models to car cabin environments

Car cabins are non-uniform and asymmetric environments in relation to both air velocity and temperature. Estimating and controlling vehicle occupant thermal comfort is therefore a challenging task. This paper focuses on evaluating the suitability of four existing thermal comfort models, namely the Predicted Mean Vote (PMV), Taniguchis model, Zhangs model and Nilssons model in a variety of car cabin conditions. A series of comfort trials were performed ranging from controlled indoor trials to on-road driving trials. The outputs of all four models were compared to the sensation index reported by the subjects situated in the driver seat. The results show that PMV and Nilssons model are generally applicable for the car cabin environment, but that they are most accurate when there is a small air temperature rate of change (of under 1.5 °C per minute), giving correlation levels of 0.91 and 0.93 for the two models respectively. Taniguchis and Zhangs models were found unsuitable for all conditions, with correlation levels ranging between 0.03 and 0.60. Nilssons model is recommended by the authors based on the level of agreement with the subjective reports, its ability to estimate both local and overall thermal sensation and the smaller number of input parameters.

Long-Term Behavioural Change Detection through Pervasive Sensing

The paper proposes an information generation and summarisation algorithm to detect behavioural change in applications such as long-term monitoring of vulnerable people. The algorithm learns the monitored subjects behaviour autonomously post-deployment and provides time-suppressed summaries of the activity types engaged with by the subject over the course of their day to day life. It transmits updates to external observers only when the summary changes by more than a defined threshold. This technique substantially reduces the number of transmission required by a wearable monitoring system, both through summarisation of the raw data into useful information and by preventing transmission of duplicated or predictable data and information. Based on evaluation using simulated activity data, the proposed algorithm results in an average of one transmission per month following an initial convergence period (reaching less than 1 transmission per day after only three days) and detects a change in behaviour after an average of 1.1 days.

Fielded Autonomous Posture Classification Systems: Design and Realistic Evaluation

Few Body Sensor Network (BSN) based posture classification systems have been fielded to date, despite laboratory based research work confirming their theoretical suitability for a range of applications. This paper reports and reflects on two algorithms which i) improve the accuracy of real-time, multi-accelerometer based posture classifiers when dealing with natural movement and transitions and ii) maximize a wearable systems battery life through distributed computation at nodes. The EWV transition filters proposed here increase the classification accuracy by 1% over unfiltered results in realistic scenarios and significantly reduces spurious classifier output in real-time visualizations. A 200 fold transmission reduction from the on-body system to an outside system was achieved in practice by combining the transition filters with an event-based design. Furthermore, a method of reducing transmissions between on-body data gathering nodes based on distributed processing of the classifier rules (but maintaining a one-way flow of communications during system use) is also described. This provides a 3.3 fold reduction in packets and a 13.5 fold reduction in data transmitted from one node to the other in a two-node wearable system.

Energy Profiling in Practical Sensor Networks: Identifying Hidden Consumers

Reducing energy consumption of wireless sensor nodes extends battery life and/or enables the use of energy harvesting and thus makes feasible many applications that might otherwise be impossible, too costly or require constant maintenance. However, theoretical approaches proposed to date that minimize Wireless Sensor Network energy needs generally lead to less than expected savings in practice. We examine the experiences of tuning the energy profile for two near-production wireless sensor systems and demonstrate the need for: microbenchmark-based energy consumption profiling; examining start-up costs; and monitoring the nodes during long-term deployments. The tuning exercise resulted in reductions in energy consumption of: 93% for a multihop Telos-based system (average power 0.029 mW); 94.7% for a single hop Ti-8051-based system during startup; and 39% for a Ti-8051 system post start-up. This paper shows that reducing the energy consumption of a node requires a whole system view, not just measurement of a typical sensing cycle. We give both generic lessons and specific application examples that provide guidance for practical WSN design and deployment.

Comfort in Cars - Estimating Equivalent Temperature for Comfort Driven Heating, Ventilation and Air Conditioning (HVAC) Control

Equivalent Temperature is generally considered an accurate predictor for thermal comfort in car cabins. However, direct measurement of this parameter is impractical in fielded applications. The paper presents an empirical, multiple linear regression based approach for estimating body segment equivalent temperatures for car cabin occupants from different sensors within the car. Body part equivalent temperature at eight segments and cabin sensor data (air temperature, surface temperature, mean radiant temperature, humidity and solar load) was gathered in a variety of environmental and cabin conditions. 38 experimental hours of trials in a controlled environment and 26 experimental hours of realistic driving trials were used for training and evaluating the estimator’s performance. The estimation errors were on average between 0.5 °C and 1.9 °C for different body parts for trials within a controlled environment, while for trials in realistic driving scenarios they ranged between 1 °C and 2 °C. This demonstrates that passenger body part equivalent temperature can be estimated using a multiple linear regression from environmental sensors and leads the way to comfort driven Heating, Ventilation and Air Conditioning control.