HILCPS

Welcome to the Human-in-the-Loop Cyber-Physical Systems Group (HILCPS), IIT Madras, India. We at HILCPS, primarily focus on Cognitive Systems Engineering, Behavioral Informatics and Cyber-Physical Systems.

Cognitive systems engineering is a speciality discipline of systems development that addresses the design of socio-technical systems. Drawing on contemporary insights from cognitive, social, and organizational psychology, we seek to design systems that are effective and robust to human errors. The focus is on amplifying the human capability to perform cognitive work by integrating technical functions with the human cognitive processes. These initiatives will result in improved reliability. It is therefore vital to understand the cognitive state of the operators which give insights about the types of errors they commit, and the workload involved in executing a task.

Our research vision in this area is to design training systems, decision support systems and Human Machine Interfaces that lead to human-error tolerant systems. As a first step, we are exploring the potential of eye-tracking systems to measure and understand the control room operator’s cognitive processes. We are developing a robust, low cost, mostly non-invasive physiological measurement and analytics system that can aid behavioural informatics. The following are some of our publications in this field, visit our publications page to know more.

The behavioural informatics platform depicted below aims to identify the work pattern and performance of ATCos (Air Traffic Controllers). ATCos primarily obtain information visually from the radar displays and communicate their decisions orally to the pilots. Therefore, it is imperative to have a complete visual and aural representation of the ATCo work environment. Suitable sensors are required to obtain this representation of the work environment. Specifically, video sensors are required to obtain the information that ATCos perceive from the radar screens. Voice sensors, on the other hand, allow us to incorporate the communication between ATCOs and pilots into our analysis. The effect of these interactions on the cognitive state of the ATCo can be determined by physiological sensors such as eye-tracking, electroencephalography, thermography, electrocardiography, and skin conductance. These techniques enable a comprehensive measurement system that is then amenable to analytics.

We wish to initially use only eye-tracking and video analytics to carry out gaze informatics and thereby uncover insights on the cognitive state of ATCos using gaze pattern, entropy, pupil power spectra and other cognitive metrics. In this work, we intend to carry out studies to:

Over the course of this work, a variety of scenarios such as high and low traffic density, conflict and non-conflict situations, and different shift timings will be simulated to understand their effect on performance. These scenarios will help us determine the situation-specific influences on the ATCos’ cognitive behaviour. Subsequently, we also wish to carry out studies on evaluating training adequacy, transplanting expert cognitive patterns to beginners, and examining the effects of fatigue on performance. The different elements of the platform will be iteratively enhanced based on their performance, including the quality and quantity of data obtained, measurement accuracy, feedback on non-intrusiveness, ease of calibration, and other protocols. Finally, large-scale studies will be conducted.

Energy systems around the world have been undergoing significant transitions over the last decade, due to environmental concerns and new geopolitical realities. Traditional fossil-fuel based sources lead to significant environmental and health impacts. Renewable energy sources, on the other hand, offer a promise of clean energy. However, the heterogeneous (spatial) and sporadic (temporal) nature of their availability leads to many design and operational challenges to the energy grid, especially in a large country like India. Also, there is a significant difference between the energy consumption of developed countries with high human development index (HDI) like United States, Germany, and the Netherlands in contrast to less developed ones with low HDI like India. It is well understood that, for less developed countries, social and economic development is positively correlated with per-capita energy consumption.

Our research vision in this area is to develop a deep understanding of the profound changes in energy systems that would be induced in the medium to long-term by socio-technical transitions. Specifically, we plan to identify the demand profile of the residential customers based on the data acquired from households of different socio-economic and geographical conditions and develop energy management systems. These usage patterns can be assessed further using a technique known as Non-Intrusive Load Monitoring (NILM). NILM disaggregates the aggregate power consumption measured at the consumer entry point to identify the operational states of individual appliances that are being used by the consumer. NILM uses power consumption level, power consumption trend and transient behaviour of appliances in the disaggregation process. These behaviours are distinct and constitute the feature vector of an appliance.

We have currently developed a NILM algorithm based on data collected from a few Indian households using our in-house meter. The data collected from these measurements can be compressed using our works on compressed sensing. We intend to use this data to estimate the states of the power distribution system. Based on the demand profiles from the households, we propose to develop an Energy Management System (EMS) that would utilize optimization approaches to identify the cost-effective energy profile.