Affective Computing

My peers and I have been assigned a cross-disciplinary project on affective computing, the ability to recognise, interpret, process and simulate human affects. Designers were tasked with producing a game idea revolving around this concept, and as programmers our job is to select and implement this in the most optimal way. The idea was that each designer would require some form of data, entirely of their choosing that would require a non-standard input device to acquire. Originally, we had split into groups for each designer. Despite the difference in ideas, it became clear that on a lower level, the data they requested almost all boiled down to the same affect.

Ideas included the ability to read levels of fear, stress and calmness. Fear almost directly relates to stress with calmness simply being the opposite. As such, we have decided to join forces and focus on two projects only. Understanding and collecting data on the anxiety of a person and measuring how focused one is, the only idea that didn't require the same metric. The project I chose, requires players navigate a paper plane through a randomly generated level with obstacles. The primary mechanic is that manoeuvrability becomes hindered with turbulence, as a direct consequence of a rise in anxiety. Players must learn to reduce stress and remain calm, in order to safely reach the end of the level.

Devices to monitor levels of anxiety already exist from companies such as Neumitra and Galvanic Limited, and work by measuring proxies such as heart rate, skin conductance and temperature. The downfall is that if a decent level of accuracy is required, so is prior calibration. This can often be as simple as recording data when a person is in a neutral state, as to determine what should be considered baseline when evaluating how calm a person is. Unfortunately these devices are not as easily attainable, compared to other more common wearable systems.

Technology from the NASA Glenn Research Center, show that it is possible to monitor heart rate using a camera alone. This is achieved by examining very slight changes of colour in the forehead, with isolated use of the green channel to avoid excessive noise. Optical absorption and illumination in the skin change depending on heart rate and are what make it possible to observe using a light sensing device such as a camera in this case. I will post in future on how this goes, currently I am looking at utilising real-time imagery from webcam's, given their high availability and low cost.