There is currently a proliferation of research activity in the field of autonomy. New tools and techniques for automation of vehicles in particular are being developed for applications across the spectrum of industry, academia and government.
Autonomous systems are indeed engineered to unburden the operator of awareness over the systems which govern the vehicle’s safety processes. While autonomous engineering is designed to appeal to the purchaser, it effectively relieves the operator of an essential duty; to remain cognitively engaged during operation of the system. In reality, many of the systems engineered into the vehicle actually diminish the driver’s capacity to remain engaged. This is known as ‘engineered distraction’ in the industry and results in cognitive lapses and a loss of situational awareness by the operators. Human error in Human Robot Teaming failures is primarily due to human cognitive and attention lapses.
Human Cognition, as it applies here, is concerned with the internal mental processes that involve as a result of some external stimulus and the resulting behavioral response.
The aim is to consider the role of humans in Human Machine Teaming in automated systems, to develop synergistic automation strategies that take advantage to the strengths of each teaming agent.
The ability to determine the mental state of human operator for improved situational awareness will allow more efficient “teaming” between machines and humans to successfully meet system mission goals.
Our goal is to develop an engineering approach to cognition that provides a rigorous robust structure to determining human cognitive states. Brain wave monitoring gives insight to the cognitive mental states of the drivers.
When humans are acting as passive monitors of autonomous systems, it is inherently difficult for them to fully understand what is going on due to lower levels of cognitive engagement. Here we propose to investigate and monitor the features that influence the human cognitive processes involved in successful oversight, intervention, and teaming with automated systems.
These collaborative works will result in the development of data collection protocols in order to evaluate situational awareness under various mental states. The mental states of interest are guided by previous literature, are of high relevance to driving scenarios, and include fatigue/drowsiness, distraction, and cognitive load.
Human cognition and emotion are very difficult to determine just by looking at the face and the behavior of a person. There has been considerable research conducted to identify human emotion via the study of brain waves. The relevant mental states of human drivers in autonomous vehicles for example may include, drowsiness, fatigue and distraction. Here we propose to use modern, non-invasive brain wave monitoring technology and quantum theory to determine, in real time, the mental state of human operators in human machine teamings. This will provide the machine with the needed situational awareness of the human to mitigate an existing automation conundrum. In this manner the vehicle can be capable of transmitting haptic cues (such as pulses and vibrations) to signal an anticipated need for human engagement.
We are collaborating to inform a quantum-like modeling approach to assessing human cognition by defining the state of the human in a geometrical fashion using state vectors placed in a specific basis.
We believe that the mathematics of quantum theory are relevant to the contextual phenomena of trust. We note that in applying the mathematical machinery of quantum theory to human cognition we do not wish to imply that the human brain and the corresponding psychological processes have a quantum nature. We simply wish to take a quantum-like modeling approach to assessing human cognition.
To better clarify our approach, we have formulated a very simple version of the quantum probability approach to cognition: Engaged-Unengaged. This is the most basic form of situational awareness for an operator. Quantum Probability theory, is a geometric theory of assigning probabilities to outcomes. The use of a quantum probability approach to human cognition is currently developing, good introductions to the broad area of Quantum Probability applications.
More general emotions, such as happiness, could be represented by subspaces of higher dimensionality. In quantum probability theory, it is easier to keep track of the (orthogonal) projections onto the primary emotional subspaces and to manipulate these projections as linear operators on the Hilbert Space. One of the most attractive properties of quantum probability theory over standard probability theory is the use of the projections to handle sequences of events or the conditioning of events by other events; this is done by simply multiplying projection operators.
An automation conundrum exists in which as more autonomy is added to a system, and its reliability and robustness increase, the lower the situation awareness of human operators and the less likely that they will be able to take over manual control when needed.
Hence the need for a Human Machine Interface (HMI) Manager that could smoothly Blend or Share Autonomy as Needed. By studying ethology, we can understand how animals interact with humans and these approaches/techniques can be used for HMI
The challenge with Autonomy is that unexpected automation transitions will occur when the automation suddenly passes control to the human operator who cognitively may not be ready to take over.
Mica Endsley defines this as the Autonomy Conundrum and points out in her paper “From Here to Autonomy”, that the goal of full system autonomy is quite difficult, and most systems will exist at some level of semi-autonomy in the foreseen future. Thus the automation conundrum potentially creates a fundamental barrier to autonomy in safety critical systems, such as driving. While recent system autonomy efforts are beginning to leverage artificial intelligence and learning algorithms to allow the platforms to better adapt to unanticipated and changing situations; it is becoming clear that the design and inclusion of human-autonomy interfaces are needed that span across many disciplines and involve professionals from diverse fields.
Ethology is the study of animal behavior. Here it inspires and informs our quest for efficient human machine teaming by observing human-animal teaming such as the horse and human rider or First Responders and service canine teams.
Complemation is designed to aid in complementing human skills and abilities rather than replacing them. (Schutte, 1999). When combined with Ethology results in ethologically inspired Blender models of automation that complements or mimics the individual teamings between animals and humans.
Modern autonomous systems are, and will be in the future, dependent on the development of successful approaches to human–autonomy teaming. Here we define and explore “the system” consisting of the platform, the software and the human “operator”. Future automation should be designed to mitigate the risk associated with the autonomy conundrum by actively assessing and managing the humans’ engagement. This website introduces new approaches to automation that involve human cognition, human-animal teaming, and complemation concepts in automation and control.