Description:
Rapid advances in artificial intelligence are driving the adoption of robotics and automation in space and on Earth, providing new solutions for exploration and surveying, passenger transport, last-mile delivery, and automated warehouses. Nonetheless, humans will remain indispensable to supervise and manage such fleets, as humans and computers have complementary strengths. However, human operators’ capacities will be challenged and exceeded as task complexity and fleet size increase. The deployment of fleets of autonomous vehicles opens up transformational opportunities in space applications and on Earth. However, large-scale deployment is held back by safety and reliability of these systems, and a lack of personnel trained in this interdisciplinary research area.
The project’s main objectives are to establish generalizability of a machine learning (ML) algorithm between simple and complex cognitive tasks in a laboratory environment; to establish generalizability of this algorithm between lab-based and simulated space exploration environments, in a virtual reality (VR) environment; to test the effectiveness of basic adaptive automation of workload level within the simulated mission environment in real-time; and to perform a proof-of-concept transfer to real-world robotic control, in an analogue exploration mission. This work will satisfy a pressing need for validating physiological markers of cognitive workload, and contribute highly qualified personnel to the Canadian workforce.
Although this research specifically aims at facilitating space missions, outcomes could be applicable to any situation where humans are required to perform critical tasks using semi-autonomous tools. It holds natural applications on Earth for mining, drone delivery, and monitoring of fleets of self-driving cars.