Optimal Control, Pairing, and Scheduling for Manned-Unmanned Vehicles Teaming based on RoboTrust Algorithms

Project began in 2017 and was completed in 2019.

Among many factors in human-robot interaction (HRI), trust is the key enabling factor in human’s acceptance and hence utilization of robot. It has been shown that trust is dynamic in nature and trust analysis is useful for task allocation in multi-robot systems. However, the trust evolution in real-time multi-robot operations has not been well addressed in the literature. It calls for further studies of dynamic evolution and propagation of trust in multiple manned-unmanned vehicles teaming. Furthermore, existing works mainly focused on qualitative analyses and the results have not been accompanied by performance guarantees.

Motivated by the need for quantitative trust measures and trust-based teaming for manned and unmanned vehicles, this project developed trust-based optimal switching control, dynamic pairing, and real-time scheduling algorithms for multiple manned and unmanned vehicles. More specifically, shared autonomous resilient and manual control has been developed for each unmanned/manned ground vehicles in the presence of sensor failures or vehicle-to-vehicle cyber-attacks. A neighbor vehicle anomaly detection approach has been devised onboard each vehicle. Computational trust models have been developed to evaluate the worthiness of each vehicle in the team and utilized in a trust-based information management system to rule out unreliable information due to vehicle-to-cloud cyber-attacks. A human operator scheduling algorithm and human-computer interface have been designed to provide decision aids for the human to supervise the vehicle team. The overall trust-based human-robot teaming framework has been shown to outperform exclusively autonomous and manual operations and improved user experience.