With an increasing number of autonomous systems simultaneously in use and limited personnel, the human’s role must evolve from controlling each individual vehicle to supervising multiple vehicles. iWatchdog™ enables a user to coordinate multiple autonomous systems in order to achieve an overall goal as a team. As the goal is being performed, the user can continuously monitor the progress in iWatchdog™ and intervene when necessary.
Building Human-AI Trust
Human-AI teaming performance is largely attributed to having an appropriate level of user trust, which can impact on the reliance and use of such systems. iWatchdog™ builds user trust through the explanation of its decisions and actions by providing sufficient information about system capabilities, situational awareness and decision trace.
Using the C-BDI™ framework, explainable BDI agents used in iWatchdog™ can explain their actions based on their goals (Desire) and beliefs in the form of a graphical plan (see image). During the run-time of the agent, the graphical plan is constructed to convey the following:
- What is it parent goal?
- What is it doing now to achieve the goal?
- What are the enabling conditions for the action and parent goal?
iWatchdog™ has been demonstrated to perform the following roles:
- Airbase surveillance and intruder detection
- Airbase logistics
- Resilient mobile field communications