Analysing and modelling human trust to a navigation robot

Human trust plays a crucial role in Human-Machine Interactions (HMIs) within autonomous systems. This paper delves into the factors that influence human trust in machines, including varying error rates and types made by the machine, as well as the human’s ability to intervene and rectify errors. To explore these factors, we conducted three scenarios involving a simulated claw robot navigating through multiple objects to detect and locate a target object. The first scenario examined the effect of changing error rates on human trust in the machine. In the second scenario, we investigated how variability in speed and accuracy of reaching the target impacted human trust. Lastly, we explored whether human trust in the machine changed when individuals had the capability to intervene and correct severe errors made by the machine. We then proposed a regression model to estimate human trust. Our results showed that human trust is significantly affected by changes in error rates. Our participants reported a higher trust on robot with low speed but higher accuracy in performing the task than the robot with high speed but lower accuracy. Interestingly, the ability to intervene and correct the robot’s errors improved the participants’ trust in the robot. Our regression result showed that we can estimate trust using different type of errors committed by machine which can be applied in real world scenarios.