LARIAT: Predictive Haptic Feedback to Improve Semi-Autonomous UGV Safety in a Case Study

While much work is being done to advance autonomous capabilities of mobile robotics, specifically unmanned ground vehicles (UGVs), some applications might currently be too complex or undesirable for full autonomy. Maintaining a human in the loop has proven to be a reliable strategy in these applications, yet there are currently limitations to the efficacy of human operators. Haptic feedback has been proposed as a method of addressing these limitations, and aiding UGV operators in safe and effective operation. This manuscript presents the experimental validation of LARIAT (Lowering Attention Requirements in semi-Autonomous Teleoperation), a portable haptic device for teleoperated semi-autonomous UGVs. This device utilizes an adapted predictive form of the Zero-Moment Point (ZMP) rollover index to inform haptic squeeze cues provided to the UGV operator for human-on-the-loop notifications. First, a brief design overview of LARIAT, implemented haptic control, and the ZMP index are presented. In addition to experimental device characterization of the just noticeable difference, we present a case study that demonstrates LARIAT's abilities to improve teleoperation performance. In an experiment involving a simulation of walking behind a semi-autonomous UGV, LARIAT reduced the number of UGV rollovers by up to 50%, with comparable or increased performance in a concurrent secondary tasks.