Another Point of View: The Effects of Camera Perspective and Augmentation on the Performance of Remote Assistants of Highly Automated Vehicles

Kurzfassung

When operated under real-world conditions, highly automated vehicles (HAVs, SAE Level 4) face many traffic situations they cannot cope with, e.g., adverse weather. Remote assistance may aid HAVs in such situations, making HAV operations more robust. In this task context, human-machine interfaces (HMIs) for remote assistants of HAVs often present traffic situations from a camera perspective similar to the driver's. However, this first-person view has been associated with shortcomings including the occlusion of relevant objects on the road or the distortion of distance and angle perception. These shortcomings may affect the performance of the remote assistant. An experimental lab study with 37 participants was carried out to investigate if three different camera perspectives affect performance of remote assistants, their situation awareness, and other variables pertaining to them. The task revolves around determining the point in time when a complex left-turn driving task can be implemented by the supervised HAV at a busy urban intersection with mixed traffic. Additionally, the interplay of camera perspectives and video augmentation that visualizes additional sensor data was investigated in an environment with and without adverse weather due resulting from fog. Results indicated that certain performance indicators including decision time were affected by camera perspective. The positive and compensatory impact of augmentation under poor visibility conditions in adverse weather found in prior studies was replicated. Findings suggest that the most suitable perspective highly depends on the specific scenario. The results will help design context-sensitive HMIs for remote assistance of HAVs.