Autonomous vehicles are thought of as tools for increasing safety on the roads. However, automated vehicles have on multiple occasions been involved in accidents. This thesis is centered around the experience of safety drivers who test autonomous trucks. The aim is to improve safety for safety driver. I approach the task with acceptance of the fact that errors will occur due to being human (formerly known as human error). Studies in the field of human factors indicate that automation can result in decreased levels of situation awareness and mental underload, and overtrust or undertrust. There is also an increased risk of mode confusion and difficulties sustaining attention, which is a problem since the automation level of the vehicle in question is rising and the safety driver will still be neededfor the foreseeable future. I take a research through design approach and use observation,contextual interviews, co-creation, workshops, the wizard of oz method, enactment, back-casting and user testing. Through these methods, my participants and I identify critical traffic situations and propose design solutions that can make the work of the safety driver safer. The proposed design solutions are as follows: reducing the mental workload of the handover process by simplifying it; adding a feature that allows safety drivers to see what the system perceives through its sensors, which can improve their situational awareness (SA) and foster appropriate trust levels; using a colored bar on a screen behind the steerin wheel to communicate the mode; and enabling safety drivers to receive instructions for engaging auto mode.