Focus area III: Planning

In real-world driving, an automated vehicle must continuously decide how to move safely and efficiently while accounting for many sources of uncertainty from perception and prediction modules. Given these uncertainties and complexities in real-world environments, ensuring reliable and robust motion planning remains a critical challenge.
The goal of the focus area Planning is to develop decision-making and motion-planning methods that explicitly account for uncertainties arising from perception and prediction. To address this, complementary approaches are investigated, including classical search- and sampling-based planners as well as deep reinforcement learning-based methods. By evaluating these approaches, their respective strengths can be analyzed and combined to form robust hybrid solutions, including effective fallback strategies for highly uncertain situations. In addition, the focus area examines the fundamental trade-offs between safety, uncertainty, and criticality in the planning process. By systematically analyzing and incorporating these trade-offs, the focus area Planning contributes to the development of trustworthy, transparent, and reliable decision-making mechanisms for automated vehicles.