This thesis focuses on motion planning algorithms for self-driving heavy-duty vehicles. Motion planning is a fundamental part of autonomous vehicles, tasked with finding the correct sequence of actions that take the vehicle towards its goal. This work focuses on the aspects that distinguish heavy-duty vehicles from passenger vehicles and require novel developments within motion planning algorithms. The proposed algorithms are studied in simulation environments and on two Scania prototype autonomous vehicles: a mining truck and a public transport bus.

We start by addressing the problem of finding the shortest paths for a vehicle in obstacle-free environments. This problem has long been studied, but the considered vehicle models have been simplistic. We propose a novel algorithm that plans paths respecting complex vehicle actuator constraints associated with the slow dynamics of heavy vehicles.

Using the previous method, we tackle the motion planning problem in environments with obstacles. Lattice-based motion planners, a popular choice for this type of scenario, come with drawbacks related to the sub-optimality of solution paths and the discretization of the goal state. We propose a novel path optimization method that significantly reduces both drawbacks. The resulting optimized paths contain less oscillatory behavior and arrive precisely at arbitrary non-discretized goal states.

We then study the problem of bus driving in urban environments. In order to successfully maneuver buses, distinct driving objectives must be used in planning algorithms. Moreover, a novel environment classification and collision avoidance scheme must be introduced. The result is a motion planning algorithm that mimics professional bus driver behavior, resulting in safer driving and increased vehicle maneuverability.

One particular challenge of driving in urban environments is common to buses and trucks with trailers, namely, that of centering the whole vehicle body on the road. In the case of buses, the long wheelbase introduces a conflict between centering the rear axle vehicle or centering the front axle. In the case of trucks with trailers, a similar conflict appears, this time between centering the truck body or centering the trailer body. We propose a framework to design motion planners that optimally trade-off between these conflicting objectives, resulting in planned paths that center the whole vehicle body, improving driving behavior.

Finally, we study the challenges of interacting with human-driven vehicles. We propose a motion planning framework that addresses the multi-modality of human behaviors, the interactive nature of traffic, and the impact of the autonomous vehicle on human drivers' decision making. The result is a motion planner that can reason about multiple future outcomes of a traffic scene, minimizing the expected cost across all outcomes. Furthermore, we show that incorporating neuroscience-based decision making models of human drivers into the motion planner results in the autonomous vehicle taking safe but assertive maneuvers, reducing the conservativeness usually seen in autonomous vehicles.