Chapter 7 Path Planning and Decision-Making

Autonomous driving requires robust planning and decision-making to navigate complex environments while ensuring safety, efficiency, and comfort. This section provides a comprehensive overview of planning methods, encompassing global path planning\index{planning}, local trajectory generation, behavioral decision-making frameworks, and associated strategic methodologies. These capabilities collectively allow autonomous vehicles to interpret their surroundings, evaluate potential maneuvers, and determine optimal driving actions in dynamic contexts.

Planning techniques for autonomous systems are typically categorized into three principal classes: search-based, sampling-based, and optimization-based methods. Each class is grounded in distinct algorithmic paradigms. Search-based methods rely on graph traversal techniques to explore feasible paths within discretized representations of the driving environment. Sampling-based approaches construct solutions by incrementally sampling the state space, favoring probabilistic completeness in high-dimensional domains. Optimization-based methods formulate planning as a constrained optimization problem, often leveraging numerical solvers to compute trajectories that balance multiple objectives such as safety margins, smoothness, and energy efficiency.