A Survey on Hybrid Motion Planning Methods for Automated Driving Systems

TL;DR

The paper tackles the challenge of designing motion planning for automated driving systems that must handle diverse and unforeseen driving tasks. It presents a comprehensive taxonomy of hybrid motion planners, organized by how they combine traditional, learning-based, and cooperative methods to address TDM and TG, and analyzes them across key challenges such as vehicle dynamics, driving context, real-time feasibility, safety, and uncertainty. By reviewing roughly 50 hybrid approaches, the authors identify four main hybrid categories—decomposed geometry, potential field-based hybrids, optimization-based hybrids, and logical–learning combinations—plus cooperative planning, and discuss their relative strengths and limitations. The study highlights that hybrid methods can mitigate individual drawbacks of traditional planners, with a trend toward integrating learning for decision making and logic-based components for safety and explainability, while calling for more work on uncertainty robustness, general-purpose frameworks, and scalable V2X-enabled planning. This work provides researchers and industry practitioners with a structured view of hybrid motion planning options and points to practical research directions to advance real-time, safe, and generalizable ADS capabilities.

Abstract

Motion planning is an essential element of the modular architecture of autonomous vehicles, serving as a bridge between upstream perception modules and downstream low-level control signals. Traditional motion planners were initially designed for specific Automated Driving Functions (ADFs), yet the evolving landscape of highly automated driving systems (ADS) requires motion for a wide range of ADFs, including unforeseen ones. This need has motivated the development of the ``hybrid" approach in the literature, seeking to enhance motion planning performance by combining diverse techniques, such as data-driven (learning-based) and logic-driven (analytic) methodologies. Recent research endeavours have significantly contributed to the development of more efficient, accurate, and safe hybrid methods for Tactical Decision Making (TDM) and Trajectory Generation (TG), as well as integrating these algorithms into the motion planning module. Owing to the extensive variety and potential of hybrid methods, a timely and comprehensive review of the current literature is undertaken in this survey article. We classify the hybrid motion planners based on the types of components they incorporate, such as combinations of sampling-based with optimization-based/learning-based motion planners. The comparison of different classes is conducted by evaluating the addressed challenges and limitations, as well as assessing whether they focus on TG and/or TDM. We hope this approach will enable the researchers in this field to gain in-depth insights into the identification of current trends in hybrid motion planning and shed light on promising areas for future research.

Figures (8)

• Figure 1: The overall architecture of the control module for AVs in the modular architecture. The input and output signals are shown by I and O in brown and green colours, respectively. The functionality of the signals is explained in Section \ref{['bckg']}.
• Figure 2: Example illustration for the route, path, waypoints, and trajectory generation in a Cartesian global coordinate system. The state of the vehicle at time $t$ is defined in terms of its location $x, y$, orientation $\theta$ and their derivatives (e.g., acceleration, jerk, etc). The trajectory (red curve) and the path (blue curve) spatially coincide, but in this illustration, they are not drawn on top of each other for presentation clarity. The trajectory/path doesn't have to contain the waypoints, which only determine the high-level route planning.
• Figure 3: Example illustration for manoeuvre-based TDM (lane-change or lane-keeping) and TG (two candidate trajectories per TDM output).
• Figure 4: Classification of the traditional methods (elements of the hybrid methods) used for motion planning and control system in AVs.
• Figure 5: Example illustrating the sampling-based (b) vs the search-based (c) method for a lane change scenario (a). The search space in this illustration is occupancy grids in (c) and trajectory samples in (b) that are generated via parametric curves. Colours show the candidate (blue), dismissed (red), and accepted (green) trajectories.