[Paper Review] Optimizing Passenger Comfort in Cost Functions for Trajectory Planning

Jean Elsner|arXiv (Cornell University)|Nov 16, 2018

Vehicle Dynamics and Control Systems3 references4 citations

TL;DR

This paper investigates how passenger comfort can be optimized in autonomous vehicle trajectory planning by integrating human-centric comfort criteria—such as acceleration, jerk, and lateral forces—into cost functions. Using the CommonRoad framework, it evaluates weighted combinations of motion metrics (e.g., lane center offset, obstacle distance, curvature) and finds that minimizing jerk and lateral deviation yields the most comfortable trajectories, though curvature weighting has minimal impact in practice.

ABSTRACT

Current advances in the development of autonomous cars suggest that driverless cars may see wide-scale deployment in the near future. Research by both industry and academia is driven by potential benefits of this new technology, including reductions in fatalities and improvements in traffic and fuel efficiency as well as greater mobility for people who will or cannot drive cars themselves. A deciding factor for the adoption of self-driving cars besides safety will be the comfort of the passengers. This report looks at cost functions currently used in motion planning methods for autonomous on-road driving. Specifically, how the human perception of how comfortable a trajectory is can be formulated within cost functions.

Motivation & Objective

• To identify and quantify passenger comfort criteria relevant to autonomous vehicle motion planning.
• To evaluate how different cost function components influence perceived comfort in trajectory planning.
• To determine the relative impact of motion metrics—such as jerk, lateral deviation, and obstacle proximity—on passenger comfort using simulation-based analysis.
• To provide a data-driven, reproducible framework for comparing comfort-optimized cost functions in autonomous driving systems.

Proposed method

• The study uses the CommonRoad framework to define and combine partial cost functions, including acceleration, jerk, steering rate, lane center offset, obstacle distance, and curvature.
• It formulates the overall cost function as a weighted sum of basis cost functions: $ J_C = \sum_{i\in\mathcal{I}} w_i J_i $, enabling systematic tuning of comfort priorities.
• A simulation-based evaluation varies individual weights (e.g., $ w_d $, $ w_o $) while holding others constant to assess their influence on trajectory behavior.
• The analysis compares trajectories under different weight configurations using metrics like lane center offset, obstacle distance, speed, and curvature.
• It employs a benchmarking approach with fixed start/goal positions and obstacles to ensure reproducibility and isolate the effect of each weight.
• The study compares results to prior work (e.g., Arnay 2016, Chu 2012) using CommonRoad notation to rank cost function configurations.

Experimental results

Research questions

• RQ1 How do different motion metrics—such as jerk, lateral deviation, and obstacle proximity—affect passenger comfort in autonomous vehicle trajectories?
• RQ2 What is the relative influence of individual cost function weights on trajectory characteristics like lane keeping and obstacle avoidance?
• RQ3 To what extent does curvature weighting contribute to perceived comfort, given that smoothness is often considered a key comfort factor?
• RQ4 How do empirically derived weights compare to those from prior studies in terms of performance and comfort trade-offs?
• RQ5 Can a motion planner that prioritizes safety and rule compliance inherently deliver high passenger comfort?

Key findings

• The lane center offset weight ($ w_d $) has a strong inverse correlation with lane deviation, confirming its direct influence on path tracking.
• The obstacle distance weight ($ w_o $) increases trajectory cost in terms of lane center offset, indicating a trade-off between safety and path adherence.
• Speed and curvature costs show no clear dominant influence from any single weight, suggesting complex, concurrent interactions among motion metrics.
• The curvature cost ($ J_\kappa $) receives minimal weight (0.01–0.02) in high-performing configurations, indicating limited practical impact on comfort.
• The highest-ranked cost function configurations, such as $ J_{RA1} $ and $ J_{RA2} $, prioritize obstacle avoidance and path tracking over curvature, suggesting that smoothness is less critical than proximity and alignment.
• The study concludes that a robust, rule-compliant motion planner may inherently provide high comfort, reducing the need for explicit comfort terms in cost functions.