GTP-UDrive: Unified Game-Theoretic Trajectory Planner and Decision-Maker for Autonomous Driving in Mixed Traffic Environments

TL;DR

GTP-UDRIVE addresses the challenge of autonomous-vehicle behavior in mixed traffic by unifying trajectory planning and decision-making within a game-theoretic framework. It adopts clothoid-based, piecewise trajectories represented by four decision variables (the two remaining waypoints) and solves for optimality using a PSO-based generalized Nash equilibrium approach that accounts for human-driver intentions. The method emphasizes safety through obstacle bounding boxes, SAT-based collision checks, and an adaptive elliptic safety zone, while maintaining efficiency via a defined I^{\text{eff}} metric; the payoff combines safety and efficiency with a constraint-penalty term. Experimental validation includes real-vehicle testing at an unsignalized intersection and Matlab simulations, demonstrating that the ego vehicle can either negotiate crossing with optimized waypoints or yield when the opponent acts first, thereby reducing collisions or unnecessary stops. Overall, GTP-UDRIVE provides a scalable, interpretable framework for integrated decision-making and trajectory optimization in mixed-traffic intersections with potential practical impact for safer, more human-friendly autonomous driving.

Abstract

Understanding the interdependence between autonomous and human-operated vehicles remains an ongoing challenge, with significant implications for the safety and feasibility of autonomous driving.This interdependence arises from inherent interactions among road users.Thus, it is crucial for Autonomous Vehicles (AVs) to understand and analyze the intentions of human-driven vehicles, and to display behavior comprehensible to other traffic participants.To this end, this paper presents GTP-UDRIVE, a unified game-theoretic trajectory planner and decision-maker considering a mixed-traffic environment. Our model considers the intentions of other vehicles in the decision-making process and provides the AV with a human-like trajectory, based on the clothoid interpolation technique.% This study investigates a solver based on Particle Swarm Optimization (PSO) that quickly converges to an optimal decision.Among highly interactive traffic scenarios, the intersection crossing is particularly challenging. Hence, we choose to demonstrate the feasibility and effectiveness of our method in real traffic conditions, using an experimental autonomous vehicle at an unsignalized intersection. Testing results reveal that our approach is suitable for 1) Making decisions and generating trajectories simultaneously. 2) Describing the vehicle's trajectory as a piecewise clothoid and enforcing geometric constraints. 3) Reducing search space dimensionality for the trajectory optimization problem.

Figures (5)

• Figure 1: Trajectory description: Starting from an initial position $\textbf{p}_{s} (x_{s},y_{s},\theta_{s})$ and aiming to reach the desired destination $\textbf{p}_{g} (x_{g},y_{g},\theta_{g})$, we investigate the conflict zone space to generate feasible trajectories
• Figure 2: Experimental Autonomous test vehicle of Institut VEDECOM
• Figure 3: All.des Marronniers intersection: Comparison between recorded left turn human driver trajectories, and GTP-UDRIVE generated trajectory (in dashed red)
• Figure 4: a) The nominal case at impact time $\textbf{t}_\textbf{impact}$, b) Post-application of GTP-UDRIVE at $\textbf{t}_\textbf{crit}$, and c) The evolution of the gap to collision GTC.
• Figure 5: Case 2: Ego vehicle stops between ${{t}_{1}}= 2.85 (s)$ and ${{t}_{2}}= 5.76 (s)$ until the opponent vehicle has safely cleared the conflict zone.