Interaction-Aware Vehicle Motion Planning with Collision Avoidance Constraints in Highway Traffic

TL;DR

Addresses collision-free, interaction-aware motion planning for autonomous vehicles in highway traffic. It integrates a Markov-chain trajectory predictor with a Pontryagin Minimum Principle (PMP) based optimizer, formulating the problem via a Hamiltonian $H$ and co-state dynamics while enforcing state constraints $h(x(t_d),t_d)=0$. A 2DOF ego-vehicle model is optimized to minimize acceleration energy $L_e(u_e,t)= \frac{1}{2}(u_x^2+u_y^2)$ under collision-avoidance constraints derived from predicted neighbor trajectories, using cubic polynomials for longitudinal paths and jump conditions at constraint times. Simulations on a two-lane highway under varying neighbor maneuvers demonstrate adaptive, collision-free trajectories that respect interactions with surrounding traffic.

Abstract

This paper proposes collision-free optimal trajectory planning for autonomous vehicles in highway traffic, where vehicles need to deal with the interaction among each other. To address this issue, a novel optimal control framework is suggested, which couples the trajectory of surrounding vehicles with collision avoidance constraints. Additionally, we describe a trajectory optimization technique under state constraints, utilizing a planner based on Pontryagin's Minimum Principle, capable of numerically solving collision avoidance scenarios with surrounding vehicles. Simulation results demonstrate the effectiveness of the proposed approach regarding interaction-based motion planning for different scenarios.

Figures (9)

• Figure 1: Optimal trajectory with state-equality constraints.
• Figure 2: Collision avoidance maneuver during lane change on a road with two lanes. The ego vehicle is depicted in green, the $1$-surrounding vehicle in yellow, and the $2$-surrounding vehicle in red. The red boxes surrounding the vehicles indicate safety-critical regions.
• Figure 3: Optimal trajectory planning with collision avoidance constraints, coupled with the predicted trajectory of surrounding vehicles.
• Figure 4: The position of the ego vehicle and surrounding vehicle 1 is recorded every 0.5 seconds, denoted by the $(x, y)$ coordinates, where surrounding vehicle 1 travels at a constant velocity on the adjacent road (scenario 1).
• Figure 5: The position of the ego vehicle and surrounding vehicle 1 is recorded every 0.5 seconds, denoted by the $(x, y)$ coordinates, where surrounding vehicle 1 decelerates on the adjacent road (scenario 2).