Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Efficient and Interaction-Aware Trajectory Planning for Autonomous Vehicles with Particle Swarm Optimization

Lin Song, David Isele, Naira Hovakimyan, Sangjae Bae

TL;DR

The paper tackles the problem of safe, smooth, interaction-aware lane-change trajectory planning in dense traffic. It proposes a PSO-based planner that leverages SGAN-predicted trajectories of surrounding vehicles and then applies polynomial smoothing to ensure dynamic feasibility, with safety enforced through a distance metric $h_i(x,y)\ge \epsilon$. The approach integrates a composite cost $f_{PSO}$ that balances reference tracking, heading, safety, comfort, and lane alignment, guided by NN predictions to shape the search. Real-time performance is demonstrated ( runtimes under $100$ ms in many cases ) and the method is shown to be competitive with, and sometimes superior to, baseline ADMM-NNMPC and MC-based strategies in terms of success rate and trajectory quality. This work advances practical, interaction-aware autonomous driving by uniting learning-based predictions with fast, derivative-free optimization and smooth trajectory generation.

Abstract

This paper introduces a novel numerical approach to achieving smooth lane-change trajectories in autonomous driving scenarios. Our trajectory generation approach leverages particle swarm optimization (PSO) techniques, incorporating Neural Network (NN) predictions for trajectory refinement. The generation of smooth and dynamically feasible trajectories for the lane change maneuver is facilitated by combining polynomial curve fitting with particle propagation, which can account for vehicle dynamics. The proposed planning algorithm is capable of determining feasible trajectories with real-time computation capability. We conduct comparative analyses with two baseline methods for lane changing, involving analytic solutions and heuristic techniques in numerical simulations. The simulation results validate the efficacy and effectiveness of our proposed approach.

Efficient and Interaction-Aware Trajectory Planning for Autonomous Vehicles with Particle Swarm Optimization

TL;DR

The paper tackles the problem of safe, smooth, interaction-aware lane-change trajectory planning in dense traffic. It proposes a PSO-based planner that leverages SGAN-predicted trajectories of surrounding vehicles and then applies polynomial smoothing to ensure dynamic feasibility, with safety enforced through a distance metric . The approach integrates a composite cost that balances reference tracking, heading, safety, comfort, and lane alignment, guided by NN predictions to shape the search. Real-time performance is demonstrated ( runtimes under ms in many cases ) and the method is shown to be competitive with, and sometimes superior to, baseline ADMM-NNMPC and MC-based strategies in terms of success rate and trajectory quality. This work advances practical, interaction-aware autonomous driving by uniting learning-based predictions with fast, derivative-free optimization and smooth trajectory generation.

Abstract

This paper introduces a novel numerical approach to achieving smooth lane-change trajectories in autonomous driving scenarios. Our trajectory generation approach leverages particle swarm optimization (PSO) techniques, incorporating Neural Network (NN) predictions for trajectory refinement. The generation of smooth and dynamically feasible trajectories for the lane change maneuver is facilitated by combining polynomial curve fitting with particle propagation, which can account for vehicle dynamics. The proposed planning algorithm is capable of determining feasible trajectories with real-time computation capability. We conduct comparative analyses with two baseline methods for lane changing, involving analytic solutions and heuristic techniques in numerical simulations. The simulation results validate the efficacy and effectiveness of our proposed approach.
Paper Structure (23 sections, 7 equations, 7 figures, 2 tables, 2 algorithms)

This paper contains 23 sections, 7 equations, 7 figures, 2 tables, 2 algorithms.

Table of Contents

  1. INTRODUCTION
  2. SYSTEM AND MODEL
  3. Vehicle Model
  4. Prediction Model
  5. Particle Swarm Optimization (PSO)
  6. APPROACH
  7. Trajectory Planning
  8. Particle Swarm Optimization (PSO) and Cost Function
  9. Swarm Formulation
  10. Cost Function for Evaluation
  11. Collision Avoidance in PSO
  12. Safety Constraints with SGAN Prediction
  13. Metrics for Inter-Vehicle Distance
  14. Safety Specification for Collision Evaluation
  15. NN Constraints as Guidance for PSO
  16. ...and 8 more sections

Figures (7)

  • Figure 1: Motivation example: lane change in dense traffic.
  • Figure 2: Illustration of the inter-vehicle distance metric.
  • Figure 3: The overview architecture of PSO with the integrated SGAN prediction module (left) and an illustrative example of the evolution process of particles in PSO which are also updated based on the NN outputs (right).
  • Figure 4: Illustrative example of modifying planned trajectory using Monte-Carlo sampling.
  • Figure 5: Illustration of an initial trajectory plan that can cause a collision with other vehicles.
  • ...and 2 more figures

Theorems & Definitions (3)

  • Remark 1
  • Remark 2
  • Remark 3