# Optimization-Based Path-Planning for Connected and non-Connected Automated Vehicles

**Authors:** Panagiotis Typaldos, Markos Papageorgiou, Ioannis Papamichail  
**arXiv:** [2104.06778](https://arxiv.org/abs/2104.06778)  
**Date:** 2021-04-14  
**Summary page:** https://www.summarizepaper.com/en/paper/2104.06778v1/

## Key points

- Optimization-based path-planning algorithm for connected and non-connected automated vehicles on multilane motorways
- Objective function incorporates vehicle advancement, passenger comfort, collision avoidance, road departures, and negative speeds
- Connected vehicles exchange real-time information with each other or the infrastructure
- Numerical solution utilizes an efficient feasible direction algorithm and a simplified Dynamic Programming algorithm
- Designed to be executed within a model predictive control (MPC) framework and embedded in the Aimsun microsimulation platform
- Testbed uses a homogeneous motorway section with three lanes and two levels of inflow simulated
- Vehicle trajectories and traffic conditions monitored over a 60-minute simulation horizon
- Automated vehicles have a planning horizon of 8 seconds with a path-planning step of 0.25 seconds
- Optimally controlled AVs drive closer to desired speed compared to ordinary vehicles
- Increased penetration rates of AVs improve overall traffic flow efficiency, benefiting manual vehicles as well
- Connected controlled vehicles demonstrate higher efficiency compared to non-connected ones due to improved real-time information and short-term prediction

## Summary

This study presents an optimization-based path-planning algorithm for connected and non-connected automated vehicles on multilane motorways. The algorithm is derived from an optimal control problem formulation, with the objective function incorporating terms related to vehicle advancement, passenger comfort, collision avoidance, road departures, and negative speeds. Connected vehicles are able to exchange real-time information about their last generated path with each other or the infrastructure. The numerical solution of the optimal control problem utilizes an efficient feasible direction algorithm, while a simplified Dynamic Programming algorithm provides the initial guess trajectory. The approach is designed to be executed within a model predictive control (MPC) framework and is embedded in the Aimsun microsimulation platform for realistic evaluation. The study uses a homogeneous motorway section as a testbed, with three lanes and two different levels of inflow simulated. Vehicle trajectories and traffic conditions are monitored over a 60-minute simulation horizon. Both automated and manually driven vehicles are randomly assigned characteristics such as vehicle type, dimensions, desired speed, initial lane, and time gap. For automated vehicles (AVs), a planning horizon of 8 seconds with a path-planning step of 0.25 seconds is used. Results show that optimally controlled AVs drive closer to their desired speed compared to ordinary Aimsun vehicles. Increased penetration rates of AVs improve the efficiency of overall traffic flow, benefiting manual vehicles as well. Additionally, connected controlled vehicles demonstrate higher efficiency compared to non-connected controlled vehicles due to improved real-time information and short-term prediction. In summary, this study presents an optimization-based path-planning algorithm which improves vehicle advancement goals while considering passenger comfort and collision avoidance objectives for connected and non-connected automated vehicles on multilane motorways. Results indicate increased efficiency in driving closer to desired speeds for optimally controlled AVs compared to ordinary vehicles; furthermore increased penetration rates of AVs enhance overall traffic flow efficiency with connected controlled vehicles demonstrating higher efficiency than non-connected ones.