InfoFusion Controller

Abstract

In this paper, we propose the InfoFusion Controller, an advanced path planning algorithm that integrates both global and local planning strategies to enhance autonomous driving in complex urban environments. The global planner utilizes the informed Theta-Rapidly-exploring Random Tree Star (Informed- TRRT*) algorithm to generate an optimal reference path, while the local planner combines Model Predictive Control (MPC) and Pure Pursuit algorithms. Mutual Information (MI) is employed to fuse the outputs of the MPC and Pure Pursuit controllers, effectively balancing their strengths and compensating for their weaknesses.

The proposed method addresses the challenges of navigating in dynamic environments with unpredictable obstacles by reducing uncertainty in local path planning and improving dynamic obstacle avoidance capabilities. Experimental results demonstrate that the InfoFusion Controller outperforms traditional methods in terms of safety, stability, and efficiency across various scenarios, including complex maps generated using SLAM techniques.

Index Terms—Path Planning, MPC, Pure Pursuit, Mutual Information

InfoFusion Controller

Pipeline of InfoFusionController: (a) Controller (Step k-1): The vehicle's position at the previous time step ( x_k-1) is determined and passed to the environment. (b) Environment (Observation): The environment provides observation data ( Z _k ) such as obstacles and road conditions. (c) Global Planning (Informed TRRT):* The Informed TRRT* algorithm generates a global reference path ( T_ref) based on the observations and current state. (d) Local Planning (Pure Pursuit): The Pure Pursuit algorithm steers the vehicle toward a point on the reference path ( T_ref). (e) Local Planning (MPC): The MPC algorithm predicts optimal control inputs to follow the reference path over a dynamic horizon. (g) Synchronization (Mutual Information): Mutual Information (MI) combines state predictions from Pure Pursuit ( S_k^P ) and MPC ( S_k^M ) to select the best path. (h) Controller (Step k+1): The system executes the selected path, moving the vehicle to the next position ( x_k+1 ).

Obstacle Avoidance

In the left trajectory plot, InfoFusion demonstrates stable obstacle avoidance compared to other algorithms. While adaptive MPC and MPC basic show slight path distortions, and Pure Pursuit makes abrupt turns when navigating around dynamic objects, InfoFusion navigates smoothly with minimal deviation from the optimal path. This indicates that the InfoFusion controller maintains both safety and stability even when dynamic obstacles are introduced.

Driving Safety

In order to compare the safety of the algorithms, we were conducted in map_medium with a lot of cornering. MPC frequently oscillates back and forth excessively. This pattern can lead to unnecessary speed changes, reducing stability during driving. On the other hand, pure_pursuit shows more stable acceleration without significant fluctuations, but it has a tendency to overly reduce speed before cornering. In contrast, the info_fusion algorithm demonstrates a more consistent pattern compared to MPC.

BibTeX

@article{choi2025infofusioncontroller,
  author    = {Choi, Seongjun and Kim, Youngbum and Nam, Namwoo and Shin, Mansun and Chae, Byunggi and Lee, Sungjin},
  title     = {InfoFusionController: Informed TRRT Star with Mutual Information based on Fusion of Pure Pursuit and MPC for Enhanced Path Planning},
  journal   = {ICCE},
  year      = {2025},
}