Enhanced pure pursuit with dynamic steering control for autonomous mobile robots and application to safe navigation in chemical plants

Key findings from this study

• The study found that PP-DSC achieved 68–82% reduction in lateral tracking deviation compared to standard Pure Pursuit across diverse trajectory shapes in open-field conditions.
• The researchers demonstrate that Safety-integrated PP-DSC provided 11–17% tracking improvement in moderate-curvature sections of an industrial biodiesel plant simulation while maintaining less than 1% computational overhead.
• The authors report that algorithm performance diverged significantly between open environments and industrial settings, with standard Pure Pursuit outperforming the safety-integrated variant by 15.6% when tight turning radii constrained maneuvering.

Overview

Pure Pursuit with Dynamic Steering Control (PP-DSC) adaptively adjusts lookahead distance and velocity based on steering angle to enhance trajectory tracking in autonomous mobile robots. The algorithm was implemented on a four-wheeled steering-type autonomous mobile robot using ROS 2 Jazzy with real-time sensor fusion from GNSS-RTK, IMU, and wheel encoders. A safety-integrated variant incorporating Fire and Explosion Index-based velocity modulation was developed and evaluated for industrial deployment in hazardous environments.

Methods and approach

The PP-DSC algorithm modifies standard Pure Pursuit by making lookahead distance and velocity responsive to instantaneous steering angle. A four-wheeled steering-type autonomous mobile robot implemented the controller via ROS 2 Jazzy middleware. Real-time localization integrated GNSS-RTK, IMU, and wheel encoder measurements. Field experiments examined performance on straight, circular, and figure-eight trajectories at velocities ranging from 1.0 to 5.0 m/s across a 64 × 20 m test area. A Safety-integrated PP-DSC variant incorporated F&EI-based safety factors to modulate velocity in designated hazard zones. Simulation validation occurred in a 92 × 65 m empty fruit bunch biodiesel plant model with 5–9 m turning radii.

Results

The study found that PP-DSC achieved mean lateral deviations of 0.05 m, 0.07 m, and 0.08 m on straight, circular, and figure-eight paths respectively. These results represented 68–82% improvement over standard Pure Pursuit mean deviations of 0.19 m, 0.40 m, and 0.27 m. In the industrial simulation, standard Pure Pursuit outperformed Safety-integrated PP-DSC by 15.6% due to tight turning radius constraints. Safety-integrated PP-DSC demonstrated 11–17% improvement in moderate-curvature sections within the industrial setting. The F&EI-based safety integration imposed computational overhead of less than 1% while enabling automatic velocity reduction in hazard zones.

Implications

The results confirm that adaptive steering-responsive control mechanisms substantially enhance trajectory tracking accuracy in open-field environments with moderate to large turning radii. The gap between open-field and industrial performance suggests that algorithm design must account for environmental geometry; tight turning constraints inherent to chemical plant layouts may favor simpler control strategies or require further algorithmic refinement. The successful integration of Fire and Explosion Index-based safety factors into navigation control demonstrates a viable pathway for embedding Process Safety Management compliance into autonomous system operation without substantial computational burden.
Industrial deployment of autonomous mobile robots in hazardous facilities requires geometry-specific controller selection rather than universal algorithm adoption. The modest computational overhead of safety integration suggests feasibility for real-time implementation on resource-constrained robotic platforms. Further investigation into adaptive strategy switching based on detected path geometry could reconcile the performance divergence observed between open and constrained environments.

Scope and limitations

This summary is based on the study abstract and available metadata. It does not include a full analysis of the complete paper, supplementary materials, or underlying datasets unless explicitly stated. Findings should be interpreted in the context of the original publication.