A Visual Servoing approach for road lane following with obstacle avoidance

de Lima, Danilo Alves; Victorino, Alessandro Correa · 2014 · OpenAlex-citations

DOI: 10.1109/itsc.2014.6957725

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the challenge of local navigation for autonomous car-like robots in urban environments, specifically focusing on road lane following combined with obstacle avoidance. Traditional Visual Servoing (VS) approaches often fail to account for velocity changes required to stop or avoid obstacles, while global navigation methods relying on GPS are susceptible to signal loss and noise. The authors propose a hybrid control strategy that integrates an Image-Based Visual Servoing (IBVS) controller for deliberative lane following with an Image-Based Dynamic Window Approach (IDWA) for reactive obstacle avoidance. This method validates VS-generated velocities against vehicle kinematic constraints and obstacle proximity, ensuring safe navigation without requiring global localization. The methodology is structured into two layers: workspace perception and navigation control. For perception, the system uses a monocular camera to extract visual features defining the road lane center, specifically the tangent of the path and its angular offset in the image plane. Obstacles are detected using laser sensors in simulations and stereo vision in real-world tests, mapped into a local occupancy grid. The navigation control layer employs a kinematic model of a front-wheel-drive vehicle. The IBVS controller computes desired linear and angular velocities to minimize image feature errors, guiding the vehicle to the lane center. These velocities are then validated by the IDWA, which searches for admissible velocities within a dynamic window defined by the vehicle’s acceleration limits and safe stopping distances relative to detected obstacles. The IDWA objective function prioritizes heading toward the visual target, distance from obstacles, and maintaining desired velocity. The approach was validated through both simulation and experiments with a full-sized autonomous electric vehicle. Simulation results demonstrated that while pure IBVS could follow lanes with small lateral errors, it lacked robustness for path reaching and obstacle avoidance. The hybrid VS+IDWA controller successfully avoided static obstacles while maintaining lane centering. When no obstacles were present, the hybrid system converged to the reference lane faster than the reactive IDWA alone. Real-world experiments on a non-structured circuit confirmed the system's viability, showing that the robot could correct feature detection errors and converge to the road center at a desired speed of 1 m/s, even with variations in feature detection. The IDWA effectively modified VS outputs only when necessary to guarantee obstacle avoidance, otherwise preserving the deliberative control inputs. The significance of this work lies in its ability to combine the precision of visual servoing with the safety guarantees of reactive obstacle avoidance, creating a robust local navigation system for autonomous vehicles. By eliminating the need for global localization and integrating kinematic constraints directly into the visual control loop, the proposed method offers a practical solution for urban navigation tasks. The results demonstrate that this hybrid approach allows electric vehicles to perform safe and accurate lane following in cluttered environments, addressing a critical gap in existing visual navigation strategies that typically treat path following and obstacle avoidance as separate or switched tasks.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• perception action locomotion