Algorithms for collision-free navigation of mobile robots in complex cluttered environments: a survey

Analysis

Why This Paper Matters

This survey, published in 2014, addresses a fundamental challenge in robotics: enabling unmanned vehicles to navigate safely through unknown, cluttered environments. At the time, the field was fragmented with many ad-hoc approaches, and this paper provided a structured overview of techniques that rigorously guarantee collision avoidance under certain assumptions. Its emphasis on reactive methods—those that rely on local sensory data rather than global maps—was particularly timely, as it aligned with the growing interest in autonomous vehicles operating in dynamic, unstructured settings.

The paper's significance is underscored by its 471 citations, indicating its role as a key reference for both researchers and practitioners. By systematically categorizing sensor models, vehicle kinematics, environment assumptions, and performance criteria, the survey helped establish a common framework for evaluating and comparing navigation algorithms. This has facilitated more principled development of collision avoidance systems, which are critical for applications ranging from warehouse robots to autonomous cars.

Technical Contributions

The survey's main contributions include:

• Comprehensive taxonomy: It organizes navigation methods into categories based on whether they handle stationary obstacles, moving obstacles, or multiple vehicles, and whether they are global or reactive.
• Rigorous collision avoidance: It highlights techniques that provide formal guarantees of safety, such as those based on velocity obstacles, potential fields, and sliding mode control.
• Focus on reactive methods: It argues for the practicality of local sensory data approaches, which are more scalable to unknown environments than global planning.
• Model predictive control (MPC): It reviews MPC-based navigation laws that optimize future trajectories while respecting constraints, offering a way to handle complex dynamics and obstacles.
• Sliding mode control: It discusses sliding mode approaches that are robust to uncertainties and disturbances, providing reliable collision avoidance in cluttered settings.

Results

As a survey, the paper does not present new experimental metrics. Instead, it synthesizes findings from numerous prior studies, comparing the trade-offs between different methods in terms of computational complexity, sensor requirements, and safety guarantees. For example, it notes that reactive methods are generally more computationally efficient but may suffer from local minima, while MPC methods offer better performance but require more computation. The survey also identifies gaps in the literature, such as the need for better handling of dynamic obstacles and multi-vehicle coordination.

Significance

The broader impact of this survey is its role in consolidating knowledge and guiding future research. By clearly articulating the assumptions and limitations of existing methods, it has helped researchers identify open problems and motivated the development of more robust algorithms. Its focus on reactive and predictive control has influenced subsequent work in model predictive control for autonomous driving and mobile robots. Moreover, the survey's emphasis on rigorous collision avoidance has contributed to the safety-critical design of autonomous systems, which is essential for real-world deployment. Even though the field has evolved with the advent of deep learning, this survey remains a valuable reference for understanding the foundational principles of collision-free navigation.