Risk-Aware Obstacle Avoidance Algorithm for Real-Time Applications

Ozan Kaya, Emir Cem Gezer, Roger Skjetne, Ingrid Bouwer Utne · 2026

Read Risk-Aware Obstacle Avoidance Algorithm for Real-Time Applications by Ozan Kaya, Emir Cem Gezer, Roger Skjetne, Ingrid Bouwer Utne (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-09204.txt.

Introduction

Autonomous surface vessels in changing seas must navigate under uncertainty, and Ozan Kaya and three co-authors offer a hybrid risk-aware architecture for the job. The system builds probabilistic risk maps capturing obstacle proximity and the behavior of moving objects, then plans with a risk-biased rapidly exploring random tree, refined by B-spline smoothing for continuous trajectories.

Three rewiring modes trade off path length against risk: shortest path, lowest risk, or a combination. Tests with static and dynamic obstacles show safe navigation, smooth paths, and adaptation as risks shift. Against conventional lidar-only or vision-only navigation, the method improves operational safety and autonomy, a promising step for uncertain marine missions.

Catalog record

CREATOROzan Kaya, Emir Cem Gezer, Roger Skjetne, Ingrid Bouwer Utne
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE23 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-09204

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Risk-Aware Obstacle Avoidance Algorithm for Real-Time Applications Ozan Kaya, Emir Cem Gezer, Roger Skjetne, Ingrid Bouwer Utne Abstract Robust navigation in changing marine environments requires autonomous systems capable of perceiving, reasoning, and acting under uncertainty. This study introduces a hybrid risk-aware navigation architecture that integrates probabilistic modeling of obstacles along the vehicle path with smooth trajectory optimization for autonomous surface vessels. The system constructs probabilistic risk maps that capture both obstacle proximity and the behavior of dynamic objects. A risk-biased Rapidly Exploring Random Tree (RRT) planner leverages these maps to generate collision-free paths, which are subsequently refined using B-spline algorithms to ensure trajectory continuity. Three distinct RRT* rewiring modes are implemented based on the cost function: minimizing the path length, minimizing risk, and optimizing a combination of the path length and total risk. The framework is evaluated in experimental scenarios containing both static and dynamic obstacles. The results demonstrate the system’s ability to navigate safely, maintain smooth trajectories, and dynamically adapt to changing environmental risks. Compared with conventional LIDAR or vision-only navigation approaches, the proposed method shows improvements in operational safety and autonomy, establishing it as a promising solution for risk-aware autonomous vehicle missions in uncertain and dynamic environments. keywords Obstacle avoidance; trajectory generation; path planning; RRT; risk mapping; maritime autonomous systems; autonomous ships; control engineering; ocean engineering; marine technology. 1 Introduction Autonomous Surface Vessels (ASVs) have been gaining a strong attention in the digital transformation of the marine sector, offering safer, cleaner, and more efficient operations ( 13 ) . Recent advances in relevant technologies have significantly improved their capability to perform missions such as monitoring, inspection, and transportation. However, robust and safe navigation in real world remains a major research challenge. The primary cause is because the consequences of hazardous in the marine domain can be very high. To achieve higher levels of autonomy ASVs must not only detect and track obstacles but also reason about environmental uncertainty and act safely while considering the risk picture ( 15 ) . Risk in marine systems and operations encompasses a wide variety of design and operational hazards. For ASV relevant navigation hazards and hazardous events are collisions, groundings, and drift hazardouss ( 5 ) . Traditional risk assessment frameworks, such as those discussed by 14 , have typically focused on the design phase and assumed relatively static operating conditions. However, recent developments in dynamic and online Risk Assessment (RA) methodologies enable a more continuous updating of the risk spectrum ( 9 ; 16 ; 1 ) . Such approaches can be integrated directly into autonomous control loops which allows ASVs to adapt their mission dynamically in response to environmental and operational changes; see e.g. ( 15 ; 10 ) . Probability theory is a key element of such dynamic risk assessment frameworks. One feasible method is Bayesian Belief Networks (BBNs) and related graphical models that provide a structured means of capturing dependencies between uncertain operational variables such as vessel velocity, obstacle proximity, control system failures, and environmental disturbances, e.g., ( 12 ; 6 ) . By frequent updating of the conditional probabilities based on situational awareness, these models can estimate the likelihood or risk in operation. Several studies have integrated accident probability estimations into path planning algorithms. 11 and 2 proposed risk-weighted A* and D* methods for underwater and surface vehicle navigation, showing improved collision avoidance under environmental uncertainties. 19 extended this concept to ship navigation using a modified A* algorithm that considered both environmental hazards and vessel dynamics. Similarly, 4 employed fuzzy logic and genetic optimization to enhance risk-aware collision avoidance performance. However, risk is typically included as a fixed cost or heuristic, which reduces the transparency of its interpretation and adaption during planning and drives the need for more open formulations that directly address risk 17 . Consequently, there is a need for integrated frameworks that combine probabilistic reasoning, adaptive risk modeling, and continuous trajectory optimization for real-world applications. This study accounts for identified hazardous events by estimating their probabilities of occurrence using a Bayesian Belief Network (BBN), resulting in a multilayered risk-aware navigation architecture that integrates probabilistic risk modeling with smooth trajectory optimization. The system constructs probabilistic risk maps by fusing environmental observations (e.g., radar, LIDAR, and tracking data) with predicted behaviors of dynamic obstacles and exploring the navigation space. The resulting hazardous event probability map captures both static and dynamic variations and is used for navigation decisions 10 . The path planning is achieved using a Risk-Aware Rapidly Exploring Random Tree (RA-RRT*) algorithm that incorporates the probabilistic risk field directly into its sampling and cost functions, considering the advantages of RRT* as investigated by 8 . To this end, we have developed a RA-RRT* path planning algorithm that incorporates hazardous probability estimations from a BBN and uses a weighting value to balance risk probability and length along the path. This enables flexible trajectory generation that can be tuned according to mission context, safety tolerance, or environmental conditions ( 15 ; 18 ) . We then generate smooth trajectories from initially sampled waypoints using B-splines which guarantees curvature continuity and eliminates sharp turning angles. This work contributes to a risk-aware obstacle avoidance framework that bridges probabilistic risk modeling with adaptive trajectory planning for ASVs. By integrating a BBN model, RRT*, and smooth trajectory optimization, the system achieves safer operation in dynamic marine environments. This paper is structured as follows: Section 2 lays down the foundation and talks about the preliminaries. Section 3 offers a methodology, including an overview of the proposed framework, the BBN model, and Risk-Aware RRT* (RA-RRT*) path planning for the trajectory. Furthermore, this section presents evaluation scenarios according to different cost function and rewiring methods. Section 4 presents the results and discussions, followed by a conclusion summarizing the key findings and outlining avenues for future research. 2 Preliminaries 2.1 Risk definition In this work, risk is represented as a collection of triplets describing hazardous events, their probabilities, and consequences ( 7 ) . The notation follows as presented by 11 . Let the set of hazardous events be denoted as R=\left\{E_{i},\ \Pr\{E_{i}\},\ C_{i}\right\}_{i=1}^{n_{R}}, (1) where E_{i} represents hazardous event i , \Pr\{E_{i}\} is the probability of that event, and C_{i} denotes the consequences. When the set includes all relevant hazardous events identified through a systematic risk management process, it is considered to provide a complete representation of risk. 2.2 Bayesian Belief Networks BBNs provide a graphical probabilistic representation of causal dependencies affecting collision or grounding. Each node represents an influencing factor. This model enables reasoning under uncertainty by continuously updating belief states when new evidence becomes available. 2.3 Problem Definition Let the path planning problem be defined on a directed graph \mathcal{G=(V,A)} , where \mathcal{V} denotes the set of nodes and \mathcal{A\subseteq V\times V} denotes the set of directed edges. A path p is defined as an ordered sequence of edges connecting a start node x_{s}\in V to a goal node x_{g}\in V . Let \mathcal{P}(x_{s},x_{g}) denote the set of all feasible paths from x_{s} to x_{g} . For a given path p\in\mathcal{P}(x_{s},x_{g}) , \mathcal{A}(p)\subseteq\mathcal{A} denote the set of edges traversed by p . Finally, let \Pr\{E_{i}\mid a\} denote the probability of hazardous event E_{i} occurring while traversing edge a(u,v)\in\mathcal{A} . Therefore we write,

Read the complete text