In order for autonomous surface vessels (ASVs) to avoid collisions at sea it is necessary predict the future trajectories of surrounding vessels. This paper investigate use historical automatic identification system (AIS) data such trajectories. The availability AIS have steadily increased in last years as a result more regulations, together with wider coverage through integration on satellites and land based receivers. Several AIS-based methods predicting vessel already exist. However,...

This paper presents a viable approach for incorporating collision avoidance strategies into existing guidance and control systems on marine vessels. We propose method that facilitates the use of simulation-based Model Predictive Control (MPC) (COLAV) Any COLAV strategy to be applied in real traffic must adhere international regulations preventing collisions at sea (COLREGS). The proposed MPC does not rely an accurate model system achieve vessel behaviors are compliant with COLREGS. Rather,...

Abstract We present results from sea trials for an autonomous surface vehicle (ASV) equipped with a collision avoidance system based on model predictive control (MPC). The were performed in the North Sea as part of ASV Challenge posed by Deltares through Dutch initiative involving different authorities, including Ministry Infrastructure and Water Management, Netherlands Coastguard, Royal Navy. To allow to operate maritime environment governed International Regulations Preventing Collisions...

This article presents a new algorithm for short-term maritime collision avoidance (COLAV) named the branching-course MPC (BC-MPC) algorithm. The is designed to be robust with respect noise on obstacle estimates, which significant source of disturbance when using exteroceptive sensors such as e.g. radars detection and tracking. Exteroceptive do not require vessel-to-vessel communication, enables COLAV toward vessels equipped automatic identification system (AIS) transponders, in addition...

Abstract In this paper, we summarize the experiences with autonomous passenger ferry development prototype milliAmpere, which has been used as a test platform in several research projects at Norwegian University of Science and Technology (NTNU) since 2017. New algorithms for motion planning, control, collision avoidance, docking, multi-target tracking localization have developed verified full-scale experiments milliAmpere. The infrastructure surrounding milliAmpere includes sensor rigs...

Sensor fusion plays a key part in autonomous surface vehicles, however, the high cost of sensors makes barrier entry this research field quite high. In work, we present complete system for sensor on milliAmpere ferry platform as well an open dataset maritime tracking across two environments. Individual and their detection pipelines are evaluated various metrics. We also evaluate performance both individually using multi-sensor extension JIPDA multi-target tracker. find that different...

When operating an autonomous surface vessel (ASV) in a marine environment it is vital that the equipped with collision avoidance (COLAV) system. This system must be able to predict trajectories of other vessels order avoid them. The increasingly available automatic identification (AIS) data can used for this task. In paper, we present data-driven approach positions 5–15 minutes into future using AIS data. predictions are given as Gaussian Mixture Models (GMMs), thus give measure uncertainty...

Collision avoidance systems are a key ingredient in developing autonomous surface vehicles (ASVs). Such require real-time information about the environment, which can be obtained from transponder-based or exteroceptive sensors located on ASV. In this paper, we present closed-loop collision (COLAV) system using maritime radar for detecting target ships, implemented 26 foot high-speed The was validated full-scale experiments Trondheimsfjorden, Norway, May 2017. probabilistic data association...

Harbor surveillance above and below the sea surface depends on sensors such as radar multibeam sonar. These attempt to detect track moderately observable targets small boats or human divers in environments which often are characterized by heavy-tailed backgrounds. Target tracking is challenging even for moderate signal-to-noise ratios (SNRs) due increased frequency of target-like outliers. A method operating an environment should exploit much data possible ensure robustness. Still,...

We propose a collision avoidance method that incorporates the interactive behavior of agents and is proactive in dealing with uncertainty future obstacles. The proposed considers interactions will be experienced by an autonomous surface vessel (ASV) environment governed international regulations for preventing collisions at sea (COLREGs). Our approach aims encouraging dynamic obstacles to cooperate according COLREGs. Therefore, we strategy assessing cooperative obstacles, result assessment...

In this article, a new approach for ship-ship collision probability estimation based on the Cross-Entropy (CE) method is introduced, which can be treated as an adaptive importance sampler. It has advantage of attaining low variance estimates small probabilities, will most often case in realistic scenarios. Furthermore, risk-based Collision Avoidance (COLAV) system being able to take obstacle kinematic uncertainty and intention into account presented, namely Probabilistic Scenario-Based Model...

The multiple hypothesis tracker (MHT) and finite set statistics (FISST) are two approaches to multitarget tracking, which both have been heralded as optimal. In this paper, we show that the Bayes filter with basis in FISST can be expressed terms MHT formalism, consisting of association hypotheses corresponding probabilities hypothesis-conditional densities targets. Furthermore, resulting MHT-like method under appropriate assumptions (Poisson clutter birth models, no target death,...

Autonomous surface vehicles and maritime autonomous ships must rely on sense-and-avoid systems for navigating safely among other ships. The main objective of this paper is to present examples such systems, their verification in full-scale collision avoidance experiments as part the research project "Sensor fusion vehicles" (Autosea). Lessons learned from progression have led increasing robustness methods, provide a foundation several important topics further near future.

The open wording of the traffic rules sea, COLREGS, and existence unwritten rules, make it essential for an autonomous ship to understand intentions other ships. This article uses a dynamic Bayesian network (DBN) model infer ships in waters based on their observed real-time behavior. Multiple intention nodes are included describe different ways can interpret conflict with behavioral outlined COLREGS. prior probability distributions adapted current situation observable characteristics such as...

Verification is a necessary step in system development, so too for autonomous surface vehicles (ASVs). However, formal and analytical verification methods are not well suited such highly complex systems. Simulation-based testing has therefore been proposed as viable approach. This would require numerous simulations to be performed evaluated, raising the need automatic evaluation. An important assessment measure with regards ASV safety, degree of International Regulations Preventing...

Abstract Reliable anti-collision control algorithms conforming with the rules regulating traffic at sea, International Regulations for Preventing Collisions Sea (COLREG), are essential deployment of autonomous vessels in waters shared other ships. The development such methods is an active field research. However, little attention has been given to how these interpreted by experienced mariners, and information can be parametrised use automatic systems This paper presents a method exploiting...

A Dynamic Positioning (DP) control system is developed, implemented and tested for a scale model of DNV GL's concept ship ReVolt. This model-scale used as test platform sensors systems in autonomous vessels, this paper focuses on the functionality implementation components required to achieve DP capabilities ReVolt ship. The consists 3-Degree Freedom reference filter, Proportional-Integral-Derivative (PID) controller with model-based feedforward, thrust allocation module from GL. ReVolt's...

Testing that ships are compliant to specified safety requirements have traditionally relied on real world data, which is not scalable and limited testable scenarios due financial ethical reasons. Low fidelity simulations been used counteract some of these problems, sufficient for emulating simpler systems such as radar detectors, but testing complex found in computer vision. In the automotive industry use game engines shown be a great platform their customizability, combination real-time...

In this article, we demonstrate how a variation of the joint integrated probabilistic data association with interacting multiple models and visibility state can be derived as special case Poisson multi-Bernoulli mixture filter hybrid representation standard approximations. The proposed method is tested on two radar sets that were recorded during maritime collision-avoidance experiments.

Collision Avoidance (COLAV) for autonomous ships is challenging since it relies on track estimates of nearby obstacles which are inherently uncertain in both state and intent. This uncertainty must be accounted the COLAV system order to ensure safe efficient operation vessel accordance with traffic rules. Here, a built Scenario-based Model Predictive Control (SB-MPC) dynamic probabilistic risk treatment presented. The probability collision all using combination Monte Carlo simulation (MCS)...

In recent years, the Poisson multi-Bernoulli mixture (PMBM) filter has been established among state-of-the art methods in target tracking. We present a method for including target-provided measurements said filter, both when using it to track extended objects and point targets. use messages from Automatic identification system as an example of measurements, radar LiDAR examples exteroceptive sensors. case, we utilize several different kinematic models parallel through interacting multiple...

Extended object tracking (EOT) plays an important role when creating autonomous systems like self-driving cars or surface vehicles. For accurate estimation of the target extent, it is to have a sensor with high resolution and low measurement noise. In this paper we use LIDAR track single elliptical in clutter contour model. The model enables extended Kalman filter (EKF) which combined generalized probabilistic data association (GPDA) filter. EKF method comes favorably out comparison random...