For greater autonomy of visual control-based solutions, especially applied to mobile robots, it is necessary to consider the existence of unevenness in the navigation surface, an intrinsic characteristic of several real applications. In general, depth information is essential for navigating three-dimensional environments and for the consistent parameter calibration of the visual models. This work proposes a new solution, including depth information in the visual path-following (VPF) problem, which allows the variation of the perception horizon at runtime while forcing the coupling between optical and geometric quantities. A new NMPC (nonlinear model predictive control) framework considering the addition of a new input to an original solution for the constrained VPF-NMPC allows the maintenance of low computational complexity. Experimental results in an outdoor environment with a medium-sized commercial robot demonstrate the correctness of the proposal.

Let G be a graph. Assume that to each vertex of a set of vertices $S\subseteq V(G)$ a robot is assigned. At each stage one robot can move to a neighbouring vertex. Then S is a mobile general position set of G if there exists a sequence of moves of the robots such that all the vertices of G are visited while maintaining the general position property at all times. The mobile general position number of G is the cardinality of a largest mobile general position set of G. We give bounds on the mobile general position number and determine exact values for certain common classes of graphs, including block graphs, rooted products, unicyclic graphs, Kneser graphs $K(n,2)$ and line graphs of complete graphs.

This paper tackles the challenge of the necessity of using the sequence of past environment states as the controller’s inputs in a vision-based robot navigation task. In this task, a robot has to follow a given trajectory without falling in pits and missing its balance in uneven terrain, when the only sensory input is the raw image captured by a camera. The robot should distinguish big pits from small holes to decide between avoiding and passing over. In non-Markov processes such as the abovementioned task, the decision is done using past sensory data to ensure admissible performance. Applying images as sensory inputs naturally causes the curse of dimensionality difficulty. On the other hand, using sequences of past images intensifies this difficulty. In this paper, a new framework called recurrent deep learning (RDL) with combination of deep learning (DL) and recurrent neural network is proposed to cope with the above challenge. At first, the proper features are extracted from the raw image using DL. Then, these represented features plus some expert-defined features are used as the inputs of a fully connected recurrent network (as target network) to generate command control of the robot. To evaluate the proposed RDL framework, some experiments are established on WEBOTS and MATLAB co-simulation platform. The simulation results demonstrate the proposed framework outperforms the conventional controller based on DL for the navigation task in the uneven terrains.

Recent development of sensor knowledge representation by the use of a certainty grid has been extensive and shown the usefulness of the grid-based concept for robot navigation.

Yet the methodology was not perfect. This paper introduces the Bayesian formula into the certainty grid representation to overcome some difficulties of ad hoc formula that has been the only way of updating the grids. The complete derivation of the proposed updating formula is given and proved to be able to accurately identify the simulated world. Also, the paper suggests two updating models: context-sensitive and context-free. Both of them were shown to be usable through simulation in real world modeling.

Several non-holonomic Dubins-car-like robots travel over paths with bounded curvatures in a plane that contains an a priori unknown region. The robots are anonymous to one another and do not use communication facilities. Any of them has access to the current minimum distance to the region and can determine the relative positions and orientations of the other robots within a finite and given visibility range. We present a distributed navigation and guidance strategy under which every robot autonomously converges to the desired minimum distance to the region with always respecting a given safety margin, the robots do not collide with one another and do not get into clusters, and the entire team ultimately sweeps over the respective equidistant curve at a speed exceeding a given threshold, thus forming a kind of a sweeping barrier at the perimeter of the region. Moreover, this strategy provides effective sub-uniform distribution of the robots over the equidistant curve. Mathematically rigorous justification of the proposed strategy is offered; its effectiveness is confirmed by extensive computer simulations and experiments with real wheeled robots.

We present a novel framework for collision free assisted navigation of a semi-autonomous vehicle in complex unknown environments with moving and steady obstacles. In the proposed system, a semi-autonomous vehicle is guided by a human operator and an automatic reactive navigator. The autonomous reactive navigation block takes control from the human operator in situations where there is the danger of collision with obstacle. A mathematically rigorous analysis of the proposed approach is provided. The performance of the proposed assisted navigation system is demonstrated via experiments with a real semi-autonomous hospital bed and extensive computer simulations.

This paper presents a method to calculate the fusing rule among three reactive behaviors, Wall Following, Corridor Following and Door Crossing, from place characterization for robot navigation. The technique is supported by a local grid of the closest area to the robot, which is built from sonar readings. The contour of this grid is extracted, represented by its FFT and, finally, it is reduced to a short feature vector with a principal component analysis (PCA). This feature vector is used to decide the fusing rule among the three behaviors. The algorithm is very fast in terms of its time performance, being then valid to be used in robot navigation, since the robot would rapidly react to new situations. It has also been successfully tested in simulated and real environments, with a Pioneer robot equipped with eight frontal sonar sensors, both in manually driven tasks and autonomous navigation tasks, proving its feasibility and effectiveness.

We determine the shortest (minimal in length) path on a unicycle-like mobile robot in a known environment with smooth (possibly non-convex) obstacles with a constraint on curvature of their boundaries. Furthermore, we propose a new reactive randomized algorithm of robot navigation in unknown environment and prove that the robot will avoid collisions and reach a steady target with probability 1. The performance of our algorithm is confirmed by computer simulations and outdoor experiments with a Pioneer P3-DX mobile wheeled robot.

Robot navigation is a large area of research, where many different approaches have already been tried, including navigation based on visual memories. The Sparse Distributed Memory (SDM) is a kind of associative memory based on the properties of high-dimensional binary spaces. It exhibits characteristics, such as tolerance to noise and incomplete data, ability to work with sequences and the possibility of one-shot learning. Those characteristics make it appealing to use for robot navigation. The approach followed here was to navigate a robot using sequences of visual memories stored into a SDM. The robot makes intelligent decisions, such as selecting only relevant images to store, adjusting memory parameters to the level of noise and inferring new paths from the learnt trajectories. The method of encoding the information may influence the tolerance of the SDM to noise and saturation. This paper reports novel results of the limits of the model under different typical navigation problems. The SDM showed to be very robust to illumination and scenario changes, occlusion and saturation. An algorithm to build a topological map of the environment based on the visual memories is also described.

When employing autonomous helicopters, it is desirable to use navigation approaches, which firmly ensure safety. In this paper, we propose and compare two approaches to navigation through environments containing obstacles. The first uses sliding mode boundary following to maintain a prespecified distance to obstacles, and the second uses a model predictive control approach to plan short horizon trajectories around detected objects, while ensuring that the helicopter is brought to a halt within the sensor visibility radius. The navigation approaches are subjected to analysis for robustness, and simulations are carried out with a realistic helicopter model for verification. Additional real-world experiments were performed with a wheeled robot to demonstrate potential for real-time application.

This paper describes an autonomous robot vehicle which can navigate through an initially unknown obstacle field to a nominated goal or systematically map its working environment. The navigation system uses combined ultrasonic beacon/odometry based localisation, optical range finders for environmental mapping, an A path planning procedure and continuous motion control. The computational support is divided between a graphics workstation 'home base' and a PC hosted transputer array on-board. The integration of all the subsystems cited above has been achieved and many successful navigation experiments completed. Possible further developments which would enhance the capabilities of the system are also discussed.

In this paper, we present a method for robot navigation toward a moving object with unknown maneuvers. Our strategy is based on the integration of the robot and the target kinematics equations with geometric rules. The tracking problem is modeled in polar coordinates using a two-dimensional system of differential equations. The control law is then derived based on this model. Our approach consists of a rendezvous course, which means that the robot reaches the moving goal without following its path. In the presence of obstacles, two navigation modes are integrated, namely the tracking and the obstacle-avoidance modes. To confirm our theoretical results, the navigation strategy is illustrated using an extensive simulation for different scenarios.

We combine a “hybrid” force-/position-control scheme with a potential field approach into a novel method for collision recovery and navigation in unknown environments. It can be implemented both on manipulators and mobile robots. The use of force sensors allows us to locally sense the environment and design a dynamic control law. Multiple Lyapunov functions are used to establish asymptotic stability of the closed-loop system. The switching conditions and stability criteria are established under reasonable assumptions on the type of obstacles present in the environment. Extensive simulation results are presented to illustrate the system behavior under the designed control scheme, and verify its stability, collision recovery, and navigation properties.

This paper deals with a method for robot navigation towards a moving goal. The goal maneuvers are not a priori known to the robot. Our method is based on the use of the kinematics equations of the robot and the goal combined with geometrical rules. First a kinematics model for the tracking problem is derived and two strategies are suggested for robot navigation, namely the velocity pursuit guidance law and the deviated pursuit guidance law. It turns out that in both cases, the robot's angular velocity is equal to the line of sight angle rate. Important properties of the navigation strategies are discussed and proven. In the presence of obstacles, two navigation modes are used: the tracking mode, which has a global aspect and the obstacle avoidance mode, which has a local aspect. In the obstacle avoidance mode, a polar diagram combining information about obstacles and directions corresponding to the pursuit is constructed. An extensive simulation study is carried out, where the efficiency of both strategies is illustrated for different scenarios.

This paper discusses the techniques and their applications in the development of a path planning system composed of three modules, namely: global vision (GVM), trajectory planning (TPM) and navigation control (NCM). The GVM captures and processes the workspace image to identify the obstacle and the robot configurations. These configurations are used by the TPM to generate the Voronoi roadmap, to compute the maximal clearance shortest feasible path and the visibility pathway between two configurations. The NCM controls the robot functionalities and navigation. To validate the path planning system, three sets of experiments have been conducted using the Lab robot Khepera, which have shown very good results.

The desert ant Cataglyphis is able to explore its desert habitat for hundreds of meters while foraging and return back to its nest precisely and in a straight line. With a body of less than 10 mm and a brain of less than one cubic millimeter they provide a challenge for hi-tech engineers. In this article, we give an overview of our attempts to model parts of its navigation behavior using robots and computer simulations. Inspired by the insect's navigation system we have developed mechanisms for path integration and visual piloting that were successfully employed on the mobile robot Sahabot 2.

This paper aims to develop a complete algorithm to determine the robot motion and the scene structure using a monocular vision system. It is based on straight lines and significant points extracted on them. In this way, an agreement between the problems to extract or to match points and the limitations of infinite lines to compute structure and motion is established. Many geometrical relations of the lines in the scene are exploited to clear up the coupling between the rotation and the translation of the camera. Several real images have been used to validate the proposed method. The algorithm has been considered for navigation of a mobile robot moving in man-made environments.