Optical Flow Based Robot Obstacle Avoidance

The objective of this research is to develop a real time obstacle detection and obstacle avoidance for autonomous navigation of mobile robots using a stereo camera in an unstructured environment. Autonomous navigation of mobile robots demands a) Exact determination of position and orientation of Robot and b) Accurate determination of size, shape, depth and range of potential obstacles in the environment. Simple kinematic model is used for mobile robot and stereo camera with pan and tilt provision is considered for long range operation. Complete 3D reconstruction of object/obstacle is obtained from the stereo matching algorithm and with triangulation method. The pose (position and orientation) of mobile robot is formulated from the static object observation with stereo reference matching points using RANSAC (RANdom SAmple Consensus) in successive frames. Potential field based obstacle avoidance formulation is carried out by using the obstacle range, size information, mobile robot position and orientation. Finally, proportional derivative navigation control loop along with obstacle avoidance algorithm is formulated and verified.

This paper studies the obstacle avoidance and navigation of mobile robots in unstructured environments. Because of the shortcomings of one single sensor, the system integrates the stereo vision sensor blumblebee2 with the lidar sensor to detect the information surrounding the mobile robot’s surroundings. And then through data fusion to get a more complete, more accurate scene distribution. Then use the improved system of ant colony optimization to increase the convergence speed and precision of the algorithm in robot path planning. Finally, the simulation experiment is carried out in the environment of Matlab and Visual Studio, and the physical experiment is carried out under the ROS platform. The experimental results show the feasibility and effectiveness of the proposed method.

In this paper is presented an algorithm for trajectory-tracking and obstacle avoidance for wheeled mobile robots (WMR). The proposed algorithm creates a trajectory composed of a global trajectory generated off-line and local obstacle avoidance trajectories that are created when an obstacle is detected by the sonar and laser sensors. Only one discrete-time sliding-mode controller is required to track the resulting trajectory and it does not require separate controllers for following the intended trajectory and avoiding the obstacle. Quantic equations are used to generate the paths used by the robot and calculating the velocity, acceleration, angular velocity and angular acceleration needed by the discrete-time sliding-mode controller to avoid the obstacles Realistic simulations and experimental results are performed to test the validity and performance of the proposed obstacle avoidance algorithm and the trajectory tracking controller.

An algorithm has been developed that efficiently solves a large class of robot navigation and obstacle avoidance problems using range information as its sole input from the environment. The system resides in a minicomputer and requires very small memory (1500 words) and computing time (1.35 s) allocations while solving simulated problems of broadly ranging spatial complexity and operational intricacy. It is thus a prime candidate for use in mobile robots or manipulators where real-time operation is desired.

In this paper, we propose a novel approach for obstacle avoidance. The configuration space of a nonholonomic mobile robot in the special Euclidean group SE(2). Therefore, we divide the whole configuration space into two sub-regions (i.e., a safe region and an obstacle avoidance region). We have derived two functions to define the robot's behavior in each region, which generate velocity vectors that drive the robot away from the obstacle and always towards the goal. We consider the existence of a switching surface between these two sub-regions, and specifically define this switching surface as a local subset of SE(2), i.e. the switching surface includes orientation, defined by the relative angle between the robot heading and bearing to the obstacle. Two different switching signals are proposed based on the switching surface. Lyapunov stability analysis proves the robot will converge to a goal position. To verify the effectiveness of the proposed approach, several simulations and experiments have been conducted.

Industry 4.0 paradigm is boosting the use of mobile robots in industrial applications. They must travel in areas with humans, obstacles and other vehicles. These robots are equipped with sensors such as lidars that allow them to perceive if there are obstacles close to them. This information can be exploited to avoid losses of performance. To do it, it is necessary to define obstacle avoidance algorithms to adjust the path of the robots maintaining a certain safety distance to the obstacles. In this work, an iterative obstacle avoidance for mobile robots is presented. The core idea is to enclose the obstacles in different bounding boxes in an iterative way and using some corners of the bounding boxes points to define the path. The algorithm also decides if the obstacles are avoided from the left or from the right. The algorithm has been intensively validated in simulation with positive results.

In this paper, a vision based adaptive controller is proposed for the obstacle avoidance of a micro aerial vehicle (MAV) by using the optical flow information. In order to employ the optical flow for indicating the distance between the MAV and surrounding, the multi-thread processing algorithm is proposed to reliably obtain the optical flow information uniformly diffused in the whole image frame. The MAV system is treated as a switched system while manipulating the different modes during the obstacle avoidance task. Based on the desired flying direction of obstacle avoidance obtained by the optical flow estimation, we design an adaptive controller such that the desired trajectory can be tracked under different switching modes. The simulations present the tracking response of the adaptive controller of a switched system, and the experiments of the overall system validate the collision-avoidance performance of the MAV.

Obstacle avoidance for mobile robot to reach the desired target from a start location is one of the most interesting research topics. However, until now, few works discuss about working of mobile robot in the dynamic and continuously changing environment. So, this issue is still the research challenge for mobile robots. Traditional algorithm for obstacle avoidance in the dynamic, complex environment had many drawbacks. As known that Q-learning, the type of reinforcement learning, has been successfully applied in computer games. However, it is still rarely used in real world applications. This research presents an effectively method for real time dynamic obstacle avoidance based on Q-learning in the real world by using three-wheeled mobile robot. The position of obstacles including many static and dynamic obstacles and the mobile robot are recognized by fixed camera installed above the working space. The input for the robot is the 2D data from the camera. The output is an action for the robot (velocities, linear and angular parameters). Firstly, the simulation is performed for Q-learning algorithm then based on trained data, The Q-table value is implemented to the real mobile robot to perform the task in the real scene. The results are compared with intelligent control method for both static and dynamic obstacles cases. Through implement experiments, the results show that, after training in dynamic environments and testing in a new environment, the mobile robot is able to reach the target position successfully and have better performance comparing with fuzzy controller.

This paper considers the real-time navigation and obstacle avoidance of mobile robots in complex environments. A novel local environment perception and navigation framework for mobile robot navigation based on the elevation information is proposed. A local elevation map is introduced to perceive and map obstacles in the local environment. Calculating obstacles’ heights and shapes by local elevation maps, the terrain traversability of the surrounding environment of a robot is determined, and a two-dimensional local occupancy grid map for robot path planning and navigation in local environment is generated. The effectiveness of our approach is validated experimentally through real-time navigation and obstacle avoidance in the artificial environment. The experimental results show that using the proposed framework, a mobile robot can safely reach its target position while avoiding the non-crossable area. For indoor mobile robots with limited computing power, in complex environments, it could provides more safe navigation and path-planning.

This paper presents new algorithms of path planning and obstacle avoidance for an autonomous mobile robot to navigate under unknown environments in the real time. Temporary targets are set up by distance variation method and then the algorithms of trajectory planning and obstacle avoidance are designed using fuzzy rules. It is shown by computer simulation that these algorithms are working well. Furthermore, an autonomous mobile robot was constructed to implement and test these algorithms in the real field. The experimental results are also satisfactory just like those of computer simulation.

Image segmentation is a key topic in computer vision, serving as a pre-step in a number of robotics tasks, including object recognition, obstacle avoidance and topological localization. In the literature, image segmentation has been employed as auxiliary information in order to improve optical flow performance. In this work, an alternative approach is proposed, in which optical flow information is used to aid image segmentation, aiming at scene understading for mobile robots. The proposed system performs dense optical flow analysis, followed by clustering of the optical flow vectors in a four dimensional space (formed by the x and y positions, angle and magnitude of each vector). Results from the clustering are used as ‘seeds’ in the segmentation process, performed by watershed segmentation in our implementation. In addition, the flow ‘image’ is combined with the original image, generating an image better suited for watershed segmentation, reducing the local minima effect often seen in this type of segmentation algorithms. The main pipeline considers the use of multi-modality cameras (visible and thermal-infrared). Since they see substantially different information, multi-modality further improves the amount of features of the resulting flows. Experimental results in urban and semi-urban scenarios with efficient segmentation illustrate the applicability of the method.

Image based Lung cancer diagnosis offers an effective and non-intrusive method for identification and diagnosis of lung cancer in a patient. This is done by segmentation of lung nodules from radiological images obtained from patients. This paper proposes a framework to segment nodule lesions in images. The novelty of the proposed framework is that we use optical flow information along with the morphological operations on images. The proposed framework applies the Farneback method of optical flow extraction algorithms for the nodule segmentation task. This work validates and demonstrates that optical flow information is useful for nodule segmentation and can be used in tandom with other features.

Robot navigation in dynamic unknown environments is a challenging issue in the field of autonomous mobile robot control. This paper presents a hybrid robust method for navigating an industrial robot in an environment that contains dynamic obstacles. The objectives are to find the shortest path, to minimize the energy consumption of robot, to make the smoothness of the generated paths and to tackle dynamic obstacles. Robots employed in industrial environments demand considerable autonomy and require high level of accuracy and manoeuvrability at the same time. Besides, no collision is tolerable along the way. A single-objective optimization method based on path criteria fails to satisfy all of the requirements. This paper proposes a hybrid algorithm including the whale optimization algorithm (WOA) for path planning, a learnable function approximation network for making smoothness of the generated paths and a fuzzy logic controller to avoid obstacle collision. In this algorithm, WOA optimizes the best path to be taken from the start to goal position. Once a sequence of points is candidate and segments of path are merged, a radial basis function is trained to provide a smooth movement path in the dynamic environment while trying to maximize the safety margin. To further improve the safety of navigation, a fuzzy-based obstacle avoidance algorithm is executed when the robot is placed in the vicinity of an obstacle. Fuzzy decisions are made based on values of distance information. The proposed hybrid method for path planning and obstacle avoidance issues was implemented and evaluated in dynamic environments including specific shaped obstacles. A GUI-based simulation platform was designed in Matlab environment for testing the proposed algorithm. Implementation results indicate that the proposed algorithm has yielded in smooth non-marginal goal-directed navigation with acceptable performance metrics. Meanwhile, collisions to dynamic obstacles were adaptively and non-rigidly avoided. Such a model-free hybrid algorithm for path planning and obstacle avoidance can improve autonomy in industrial operation and decrease computational complexity.

This paper presents a hierarchical control architecture for autonomous mobile robots (AMRs) that enhances obstacle avoidance and trajectory smoothness in industrial-like navigation tasks. A global A*-based planner generates a collision-free path on an occupancy grid map and triggers replanning when newly detected obstacles invalidate the current route; dynamic obstacles are handled by projecting their short horizon predicted occupancy as virtual walls on the grid. A timed elastic band (TEB) local planner refines the global path in the spatiotemporal domain to satisfy nonholonomic constraints, velocity and acceleration limits, and obstacle- avoidance requirements, producing a smooth reference trajectory with moderated curvature variations. For tracking, a four-wheel steering (4WS) 2-degrees of freedom (DOF) lateral dynamics model is controlled by a discrete-time linear quadratic regulator (LQR) feedback controller combined with an Ackermann-based feedforward term to proactively compensate the reference curvature and stabilize lateral and heading errors under steering constraints. Simulations in MATLAB and a high-fidelity multi-DOF vehicle dynamics simulation program compare five schemes under both static and dynamic obstacle scenarios: A* with Pure Pursuit (short/medium/long look-ahead), A* with LQR/feedforward, and the proposed A*-TEB-LQR/feedforward architecture. The proposed method successfully avoids obstacles and achieves the smallest average curvature and shortest arrival time among collision-free algorithms, while Pure Pursuit suffers from look-ahead sensitivity and A*-LQR/feedforward shows increased local curvature near obstacles.

Intelligent robotics is gaining significance in Maintenance, Repair, and Overhaul (MRO) hangar operations, where mobile robots navigate complex and dynamic environments for Aircraft visual inspection. Aircraft hangars are usually busy and changing, with objects of varying shapes and sizes presenting harsh obstacles and conditions that can lead to potential collisions and safety hazards. This makes Obstacle detection and avoidance critical for safe and efficient robot navigation tasks. Conventional methods have been applied with computational issues, while learning-based approaches are limited in detection accuracy. This paper proposes a vision-based navigation model that integrates a pre-trained Yolov5 object detection model into a Robot Operating System (ROS) navigation stack to optimise obstacle detection and avoidance in a complex environment. The experiment is validated and evaluated in ROS-Gazebo simulation and turtlebot3 waffle-pi robot platform. The results showed that the robot can increasingly detect and avoid obstacles without colliding while navigating through different checkpoints to the target location.