scispace - formally typeset
Search or ask a question

Showing papers on "Obstacle avoidance published in 2018"


Journal ArticleDOI
TL;DR: This article is to present a comprehensive literature review of the vision-based methods for UAV navigation, specifically on visual localization and mapping, obstacle avoidance and path planning, which compose the essential parts of visual navigation.
Abstract: Research on unmanned aerial vehicles (UAV) has been increasingly popular in the past decades, and UAVs have been widely used in industrial inspection, remote sensing for mapping & surveying, rescui...

222 citations


Journal ArticleDOI
TL;DR: In this article, a real-time dynamic path planning method for autonomous driving that avoids both static and moving obstacles is presented, which determines not only an optimal path, but also the appropriate acceleration and speed for a vehicle.

215 citations


Journal ArticleDOI
25 Jun 2018
TL;DR: A novel, intuitively interpretable, 3-D point cloud representation called3-D modified Fisher vectors is proposed, which combines a coarse discrete grid structure with continuous generalized Fisher vectors and enables the design of a new CNN architecture for real-time point cloud classification.
Abstract: Modern robotic systems are often equipped with a direct three-dimensional (3-D) data acquisition device, e.g., LiDAR, which provides a rich 3-D point cloud representation of the surroundings. This representation is commonly used for obstacle avoidance and mapping. Here, we propose a new approach for using point clouds for another critical robotic capability, semantic understanding of the environment (i.e., object classification). Convolutional neural networks (CNNs), that perform extremely well for object classification in 2-D images, are not easily extendible to 3-D point clouds analysis. It is not straightforward due to point clouds’ irregular format and a varying number of points. The common solution of transforming the point cloud data into a 3-D voxel grid needs to address severe accuracy versus memory size tradeoffs. In this letter, we propose a novel, intuitively interpretable, 3-D point cloud representation called 3-D modified Fisher vectors. Our representation is hybrid as it combines a coarse discrete grid structure with continuous generalized Fisher vectors. Using the grid enables us to design a new CNN architecture for real-time point cloud classification. In a series of performance analysis experiments, we demonstrate competitive results or even better than state of the art on challenging benchmark datasets while maintaining robustness to various data corruptions.

187 citations


Journal ArticleDOI
TL;DR: The IABC algorithm is proposed to improve the evacuation efficiency and provide support for building design and evacuation management by employing the strategies of grouping and exit selection and uses the evacuation time of the individuals as the evaluation metric.

180 citations


Journal ArticleDOI
TL;DR: A concise deep reinforcement learning obstacle avoidance (CDRLOA) algorithm is proposed with the powerful deep Q-networks architecture to overcome the usability issue caused by the complicated control law in the traditional analytic approach.

166 citations


Journal ArticleDOI
TL;DR: This paper proposes a decentralized behavior-based formation control algorithm for multiple robots considering obstacle avoidance using only the information of the relative position of a robot between neighboring robots and obstacles that can significantly simplify the step of the information process.
Abstract: This paper proposes a decentralized behavior-based formation control algorithm for multiple robots considering obstacle avoidance. Using only the information of the relative position of a robot between neighboring robots and obstacles, the proposed algorithm achieves formation control based on a behavior-based algorithm. In addition, the robust formation is achieved by maintaining the distance and angle of each robot toward the leader robot without using information of the leader robot. To avoid the collisions with obstacles, the heading angles of all robots are determined by introducing the concept of an escape angle, which is related with three boundary layers between an obstacle and the robot. The layer on which the robot is located determines the start time of avoidance and escape angle; this, in turn, generates the escape path along which a robot can move toward the safe layer. In this way, the proposed method can significantly simplify the step of the information process. Finally, simulation results are provided to demonstrate the efficiency of the proposed algorithm.

156 citations


Journal ArticleDOI
TL;DR: The most significant advances in this field are presented, able to solve fundamental technical limitations; such as visual odometry, obstacle detection, mapping and localization, et cetera.
Abstract: This paper presents a complete review of computer vision algorithms and vision-based intelligent applications, that are developed in the field of the Unmanned Aerial Vehicles (UAVs) in the latest decade. During this time, the evolution of relevant technologies for UAVs; such as component miniaturization, the increase of computational capabilities, and the evolution of computer vision techniques have allowed an important advance in the development of UAVs technologies and applications. Particularly, computer vision technologies integrated in UAVs allow to develop cutting-edge technologies to cope with aerial perception difficulties; such as visual navigation algorithms, obstacle detection and avoidance and aerial decision-making. All these expert technologies have developed a wide spectrum of application for UAVs, beyond the classic military and defense purposes. Unmanned Aerial Vehicles and Computer Vision are common topics in expert systems, so thanks to the recent advances in perception technologies, modern intelligent applications are developed to enhance autonomous UAV positioning, or automatic algorithms to avoid aerial collisions, among others. Then, the presented survey is based on artificial perception applications that represent important advances in the latest years in the expert system field related to the Unmanned Aerial Vehicles. In this paper, the most significant advances in this field are presented, able to solve fundamental technical limitations; such as visual odometry, obstacle detection, mapping and localization, et cetera. Besides, they have been analyzed based on their capabilities and potential utility. Moreover, the applications and UAVs are divided and categorized according to different criteria.

155 citations


Journal ArticleDOI
Hongyan Guo1, Chen Shen1, Hui Zhang2, Hong Chen1, Rui Jia1 
TL;DR: The results illustrate that the proposed MPC-based simultaneous trajectory planning and tracking approach achieves acceptable obstacle avoidance performance for an intelligent vehicle.
Abstract: As a typical example of cyber-physical systems, intelligent vehicles are receiving increasing attention, and the obstacle avoidance problem for such vehicles has become a hot topic of discussion. This paper presents a simultaneous trajectory planning and tracking controller for use under cruise conditions based on a model predictive control (MPC) approach to address obstacle avoidance for an intelligent vehicle. The reference trajectory is parameterized as a cubic function in time and is determined by the lateral position and velocity of the intelligent vehicle and the velocity and yaw angle of the obstacle vehicle at the start point of the lane change maneuver. Then, the control sequence for the vehicle is incorporated into the expression for the reference trajectory that is used in the MPC optimization problem by treating the lateral velocity of the intelligent vehicle at the end point of the lane change as an intermediate variable. In this way, trajectory planning and tracking are both captured in a single MPC optimization problem. To evaluate the effectiveness of the proposed simultaneous trajectory planning and tracking approach, joint veDYNA-Simulink simulations were conducted in the unconstrained and constrained cases under leftward and rightward lane change conditions. The results illustrate that the proposed MPC-based simultaneous trajectory planning and tracking approach achieves acceptable obstacle avoidance performance for an intelligent vehicle.

153 citations


Journal ArticleDOI
13 Feb 2018-Sensors
TL;DR: An autonomous obstacle avoidance dynamic path-planning method for a robotic manipulator based on an improved RRT algorithm, called Smoothly RRT (S-RRT), that can increase the sampling speed and efficiency of RRT dramatically and provide theoretical reference value for other type of robots’ path planning.
Abstract: In a future intelligent factory, a robotic manipulator must work efficiently and safely in a Human-Robot collaborative and dynamic unstructured environment. Autonomous path planning is the most important issue which must be resolved first in the process of improving robotic manipulator intelligence. Among the path-planning methods, the Rapidly Exploring Random Tree (RRT) algorithm based on random sampling has been widely applied in dynamic path planning for a high-dimensional robotic manipulator, especially in a complex environment because of its probability completeness, perfect expansion, and fast exploring speed over other planning methods. However, the existing RRT algorithm has a limitation in path planning for a robotic manipulator in a dynamic unstructured environment. Therefore, an autonomous obstacle avoidance dynamic path-planning method for a robotic manipulator based on an improved RRT algorithm, called Smoothly RRT (S-RRT), is proposed. This method that targets a directional node extends and can increase the sampling speed and efficiency of RRT dramatically. A path optimization strategy based on the maximum curvature constraint is presented to generate a smooth and curved continuous executable path for a robotic manipulator. Finally, the correctness, effectiveness, and practicability of the proposed method are demonstrated and validated via a MATLAB static simulation and a Robot Operating System (ROS) dynamic simulation environment as well as a real autonomous obstacle avoidance experiment in a dynamic unstructured environment for a robotic manipulator. The proposed method not only provides great practical engineering significance for a robotic manipulator's obstacle avoidance in an intelligent factory, but also theoretical reference value for other type of robots' path planning.

149 citations


Journal ArticleDOI
TL;DR: This paper addresses formation control with obstacle avoidance problem for a class of second-order stochastic nonlinear multiagent systems under directed topology and proves that control objective can be achieved, of which obstacle avoidance is proven by finding an energy function satisfying that its time derivative is positive.
Abstract: This paper addresses formation control with obstacle avoidance problem for a class of second-order stochastic nonlinear multiagent systems under directed topology. Different with deterministic multiagent systems, stochastic cases are more practical and challenging because the exogenous disturbances depicted by the Wiener process are considered. In order to achieve control objective, both the leader-follower formation approach and the artificial potential field (APF) method are combined together, where the artificial potential is utilized to solve obstacle avoidance problem. For obtaining good system robustness to the undesired side effects of the artificial potential, $H_\infty$ analysis is implemented. Based on the Lyapunov stability theory, it is proven that control objective can be achieved, of which obstacle avoidance is proven by finding an energy function satisfying that its time derivative is positive. Finally, a numerical simulation is carried out to further demonstrate the effectiveness of the proposed formation schemes.

127 citations


Journal ArticleDOI
TL;DR: The proposed controller solves the challenges of navigation, minimizes the computational calculations, and avoids random moving of fireflies, and the performance of proposed controller is better in terms of path optimality when compared to other intelligent navigational approaches.

Journal ArticleDOI
18 Jun 2018
TL;DR: This research presents a probabilistic procedure that can be used to estimate the altitude and speed of UAVs using a 3D camera system, and shows the effectiveness of this method for short-term, low-frequency navigation.
Abstract: Unmanned aerial vehicles (UAVs) have recently attracted the attention of researchers due to their numerous potential civilian applications. However, current robot navigation technologies need furth...

Journal ArticleDOI
TL;DR: An obstacle avoidance algorithm for low speed autonomous vehicles (AV), with guaranteed safety, constructed based on a barrier function method, which works in a plug-and-play fashion with any lower level navigation algorithm.
Abstract: This paper presents an obstacle avoidance algorithm for low speed autonomous vehicles (AV), with guaranteed safety. A supervisory control algorithm is constructed based on a barrier function method, which works in a plug-and-play fashion with any lower level navigation algorithm. When the risk of collision is low, the barrier function is not active; when the risk is high, based on the distance to an “avoidable set,” the barrier function controller will intervene, using a mixed integer program to ensure safety with minimal control effort. This method is applied to solve the navigation and pedestrian avoidance problem of a low speed AV. Its performance is compared with two benchmark algorithms: a potential field method and the Hamilton–Jacobi method.

Journal ArticleDOI
TL;DR: An integrated biologically inspired self-organizing map (SOM) algorithm is proposed for task assignment and path planning of an autonomous underwater vehicle (AUV) system in 3-D underwater environments with obstacle avoidance.
Abstract: An integrated biologically inspired self-organizing map (SOM) algorithm is proposed for task assignment and path planning of an autonomous underwater vehicle (AUV) system in 3-D underwater environments with obstacle avoidance. The algorithm embeds the biologically inspired neural network (BINN) into the SOM neural networks. The task assignment and path planning aim to arrange a team of AUVs to visit all appointed target locations, while assuring obstacle avoidance without speed jump. The SOM neuron network is developed to assign a team of AUVs to achieve multiple target locations in underwater environments. Then, in order to avoid obstacles and speed jump for each AUV that visits the corresponding target location, the BINN is utilized to update weights of the winner of SOM, and achieve AUVs path planning and effective navigation. The effectiveness of the proposed hybrid model is validated by simulation studies.

Journal ArticleDOI
TL;DR: The proposed general feedback control architecture includes an estimator design for fusion of database information, exteroceptive as well as proprioceptive measurements, a geometric corridor planner based on graph theory for the avoidance of multiple, potentially dynamically moving objects, and a spatial-based predictive controller.
Abstract: This paper presents an integrated control approach for autonomous driving comprising a corridor path planner that determines constraints on vehicle position, and a linear time-varying model predictive controller combining path planning and tracking in a road-aligned coordinate frame. The capabilities of the approach are illustrated in obstacle-free curved road-profile tracking, in an application coupling adaptive cruise control (ACC) with obstacle avoidance (OA), and in a typical driving maneuver on highways. The vehicle is modeled as a nonlinear dynamic bicycle model with throttle, brake pedal position, and steering angle as control inputs. Proximity measurements are assumed to be available within a given range field surrounding the vehicle. The proposed general feedback control architecture includes an estimator design for fusion of database information (maps), exteroceptive as well as proprioceptive measurements, a geometric corridor planner based on graph theory for the avoidance of multiple, potentially dynamically moving objects, and a spatial-based predictive controller. Switching rules for transitioning between four different driving modes, i.e., ACC, OA, obstacle-free road tracking (RT), and controlled braking (Brake), are discussed. The proposed method is evaluated on test cases, including curved and highway two-lane road tracks with static as well as moving obstacles.

Journal ArticleDOI
TL;DR: The task motion and self-motion (CTS) methods are coordinated to enhance the intelligence of the shared control system by equipping the robot with an autonomous obstacle avoidance function.
Abstract: This paper reports the development of an intelligent shared control system for a robotic manipulator that is commanded by the user's mind. The target objects are detected by a vision system and then displayed to the user in a video that shows them fused with flicking diamonds that are designed to excite electroencephalograph (EEG) signals at different frequency bands. Through the analysis of the invoked EEG signals, a brain–computer interface is developed to infer the exact object that is required by the user. These results are then transferred to the shared control system, which is enabled by visual servoing techniques to achieve accurate object manipulation. The task motion and self-motion (CTS) methods are coordinated to enhance the intelligence of the shared control system by equipping the robot with an autonomous obstacle avoidance function. Extensive experimental studies are performed to verify that the adaptive object tracking algorithm, the CTS method, and the least-squares method are helpful in improving the performance of the intelligent robotic system.

Journal ArticleDOI
TL;DR: Double Q-network (DDQN) deep reinforcement learning proposed by DeepMind in 2016 is applied to dynamic path planning of unknown environment and the agent is able to reach the local target position successfully in unknown dynamic environment.
Abstract: Dynamic path planning of unknown environment has always been a challenge for mobile robots. In this paper, we apply double Q-network (DDQN) deep reinforcement learning proposed by DeepMind in 2016 to dynamic path planning of unknown environment. The reward and punishment function and the training method are designed for the instability of the training stage and the sparsity of the environment state space. In different training stages, we dynamically adjust the starting position and target position. With the updating of neural network and the increase of greedy rule probability, the local space searched by agent is expanded. Pygame module in PYTHON is used to establish dynamic environments. Considering lidar signal and local target position as the inputs, convolutional neural networks (CNNs) are used to generalize the environmental state. Q-learning algorithm enhances the ability of the dynamic obstacle avoidance and local planning of the agents in environment. The results show that, after training in different dynamic environments and testing in a new environment, the agent is able to reach the local target position successfully in unknown dynamic environment.

Proceedings ArticleDOI
26 Jun 2018
TL;DR: Deep Reinforcement Learning autonomous navigation and obstacle avoidance of self-driving cars, applied with Deep Q Network to a simulated car an urban environment using two types of sensor data as input: camera sensor and laser sensor in front of the car.
Abstract: Deep Reinforcement Learning has led us to newer possibilities in solving complex control and navigation related tasks. The paper presents Deep Reinforcement Learning autonomous navigation and obstacle avoidance of self-driving cars, applied with Deep Q Network to a simulated car an urban environment. The approach uses two types of sensor data as input: camera sensor and laser sensor in front of the car. It also designs a cost-efficient high-speed car prototype capable of running the same algorithm in real-time. The design uses a camera and a Hokuyo Lidar sensor in the car front. It uses embedded GPU (Nvidia-TX2) for running deep-learning algorithms based on sensor inputs.

Journal ArticleDOI
TL;DR: A multiobjective path planning (MOPP) framework to explore a suitable path for a UAV operating in a dynamic urban environment, where safety level is considered in the proposed framework to guarantee the safety of UAV in addition to travel time is proposed.
Abstract: This paper is concerned with path planning for unmanned aerial vehicles (UAVs) flying through low altitude urban environment. Although many different path planning algorithms have been proposed to find optimal or near-optimal collision-free paths for UAVs, most of them either do not consider dynamic obstacle avoidance or do not incorporate multiple objectives. In this paper, we propose a multiobjective path planning (MOPP) framework to explore a suitable path for a UAV operating in a dynamic urban environment, where safety level is considered in the proposed framework to guarantee the safety of UAV in addition to travel time. To this aim, two types of safety index maps (SIMs) are developed first to capture static obstacles in the geography map and unexpected obstacles that are unavailable in the geography map. Then an MOPP method is proposed by jointly using offline and online search, where the offline search is based on the static SIM and helps shorten the travel time and avoid static obstacles, while the online search is based on the dynamic SIM of unexpected obstacles and helps bypass unexpected obstacles quickly. Extensive experimental results verify the effectiveness of the proposed framework under the dynamic urban environment.

Journal ArticleDOI
TL;DR: The main contribution of this paper is to prove in a constructive way that a gradient-descent coordination control law designed for single integrators can be easily modified to adapt for various motion constraints such as nonholonomic dynamics, linear/angular velocity saturation, and other path constraints while preserving the convergence of the entire multiagent system.
Abstract: This paper proposes a general approach to design convergent coordination control laws for multiagent systems subject to motion constraints. The main contribution of this paper is to prove in a constructive way that a gradient-descent coordination control law designed for single integrators can be easily modified to adapt for various motion constraints such as nonholonomic dynamics, linear/angular velocity saturation, and other path constraints while preserving the convergence of the entire multiagent system. The proposed approach is applicable to a wide range of coordination tasks such as rendezvous and formation control in two and three dimensions. As a special application, the proposed approach solves the problem of distance-based formation control subject to nonholonomic and velocity saturation constraints.

Journal ArticleDOI
24 Apr 2018
TL;DR: The use of the “Neural Network Toolbox” tool already present in MATLAB is tested to design an artificial neural network with supervised learning suitable for classification and pattern recognition by using data collected by an ultrasonic sensor.
Abstract: Artificial intelligence is the ability of a computer to perform the functions and reasoning typical of the human mind. In its purely informatic aspect, it includes the theory and techniques for the development of algorithms that allow machines to show an intelligent ability and/or perform an intelligent activity, at least in specific areas. In particular, there are automatic learning algorithms based on the same mechanisms that are thought to be the basis of all the cognitive processes developed by the human brain. Such a powerful tool has already started to produce a new class of self-driving vehicles. With the projections of population growth that will increase until the year 2100 up to 11.2 billion, research on innovating agricultural techniques must be continued. In order to improve the efficiency regarding precision agriculture, the use of autonomous agricultural machines must become an important issue. For this reason, it was decided to test the use of the “Neural Network Toolbox” tool already present in MATLAB to design an artificial neural network with supervised learning suitable for classification and pattern recognition by using data collected by an ultrasonic sensor. The idea is to use such a protocol to retrofit kits for agricultural machines already present on the market.

Proceedings ArticleDOI
21 May 2018
TL;DR: In this paper, a simple controller is used as an alternative, switchable policy to speed up training of DRL for local planning and navigation problems, which can then take over to perform more complex actions.
Abstract: Deep Reinforcement Learning (DRL) has been applied successfully to many robotic applications. However, the large number of trials needed for training is a key issue. Most of existing techniques developed to improve training efficiency (e.g. imitation) target on general tasks rather than being tailored for robot applications, which have their specific context to benefit from. We propose a novel framework, Assisted Reinforcement Learning, where a classical controller (e.g. a PID controller) is used as an alternative, switchable policy to speed up training of DRL for local planning and navigation problems. The core idea is that the simple control law allows the robot to rapidly learn sensible primitives, like driving in a straight line, instead of random exploration. As the actor network becomes more advanced, it can then take over to perform more complex actions, like obstacle avoidance. Eventually, the simple controller can be discarded entirely. We show that not only does this technique train faster, it also is less sensitive to the structure of the DRL network and consistently outperforms a standard Deep Deterministic Policy Gradient network. We demonstrate the results in both simulation and real-world experiments.

Journal ArticleDOI
TL;DR: An autonomous path planning method for the robotic manipulator based on an improved RRT algorithm that can significantly improve the success rate and efficiency of the planning without losing other performance.
Abstract: With the development of modern manufacturing industry, the application scenarios of industrial robot are becoming more and more complex. Manual programming of industrial robot requires a great deal of effort and time. Therefore, an autonomous path planning is an important development direction of industrial robot. Among the path planning methods, the rapidly-exploring random tree (RRT) algorithm based on random sampling has been widely applied for a high-dimensional robotic manipulator because of its probability completeness and outstanding expansion. However, especially in the complex scenario, the existing RRT planning algorithms still have a low planning efficiency and some are easily fall into a local minimum. To tackle these problems, this paper proposes an autonomous path planning method for the robotic manipulator based on an improved RRT algorithm. The method introduces regression mechanism to prevent over-searching configuration space. In addition, it adopts an adaptive expansion mechanism to continuously improve reachable spatial information by refining the boundary nodes in joint space, avoiding repeatedly searching for extended nodes. Furthermore, it avoids the unnecessary iteration of the robotic manipulator forward kinematics solution and its time-consuming collision detection in Cartesian space. The method can rapidly plan a path to a target point and can be accelerated out of a local minimum area to improve path planning efficiency. The improved RRT algorithm proposed in this paper is simulated in a complex environment. The results reveal that the proposed algorithm can significantly improve the success rate and efficiency of the planning without losing other performance.

Proceedings ArticleDOI
12 Jun 2018
TL;DR: A novel modeling framework for obstacle avoidance which allows us to easily account for generic, possibly nonconvex, obstacles involving polytopes, ellipsoids, semialgebraic sets and generic sets described by a set of nonlinear inequalities.
Abstract: We employ the proximal averaged Newton-type method for optimal control (PANOC) to solve obstacle avoidance problems in real time. We introduce a novel modeling framework for obstacle avoidance which allows us to easily account for generic, possibly nonconvex, obstacles involving polytopes, ellipsoids, semialgebraic sets and generic sets described by a set of nonlinear inequalities. PANOC is particularly well-suited for embedded applications as it involves simple steps, its implementation comes with a low memory footprint and its fast convergence meets the tight runtime requirements of fast dynamical systems one encounters in modern mechatronics and robotics. The proposed obstacle avoidance scheme is tested on a lab-scale autonomous vehicle.

Journal ArticleDOI
TL;DR: In this work, collision detection and path planning methods for USVs are presented and it is concluded that almost all the existing method do not address sea or weather conditions, or do not involve the dynamics of the vessel while defining the path.
Abstract: The adoption of a robust collision avoidance module is required to realise fully autonomous Unmanned Surface Vehicles (USVs). In this work, collision detection and path planning methods for USVs are presented. Attention is focused on the difference between local and global path planners, describing the most common techniques derived from classical graph search theory. In addition, a dedicated section is reserved for intelligent methods, such as artificial neural networks and evolutionary algorithms. Born as optimisation methods, they can learn a close-to-optimal solution without requiring large computation effort under certain constraints. Finally, the deficiencies of the existing methods are highlighted and discussed. It has been concluded that almost all the existing method do not address sea or weather conditions, or do not involve the dynamics of the vessel while defining the path. Therefore, this research area is still far from being considered fully explored.

Journal ArticleDOI
TL;DR: This is the fastest lightweight aerial vehicle to perform collision avoidance using three‐dimensional geometric information and a complete working system detecting obstacles at 120 Hz and avoiding trees at up to 14 m/s (31 MPH).
Abstract: We present the design and implementation of a small autonomous unmanned aerial vehicle capable of high-speed flight through complex natural environments. Using only onboard GPS-denied sensing and computation, we perform obstacle detection, planning, and feedback control in real time. We present a novel integrated approach to perception and control using pushbroom stereo, which exploits forward motion to enable efficient obstacle detection and avoidance using lightweight processors on an unmanned aerial vehicle. Our use of model-based planning and control techniques allows us to track precise trajectories that avoid obstacles identified by the vision system. We demonstrate a complete working system detecting obstacles at 120 Hz and avoiding trees at up to 14 m/s (31 MPH). To the best of our knowledge, this is the fastest lightweight aerial vehicle to perform collision avoidance using three-dimensional geometric information.

Journal ArticleDOI
25 Jun 2018
TL;DR: The proposed end-to-end modular reinforcement learning architecture for a navigation task in complex dynamic environments with rapidly moving obstacles can efficiently avoid moving obstacles and complete the navigation task at a high success rate.
Abstract: In this paper, we propose an end-to-end modular reinforcement learning architecture for a navigation task in complex dynamic environments with rapidly moving obstacles. In this architecture, the main task is divided into two subtasks: local obstacle avoidance and global navigation. For obstacle avoidance, we develop a two-stream Q-network, which processes spatial and temporal information separately and generates action values. The global navigation subtask is resolved by a conventional Q-network framework. An online learning network and an action scheduler are introduced to first combine two pretrained policies, and then continue exploring and optimizing until a stable policy is obtained. The two-stream Q-network obtains better performance than the conventional deep Q-learning approach in the obstacle avoidance subtask. Experiments on the main task demonstrate that the proposed architecture can efficiently avoid moving obstacles and complete the navigation task at a high success rate. The modular architecture enables parallel training and also demonstrates good generalization capability in different environments.

Journal ArticleDOI
TL;DR: The proposed method can perform safe and timely dynamic avoidance for redundant manipulators in human-robot interaction and is implemented in Robot Operating System (ROS) using C++.
Abstract: In order to avoid dynamic obstacle timely during manufacturing tasks performed by manipulators, a novel method based on distance calculation and discrete detection is proposed. The nearest distances between the links of a manipulator and the convex hull of an arbitrarily-shaped dynamic obstacle obtained from Kinect-V2 camera in real-time are calculated by Gilbert–Johnson–Keerthi algorithm, and the minimum one is defined as the closest distance between the manipulator and the obstacle. When the closest distance is less than a safe value, whether the dynamic obstacle is located in the global path of the manipulator is determined by improved discrete collision detection, which can adjust detection step-size adaptively for accuracy and efficiency. If the obstacle will collide with the manipulator, set a local goal and re-plan a local path for the manipulator. The proposed method is implemented in Robot Operating System (ROS) using C++. The experiments indicate that the proposed method can perform safe and timely dynamic avoidance for redundant manipulators in human-robot interaction.

Proceedings ArticleDOI
01 Oct 2018
TL;DR: This work introduces a self-supervised CNN-based approach for indoor robot navigation that addresses the problem of real-time obstacle avoidance, by employing a regression CNN that predicts the agent's distance-to-collision in view of the raw visual input of its on-board monocular camera.
Abstract: Nowadays, Unmanned Aerial Vehicles (UAVs)are becoming increasingly popular facilitated by their extensive availability. Autonomous navigation methods can act as an enabler for the safe deployment of drones on a wide range of real-world civilian applications. In this work, we introduce a self-supervised CNN-based approach for indoor robot navigation. Our method addresses the problem of real-time obstacle avoidance, by employing a regression CNN that predicts the agent's distance-to-collision in view of the raw visual input of its on-board monocular camera. The proposed CNN is trained on our custom indoor-flight dataset which is collected and annotated with real-distance labels, in a self-supervised manner using external sensors mounted on an UAV. By simultaneously processing the current and previous input frame, the proposed CNN extracts spatio-temporal features that encapsulate both static appearance and motion information to estimate the robot's distance to its closest obstacle towards multiple directions. These predictions are used to modulate the yaw and linear velocity of the UAV, in order to navigate autonomously and avoid collisions. Experimental evaluation demonstrates that the proposed approach learns a navigation policy that achieves high accuracy on real-world indoor flights, outperforming previously proposed methods from the literature.

Journal ArticleDOI
TL;DR: In this paper, a nonlinear model predictive control (MPC) formulation for obstacle avoidance in high-speed, large-size autono-mous ground vehicles (AGVs) with high center of gravity (CoG) that operate in unstructured environments, such as military vehicles, is presented.
Abstract: This paper presents a nonlinear model predictive control (MPC) formulation for obstacle avoidance in high-speed, large-size autono-mous ground vehicles (AGVs) with high centre of gravity (CoG) that operate in unstructured environments, such as military vehicles. The term ‘unstructured’ in this context denotes that there are no lanes or traffic rules to follow. Existing MPC formulations for passenger vehicles in structured environments do not readily apply to this context. Thus, a new nonlinear MPC formulation is developed to navigate an AGV from its initial position to a target position at high-speed safely. First, a new cost function formulation is used that aims to find the shortest path to the target position, since no reference trajectory exists in unstructured environments. Second, a region partitioning approach is used in conjunction with a multi-phase optimal control formulation to accommodate the complicated forms the obstacle-free region can assume due to the presence of multiple obstacle...