Algorithms for collision-free navigation of mobile robots in complex cluttered environments: a survey

HAVEN: Haptic And Visual Environment Navigation by a Shape-Changing Mobile Robot with Multimodal Perception

Many animals exhibit agile mobility in obstructed environments due to their ability to tune their bodies to negotiate and manipulate obstacles and apertures. Most mobile robots are rigid structures and avoid obstacles where possible. In this work, we introduce a new framework named Haptic And Visual Environment Navigation (HAVEN) Architecture to combine vision and proprioception for a deformable mobile robot to be more agile in obstructed environments. The algorithms enable the robot to be autonomously (a) predictive by analysing visual feedback from the environment and preparing its body accordingly, (b) reactive by responding to proprioceptive feedback, and (c) active by manipulating obstacles and gap sizes using its deformable body. The robot was tested approaching differently sized apertures in obstructed environments ranging from greater than its shape to smaller than its narrowest possible size. The experiments involved multiple obstacles with different physical properties. The results show higher navigation success rates and an average 32% navigation time reduction when the robot actively manipulates obstacles using its shape-changing body.

An improved RRT behavioral planning method for robots based on PTM algorithm

For multi-dimensional high-order nonlinear systems with unstable path quality in parameter and extension terms, we developed a new fast search random tree strategy. First, we established a high-order Lipschitz vector field dynamic system to adapt to high-order systems of multi-degree-of-freedom robots, with the complex obstacle function being one of its key components. Secondly, we designed a classification gap filtering network layer (Classification LSTM) to screen training data models and ensure the global stability of data in path design. Additionally, the visual sensors deployed in the unit area effectively implement the path marking backtracking strategy and dead zone path simplification. Finally, three examples are provided to verify the effectiveness of this design method.

A Retrospective Analysis of Indoor CO2 Measurements Obtained with a Mobile Robot during the COVID-19 Pandemic

This work presents a retrospective analysis of indoor CO2 measurements obtained with a mobile robot in an educational building after the COVID-19 lockdown (May 2021), at a time when public activities resumed with mandatory local pandemic restrictions. The robot-based CO2 measurement system was assessed as an alternative to the deployment of a net of sensors in a building in the pandemic period, in which there was a global stock outage of CO2 sensors. The analysis of the obtained measurements confirms that a mobile system can be used to obtain interpretable information on the CO2 levels inside the rooms of a building during a pandemic outbreak.

The Impact of LiDAR Configuration on Goal-Based Navigation within a Deep Reinforcement Learning Framework

Over the years, deep reinforcement learning (DRL) has shown great potential in mapless autonomous robot navigation and path planning. These DRL methods rely on robots equipped with different light detection and range (LiDAR) sensors with a wide field of view (FOV) configuration to perceive their environment. These types of LiDAR sensors are expensive and are not suitable for small-scale applications. In this paper, we address the performance effect of the LiDAR sensor configuration in DRL models. Our focus is on avoiding static obstacles ahead. We propose a novel approach that determines an initial FOV by calculating an angle of view using the sensor's width and the minimum safe distance required between the robot and the obstacle. The beams returned within the FOV, the robot's velocities, the robot's orientation to the goal point, and the distance to the goal point are used as the input state to generate new velocity values as the output action of the DRL. The cost function of collision avoidance and path planning is defined as the reward of the DRL model. To verify the performance of the proposed method, we adjusted the proposed FOV by ±10° giving a narrower and wider FOV. These new FOVs are trained to obtain collision avoidance and path planning DRL models to validate the proposed method. Our experimental setup shows that the LiDAR configuration with the computed angle of view as its FOV performs best with a success rate of 98% and a lower time complexity of 0.25 m/s. Additionally, using a Husky Robot, we demonstrate the model's good performance and applicability in the real world.

Motion planning around obstacles with convex optimization

From quadrotors delivering packages in urban areas to robot arms moving in confined warehouses, motion planning around obstacles is a core challenge in modern robotics. Planners based on optimization can design trajectories in high-dimensional spaces while satisfying the robot dynamics. However, in the presence of obstacles, these optimization problems become nonconvex and very hard to solve, even just locally. Thus, when facing cluttered environments, roboticists typically fall back to sampling-based planners that do not scale equally well to high dimensions and struggle with continuous differential constraints. Here, we present a framework that enables convex optimization to efficiently and reliably plan trajectories around obstacles. Specifically, we focus on collision-free motion planning with costs and constraints on the shape, the duration, and the velocity of the trajectory. Using recent techniques for finding shortest paths in Graphs of Convex Sets (GCS), we design a practical convex relaxation of the planning problem. We show that this relaxation is typically very tight, to the point that a cheap postprocessing of its solution is almost always sufficient to identify a collision-free trajectory that is globally optimal (within the parameterized class of curves). Through numerical and hardware experiments, we demonstrate that our planner, which we name GCS, can find better trajectories in less time than widely used sampling-based algorithms and can reliably design trajectories in high-dimensional complex environments.

Autonomous driving using imitation learning with look ahead point for semi structured environments

Semi-structured environments are difficult for autonomous driving because there are numerous unknown obstacles in drivable area without lanes, and its width and curvature considerably change. In such environments, searching for a path on a real-time is difficult, and localization data are inaccurate, reducing path tracking accuracy. Instead, alternative methods that reactively avoid obstacles in real-time using candidate paths or an artificial potential field have been studied. However, these require heuristics to identify specific parameters for handling various environments and are vulnerable to inaccurate input data. To address these limitations, this study proposes a method in which a vehicle drives toward drivable area using vision and deep learning. The proposed imitation learning method learns the look-ahead point where the vehicle should reach on a vision-based occupancy grid map to obtain a safe policy with a clear state action pattern relationship. Furthermore, using this point, the data aggregation (DAgger) algorithm with weighted loss function is proposed, which imitates expert behavior more accurately, especially in unsafe or near-collision situations. Experimental results in actual semi-structured environments demonstrated the limitations of general model-based methods and the effectiveness of the proposed imitation learning method. Moreover, simulation experiments showed that DAgger with the weight obtains a safer policy than existing DAgger algorithms.

An Optimized Probabilistic Roadmap Algorithm for Path Planning of Mobile Robots in Complex Environments with Narrow Channels

In this paper, we propose a new path planning algorithm based on the probabilistic roadmaps method (PRM), in order to effectively solve the autonomous path planning of mobile robots in complex environments with multiple narrow channels. The improved PRM algorithm mainly improves the density and distribution of sampling points in the narrow channel, through a combination of the learning process of the PRM algorithm and the APF algorithm. We also shortened the required time and path length by optimizing the query process. The first key technology to improve the PRM algorithm involves optimizing the number and distribution of free points and collision-free lines in the free workspace. To ensure full visibility of the narrow channel, we extend the obstacles through the diagonal distance of the mobile robot while ignoring the safety distance. Considering the safety distance during movement, we re-classify the all sampling points obtained by the quasi-random sampling principle into three categories: free points, obstacle points, and adjacent points. Next, we transform obstacle points into the free points of the narrow channel by combining the APF algorithm and the characteristics of the narrow channel, increasing the density of sampling points in the narrow space. Then, we include potential energy judgment into the construction process of collision-free lines shortening the required time and reduce collisions with obstacles. Optimizing the query process of the PRM algorithm is the second key technology. To reduce the required time in the query process, we adapt the bidirectional A* algorithm to query these local paths and obtain an effective path to the target point. We also combine the path pruning technology with the potential energy function to obtain a short path without collisions. Finally, the experimental results demonstrate that the new PRM path planning technology can improve the density of free points in narrow spaces and achieve an optimized, collision-free path in complex environments with multiple narrow channels.

Toward Safer Navigation of Heterogeneous Mobile Robots in Distributed Scheme: A Novel Time-to-Collision-Based Method

For safe and efficient navigation of heterogeneous multiple mobile robots (HMRs), it is essential to incorporate dynamics (mass and inertia) in motion control algorithms. Many methods rely only on kinematics or point-mass models, resulting in conservative results or occasionally failure. This is especially true for robots with different masses. In this article, we develop a novel navigation methodology for a distributed scheme by incorporating the robots' dynamics through calculating the time to collision (TTC) and designing a new controller accordingly that avoids collisions. We first propose a new predictive collision term by TTC that will be used to quantify imminent collisions among HMRs. Subsequently, using this term, we develop a novel nonlinear controller that explicitly incorporates TTC in the design and guarantees collision-free motion. Simulations and experiments were performed to demonstrate the effectiveness of the developed methods. We first compared the results of our proposed approach with controllers that only consider the robots' kinematics. It was shown that the proposed control strategy (a TTC-based controller) proves to be less conservative when determining safe motions. Specifically, for environments with limited space, it was demonstrated that using robots' kinematics may result in a collision, while our strategy results in safe motion. We also performed experiments that proved collision-free navigation of HMRs with this approach. The outcomes of this work provide more reliable motion control for HMRs, especially when the robots' masses or inertias are significantly different, for example, warehouses. The developments in this work are also applicable to vehicles and can therefore be beneficial in automated collision avoidance in autonomous driving and intelligent transportation.

On the Throughput of the Common Target Area for Robotic Swarm Strategies

A robotic swarm may encounter traffic congestion when many robots simultaneously attempt to reach the same area. This work proposes two measures for evaluating the access efficiency of a common target area as the number of robots in the swarm rises: the maximum target area throughput and its maximum asymptotic throughput. Both are always finite as the number of robots grows, in contrast to the arrival time at the target per number of robots that tends to infinity. Using them, one can analytically compare the effectiveness of different algorithms. In particular, three different theoretical strategies proposed and formally evaluated for reaching a circular target area: (i) forming parallel queues towards the target area, (ii) forming a hexagonal packing through a corridor going to the target, and (iii) making multiple curved trajectories towards the boundary of the target area. The maximum throughput and the maximum asymptotic throughput (or bounds for it) for these strategies are calculated, and these results are corroborated by simulations. The key contribution is not the proposal of new algorithms to alleviate congestion but a fundamental theoretical study of the congestion problem in swarm robotics when the target area is shared.

A hybrid inductive learning-based and deductive reasoning-based 3-D path planning method in complex environments

Traditional path planning methods, such as sampling-based and iterative approaches, allow for optimal path’s computation in complex environments. Nonetheless, environment exploration is subject to rules which can be obtained by domain experts and could be used for improving the search. The present work aims at integrating inductive techniques that generate path candidates with deductive techniques that choose the preferred ones. In particular, an inductive learning model is trained with expert demonstrations and with rules translated into a reward function, while logic programming is used to choose the starting point according to some domain expert’s suggestions. We discuss, as use case, 3-D path planning for neurosurgical steerable needles. Results show that the proposed method computes optimal paths in terms of obstacle clearance and kinematic constraints compliance, and is able to outperform state-of-the-art approaches in terms of safety distance-from-obstacles respect, smoothness, and computational time.

Motion planning of unmanned aerial vehicles in dynamic 3D space: a potential force approach

This research focuses on a collision-free real-time motion planning system for unmanned aerial vehicles (UAVs) in complex three-dimensional (3D) dynamic environments based on generalized potential force functions. The UAV must survive in such a complex heterogeneous environment while tracking a dynamic target and avoiding multiple stationary or dynamic obstacles, especially at low hover flying conditions. The system framework consists of two parts. The first part is the target tracking part employing a generalized extended attractive potential force into 3D space. In contrast, the second part is the obstacle avoidance part employing a generalized extended repulsive potential force into 3D space. These forces depend on the relative position and relative velocity between the UAV and respective obstacles. As a result, the UAV is attracted to a moving or stationary target and repulsed away from moving or static obstacles simultaneously in 3D space. Accordingly, it changes its altitude and projected planner position concurrently. A real-time implementation for the system is conducted in the SPACE laboratory to perform motion planning in 3D space. The system performance is validated in real-time experiments using three platforms: two parrot bebop drones and one turtlebot robot. The pose information of the vehicles is estimated using six Vicon cameras during real-time flights. The demonstrated results show the motion planning system’s effectiveness. Also, we propose a successful mathematical solution of the local minima problem associated with the potential field method in both stationary and dynamic environments.

Optimal Navigation of an Unmanned Surface Vehicle and an Autonomous Underwater Vehicle Collaborating for Reliable Acoustic Communication with Collision Avoidance

This paper focuses on safe navigation of an unmanned surface vehicle in proximity to a submerged autonomous underwater vehicle so as to maximise short-range, through-water data transmission while minimising the probability that the two vehicles will accidentally collide. A sliding mode navigation law is developed, and a rigorous proof of optimality of the proposed navigation law is presented. The developed navigation algorithm is relatively computationally simple and easily implementable in real time. Illustrative examples with extensive computer simulations demonstrate the effectiveness of the proposed method.

A 3D Vision Cone Based Method for Collision Free Navigation of a Quadcopter UAV among Moving Obstacles

In the near future, it’s expected that unmanned aerial vehicles (UAVs) will become ubiquitous surrogates for human-crewed vehicles in the field of border patrol, package delivery, etc. Therefore, many three-dimensional (3D) navigation algorithms based on different techniques, e.g., model predictive control (MPC)-based, navigation potential field-based, sliding mode control-based, and reinforcement learning-based, have been extensively studied in recent years to help achieve collision-free navigation. The vast majority of the 3D navigation algorithms perform well when obstacles are sparsely spaced, but fail when facing crowd-spaced obstacles, which causes a potential threat to UAV operations. In this paper, a 3D vision cone-based reactive navigation algorithm is proposed to enable small quadcopter UAVs to seek a path through crowd-spaced 3D obstacles to the destination without collisions. The proposed algorithm is simulated in MATLAB with different 3D obstacles settings to demonstrate its feasibility and compared with the other two existing 3D navigation algorithms to exhibit its superiority. Furthermore, a modified version of the proposed algorithm is also introduced and compared with the initially proposed algorithm to lay the foundation for future work.

Intelligent cooperative collision avoidance via fuzzy potential fields

This paper proposes an intelligent cooperative collision avoidance approach combining the enhanced potential field (EPF) with a fuzzy inference system (FIS) to resolve local minima and goal non-reachable with obstacles nearby issues and provide a near-optimal collision-free trajectory. A genetic algorithm is utilized to optimize parameters of membership function and rule base of the FISs. This work uses a single scenario containing all issues and interactions among unmanned aerial vehicles (UAVs) for training. For validating the performance, two scenarios containing obstacles with different shapes and several UAVs in small airspace are considered. Multiple simulation results show that the proposed approach outperforms the conventional EPF approach statistically.

Towards Fully Autonomous UAVs: A Survey

Unmanned Aerial Vehicles have undergone rapid developments in recent decades. This has made them very popular for various military and civilian applications allowing us to reach places that were previously hard to reach in addition to saving time and lives. A highly desirable direction when developing unmanned aerial vehicles is towards achieving fully autonomous missions and performing their dedicated tasks with minimum human interaction. Thus, this paper provides a survey of some of the recent developments in the field of unmanned aerial vehicles related to safe autonomous navigation, which is a very critical component in the whole system. A great part of this paper focus on advanced methods capable of producing three-dimensional avoidance maneuvers and safe trajectories. Research challenges related to unmanned aerial vehicle development are also highlighted.

Occlusion-Aware UAV Path Planning for Reconnaissance and Surveillance

Unmanned Aerial Vehicles (UAVs) have become necessary tools for a wide range of activities including but not limited to real-time monitoring, surveillance, reconnaissance, border patrol, search and rescue, civilian, scientific and military missions, etc. Their advantage is unprecedented and irreplaceable, especially in environments dangerous to humans, for example, in radiation or pollution-exposed areas. Two path-planning algorithms for reconnaissance and surveillance are proposed in this paper, which ensures every point on the target ground area can be seen at least once in a complete surveillance circle. Moreover, the geometrically complex environments with occlusions are considered in our research. Compared with many existing methods, we decompose this problem into a waypoint-determination problem and an instance of the traveling-salesman problem.

A Hybrid Approach for Autonomous Collision-Free UAV Navigation in 3D Partially Unknown Dynamic Environments

In the past decades, unmanned aerial vehicles (UAVs) have emerged in a wide range of applications. Owing to the advances in UAV technologies related to sensing, computing, power, etc., it has become possible to carry out missions autonomously. A key component to achieving this goal is the development of safe navigation methods, which is the main focus of this work. A hybrid navigation approach is proposed to allow safe autonomous operations in three-dimensional (3D) partially unknown and dynamic environments. This method combines a global path planning algorithm, namely RRT-Connect, with a reactive control law based on sliding mode control to provide quick reflex-like reactions to newly detected obstacles. The performance of the suggested approach is validated using simulations.

A method for autonomous collision-free navigation of a quadrotor UAV in unknown tunnel-like environments

Unmanned aerial vehicles (UAVs) have become essential tools for exploring, mapping and inspection of unknown three-dimensional (3D) tunnel-like environments which is a very challenging problem. A computationally light navigation algorithm is developed in this paper for quadrotor UAVs to autonomously guide the vehicle through such environments. It uses sensors observations to safely guide the UAV along the tunnel axis while avoiding collisions with its walls. The approach is evaluated using several computer simulations with realistic sensing models and practical implementation with a quadrotor UAV. The proposed method is also applicable to other UAV types and autonomous underwater vehicles.

Motion Control of a Flock of 1-Trailer Robots with Swarm Avoidance

This paper addresses the motion planning and control problem of a system of 1-trailer robots navigating a dynamic environment cluttered with obstacles including a swarm of boids. A set of nonlinear continuous control laws is proposed via the Lyapunov-based Control Scheme for collision, obstacle, and swarm avoidances. Additionally, a leader–follower strategy is utilized to allow the flock to split and rejoin when approaching obstacles. The effectiveness of the control laws is demonstrated through numerical simulations, which show the split and rejoin maneuvers by the flock when avoiding obstacles while the swarm exhibits emergent behaviors.

Non-Communication Decentralized Multi-Robot Collision Avoidance in Grid Map Workspace with Double Deep Q-Network

This paper presents a novel decentralized multi-robot collision avoidance method with deep reinforcement learning, which is not only suitable for the large-scale grid map workspace multi-robot system, but also directly processes Lidar signals instead of communicating between the robots. According to the particularity of the workspace, we handcrafted a reward function, which considers both the collision avoidance among the robots and as little as possible change of direction of the robots during driving. Using Double Deep Q-Network (DDQN), the policy was trained in the simulation grid map workspace. By designing experiments, we demonstrated that the learned policy can guide the robot well to effectively travel from the initial position to the goal position in the grid map workspace and to avoid collisions with others while driving.

Internal Wind Turbine Blade Inspections Using UAVs: Analysis and Design Issues

Interior and exterior wind turbine blade inspections are necessary to extend the lifetime of wind turbine generators. The use of unmanned vehicles is an alternative to exterior wind turbine blade inspections performed by technicians that require the use of cranes and ropes. Interior wind turbine blade inspections are even more challenging due to the confined spaces, lack of illumination, and the presence of potentially harmful internal structural components. Additionally, the cost of manned interior wind turbine blade inspections is a major limiting factor. This paper analyses all aspects of the viability of using manually controlled or autonomous aerial vehicles for interior wind turbine blade inspections. We discuss why the size, weight, and flight time of a vehicle, in addition to the structure of the wind turbine blade, are the main limiting factors in performing internal blade inspections. We also describe the design issues that must be considered to provide autonomy to unmanned vehicles and the control system, the sensors that can be used, and introduce some of the algorithms for localization, obstacle avoidance and path planning that are best suited for the task. Lastly, we briefly describe which non-destructive test instrumentation can be used for the purpose.

Autonomous Navigation of a Team of Unmanned Surface Vehicles for Intercepting Intruders on a Region Boundary

We study problems of intercepting single and multiple invasive intruders on a boundary of a planar region by employing a team of autonomous unmanned surface vehicles. First, the problem of intercepting a single intruder has been studied and then the proposed strategy has been applied to intercepting multiple intruders on the region boundary. Based on the proposed decentralised motion control algorithm and decision making strategy, each autonomous vehicle intercepts any intruder, which tends to leave the region by detecting the most vulnerable point of the boundary. An efficient and simple mathematical rules based control algorithm for navigating the autonomous vehicles on the boundary of the see region is developed. The proposed algorithm is computationally simple and easily implementable in real life intruder interception applications. In this paper, we obtain necessary and sufficient conditions for the existence of a real-time solution to the considered problem of intruder interception. The effectiveness of the proposed method is confirmed by computer simulations with both single and multiple intruders.

Research on autonomous collision avoidance of merchant ship based on inverse reinforcement learning

To learn the optimal collision avoidance policy of merchant ships controlled by human experts, a finite-state Markov decision process model for ship collision avoidance is proposed based on the analysis of collision avoidance mechanism, and an inverse reinforcement learning (IRL) method based on cross entropy and projection is proposed to obtain the optimal policy from expert’s demonstrations. Collision avoidance simulations in different ship encounters are conducted and the results show that the policy obtained by the proposed IRL has a good inversion effect on two kinds of human experts, which indicate that the proposed method can effectively learn the policy of human experts for ship collision avoidance.

Energy-Efficient Autonomous Navigation of Solar-Powered UAVs for Surveillance of Mobile Ground Targets in Urban Environments

In this paper, we consider the navigation of a group of solar-powered unmanned aerial vehicles (UAVs) for periodical monitoring of a set of mobile ground targets in urban environments. We consider the scenario where the number of targets is larger than that of the UAVs, and the targets spread in the environment, so that the UAVs need to carry out a periodical surveillance. The existence of tall buildings in urban environments brings new challenges to the periodical surveillance mission. They may not only block the Line-of-Sight (LoS) between a UAV and a target, but also create some shadow region, so that the surveillance may become invalid, and the UAV may not be able to harvest energy from the sun. The periodical surveillance problem is formulated as an optimization problem to minimize the target revisit time while accounting for the impact of the urban environment. A nearest neighbour based navigation method is proposed to guide the movements of the UAVs. Moreover, we adopt a partitioning scheme to group targets for the purpose of narrowing UAVs’ moving space, which further reduces the target revisit time. The effectiveness of the proposed method is verified via computer simulations.

Autonomous Navigation of a Solar-Powered UAV for Secure Communication in Urban Environments with Eavesdropping Avoidance

This paper considers the navigation of a solar-powered unmanned aerial vehicle (UAV) for securing the communication with an intended ground node in the presence of eavesdroppers in urban environments. To complete this task, the UAV needs to not only fly safely in the complex urban environment, but also take into account the communication performance with the intended node and eavesdroppers. To this end, we formulate a multi-objective optimization problem to plan the UAV path. This problem jointly considers the maximization of the residual energy of the solar-powered UAV at the end of the mission, the maximization of the time period in which the UAV can securely communicate with the intended node and the minimization of the time to reach the destination. We pay attention to the impact of the buildings in the urban environments, which may block the transmitted signals and also create some shadow region where the UAV cannot harvest energy. A Rapidly-exploring Random Tree (RRT) based path planning scheme is presented. This scheme captures the nonlinear UAV motion model, and is computationally efficient considering the randomness nature. From the generated tree, a set of possible paths can be found. We evaluate the security of the wireless communication, compute the overall energy consumption as well as the harvested amount for each path and calculate the time to complete the flight. Compared to a general RRT scheme, the proposed method enables a large time window for the UAV to securely transmit data.

One Way to Fill All the Concave Region in Grid-Based Map

The search space of the path planning problem can greatly affect the running time and memory consumption, for example, the concave obstacle in grid-based map usually leads to the invalid search space. In this paper, the filling container algorithm is proposed to alleviate the concave area problem in 2D map space, which is inspired from the scenario of pouring water into a cup. With this method, concave areas can be largely excluded by scanning the map repeatedly. And the effectiveness has been proved in our experiments.

Fenceless obstacle avoidance method for efficient and safe human–robot collaboration in a shared work space

In a given manufacturing setting where workers or robots are coexisting in a confined area and their movements are not coordinated due to loss in communication or because they are freely ranging relative to each other, the development of an onboard safeguard system for a robot becomes a necessity to reduce accidents while the production efficiency is uncompromised. This article develops a two-dimensional dynamics model that predicts the relative position between a robot’s end-of-arm tooling and an approaching object or threat. The safety strategy applied to the robot is derived from the calculation of three parameters: the time of collision predicted from the linear motion between the approaching object and the robot’s end-of-arm tooling, the relative absolute distance, and the overlapping area ratio. These parameters combined are updated in a cost function that is sufficiently alarming the collision severity of an approaching object in real time. This model enables deployment a safe and a productive collaborative interaction in the manufacturing environment where workers and robots are seemingly moving in close proximity within an open workspace with less safeguard barriers.

Reactive Autonomous Navigation of UAVs for Dynamic Sensing Coverage of Mobile Ground Targets

This paper addresses a problem of autonomous navigation of unmanned aerial vehicles (UAVs) for the surveillance of multiple moving ground targets. The ground can be flat or uneven. A reactive real-time sliding mode control algorithm is proposed that navigates a team of communicating UAVs, equipped with ground-facing video cameras, towards moving targets to increase some measure of sensing coverage of the targets by the UAVs. Moreover, the Voronoi partitioning technique is adopted to reduce the movement range of the UAVs and decrease the revisit times of the targets. Extensive computer simulations, from the simple case with one UAV and multiple targets to the complex case with multiple UAVs and multiple targets, are conducted to demonstrate the performance of the developed autonomous navigation algorithm. The scenarios where the terrain is uneven are also considered. As shown in the simulation results, although the additional VP technique leads to some extra computation burden, the usage of the VP technique considerably reduces the target revisit time compared to the algorithm without this technique.

Learning Reward Function with Matching Network for Mapless Navigation

Deep reinforcement learning (DRL) has been successfully applied in mapless navigation. An important issue in DRL is to design a reward function for evaluating actions of agents. However, designing a robust and suitable reward function greatly depends on the designer’s experience and intuition. To address this concern, we consider employing reward shaping from trajectories on similar navigation tasks without human supervision, and propose a general reward function based on matching network (MN). The MN-based reward function is able to gain the experience by pre-training through trajectories on different navigation tasks and accelerate the training speed of DRL in new tasks. The proposed reward function keeps the optimal strategy of DRL unchanged. The simulation results on two static maps show that the DRL converge with less iterations via the learned reward function than the state-of-the-art mapless navigation methods. The proposed method performs well in dynamic maps with partially moving obstacles. Even when test maps are different from training maps, the proposed strategy is able to complete the navigation tasks without additional training.

Goal-Oriented Obstacle Avoidance with Deep Reinforcement Learning in Continuous Action Space

In this paper, we propose a goal-oriented obstacle avoidance navigation system based on deep reinforcement learning that uses depth information in scenes, as well as goal position in polar coordinates as state inputs. The control signals for robot motion are output in a continuous action space. We devise a deep deterministic policy gradient network with the inclusion of depth-wise separable convolution layers to process the large amounts of sequential depth image information. The goal-oriented obstacle avoidance navigation is performed without prior knowledge of the environment or a map. We show that through the proposed deep reinforcement learning network, a goal-oriented collision avoidance model can be trained end-to-end without manual tuning or supervision by a human operator. We train our model in a simulation, and the resulting network is directly transferred to other environments. Experiments show the capability of the trained network to navigate safely around obstacles and arrive at the designated goal positions in the simulation, as well as in the real world. The proposed method exhibits higher reliability than the compared approaches when navigating around obstacles with complex shapes. The experiments show that the approach is capable of avoiding not only static, but also dynamic obstacles.

A Pedestrian Avoidance Method Considering Personal Space for a Guide Robot

Many methods have been proposed for avoiding obstacles in robotic systems. However, a robotic system that moves without colliding with obstacles and people, while still being mentally safe to the persons nearby, has not yet been realized. In this paper, we describe the development of a method for a mobile robot to avoid a pedestrian approaching from the front and to pass him/her by while preserving the “public distance” of personal space. We assume a robot that moves along a prerecorded path. When the robot detects a pedestrian using a laser range finder (LRF), it calculates the trajectory to avoid the pedestrian considering their personal space, passes by the pedestrian, and returns to the original trajectory. We introduce a virtual target to control the robot moving along the path, such that it can use the same control strategy as for human-following behavior. We carry out experiments to evaluate the method along three routes, in which the robot functioned without problems. The distance between the robot and the pedestrian was 9.3 m, on average, when the robot started to use avoiding behavior, which is large enough to keep a public distance from a pedestrian. When the robot passed by the pedestrian, the minimum distance between them was 1.19 m, which was large enough for passing safely.

On-Board Real-Time Trajectory Planning for Fixed Wing Unmanned Aerial Vehicles in Extreme Environments

A team from the University of Bristol have developed a method of operating fixed wing Unmanned Aerial Vehicles (UAVs) at long-range and high-altitude over Volcán de Fuego in Guatemala for the purposes of volcanic monitoring and ash-sampling. Conventionally, the mission plans must be carefully designed prior to flight, to cope with altitude gains in excess of 3000 m, reaching 9 km from the ground control station and 4500 m above mean sea level. This means the climb route cannot be modified mid-flight. At these scales, atmospheric conditions change over the course of a flight and so a real-time trajectory planner (RTTP) is desirable, calculating a route on-board the aircraft. This paper presents an RTTP based around a genetic algorithm optimisation running on a Raspberry Pi 3 B+, the first of its kind to be flown on-board a UAV. Four flights are presented, each having calculated a new and valid trajectory on-board, from the ground control station to the summit region of Volcań de Fuego. The RTTP flights are shown to have approximately equivalent efficiency characteristics to conventionally planned missions. This technology is promising for the future of long-range UAV operations and further development is likely to see significant energy and efficiency savings.

Navigation in Unknown Dynamic Environments Based on Deep Reinforcement Learning

In this paper, we propose a novel Deep Reinforcement Learning (DRL) algorithm which can navigate non-holonomic robots with continuous control in an unknown dynamic environment with moving obstacles. We call the approach MK-A3C (Memory and Knowledge-based Asynchronous Advantage Actor-Critic) for short. As its first component, MK-A3C builds a GRU-based memory neural network to enhance the robot’s capability for temporal reasoning. Robots without it tend to suffer from a lack of rationality in face of incomplete and noisy estimations for complex environments. Additionally, robots with certain memory ability endowed by MK-A3C can avoid local minima traps by estimating the environmental model. Secondly, MK-A3C combines the domain knowledge-based reward function and the transfer learning-based training task architecture, which can solve the non-convergence policies problems caused by sparse reward. These improvements of MK-A3C can efficiently navigate robots in unknown dynamic environments, and satisfy kinetic constraints while handling moving objects. Simulation experiments show that compared with existing methods, MK-A3C can realize successful robotic navigation in unknown and challenging environments by outputting continuous acceleration commands.

Real-Time Robust and Optimized Control of a 3D Overhead Crane System

A new and advanced control system for three-dimensional (3D) overhead cranes is proposed in this study using state feedback control in discrete time to deliver high performance trajectory tracking with minimum load swings in high-speed motions. By adopting the independent joint control strategy, a new and simplified model is developed where the overhead crane actuators are used to design the controller, with all the nonlinear equations of motions being viewed as disturbances affecting each actuator. A feedforward control is then designed to tackle these disturbances via computed torque control technique. A new load swing control is designed along with a new motion planning scheme to robustly minimize load swings as well as allowing fast load transportation without violating system’s constraints through updating reference trolley accelerations. The stability and performance analysis of the proposed discrete-time control system are demonstrated and validated analytically and practically.

Sensor-based Navigation of Omnidirectional Wheeled Robots Dealing with both Collisions and Occlusions

Navigation tasks are often subject to several constraints that can be related to the sensors (visibility) or come from the environment (obstacles). In this paper, we propose a framework for autonomous omnidirectional wheeled robots that takes into account both collision and occlusion risk, during sensor-based navigation. The task consists in driving the robot towards a visual target in the presence of static and moving obstacles. The target is acquired by fixed – limited field of view – on-board cameras, while the surrounding obstacles are detected by lidar scanners. To perform the task, the robot has not only to keep the target in view while avoiding the obstacles, but also to predict its location in the case of occlusion. The effectiveness of our approach is validated through several experiments.

Autonomous Area Exploration and Mapping in Underground Mine Environments by Unmanned Aerial Vehicles

In this paper, we propose a method of using an autonomous flying robot to explore an underground tunnel environment and build a 3D map. The robot model we use is an extension of a 2D non-holonomic robot. The measurements and sensors we considered in the presented method are simple and valid in practical unmanned aerial vehicle (UAV) engineering. The proposed safe exploration algorithm belongs to a class of probabilistic area search, and with a mathematical proof, the performance of the algorithm is analysed. Based on the algorithm, we also propose a sliding control law to apply the algorithm to a real quadcopter in experiments. In the presented experiment, we use a DJI Guidance sensing system and an Intel depth camera to complete the localization, obstacle detection and 3D environment information capture. Furthermore, the simulations show that the algorithm can be implemented in sloping tunnels and with multiple UAVs.

The Design and Application of a Track-type Autonomous Inspection Robot for Electrical Distribution Room

Electrical distribution equipment inspection is crucial for the electric power industry. With the rapid increase in the number of electrical distribution rooms, an unattended inspection method, for example, autonomous inspection robot, is eagerly desired by the industry to make up for the deficiencies of traditional manual inspection in effectiveness and validity. Existing inspection robots designed for indoor substations are generally lack of practicality, due to the factors such as inspection requirements and robot weight. To bridge the gap between prototype and practicality, in this work, we design the first completely autonomous robotic system, LongSword, which provides a satisfying technical solution for equipment inspection with an optical zoom camera, a thermal imaging camera or a partial discharge detector. Firstly, we design a novel and flexible hardware architecture which allows the robot to move, lift, and rotate in the station to reach any desired position. Secondly, we develop an intelligent software framework which consists of several modules to achieve accurate equipment recognition and reliable failure diagnosis. Thirdly, we achieve an apposite integration of the existing technologies to implement an applicable robotic system that can fulfill the requirements of indoor equipment inspection. There are over 200 LongSwords currently serving about 160 electrical distribution rooms, some of which have been working for more than 1 year. The average precision of device status recognition is up to 99.70%, and the average inspection time of a single device is as short as 13.5 s. The feedback from workers shows that LongSword can significantly improve the efficiency and reliability of equipment inspection, which accelerates the process of setting up unmanned stations.

Self-supervised learning of the biologically-inspired obstacle avoidance of hexapod walking robot

In this paper, we propose an integrated biologically inspired visual collision avoidance approach that is deployed on a real hexapod walking robot. The proposed approach is based on the Lobula giant movement detector (LGMD), a neural network for looming stimuli detection that can be found in visual pathways of insects, such as locusts. Although a superior performance of the LGMD in the detection of intercepting objects has been shown in many collision avoiding scenarios, its direct integration with motion control is an unexplored topic. In our work, we propose to utilize the LGMD neural network for visual interception detection with a central pattern generator (CPG) for locomotion control of a hexapod walking robot that are combined in the controller based on the long short-term memory (LSTM) recurrent neural network. Moreover, we propose self-supervised learning of the integrated controller to autonomously find a suitable setting of the system using a realistic robotic simulator. Thus, individual neural networks are trained in a simulation to enhance the performance of the controller that is then experimentally verified with a real hexapod walking robot in both collision and interception avoidance scenario and navigation in a cluttered environment.

Asymptotically Optimal Deployment of Drones for Surveillance and Monitoring

This paper studies the problem of placing a set of drones for surveillance of a ground region. The main goal is to determine the minimum number of drones necessary to be deployed at a given altitude to monitor the region. An easily implementable algorithm to estimate the minimum number of drones and determine their locations is developed. Moreover, it is proved that this algorithm is asymptotically optimal in the sense that the ratio of the number of drones required by this algorithm and the minimum number of drones converges to one as the area of the ground region tends to infinity. The proof is based on Kershner’s theorem from combinatorial geometry. Illustrative examples and comparisons with other existing methods show the efficiency of the developed algorithm.