Obstacle Collision Research Articles

Velocity Obstacle Based on Vertical Ellipse for Multi-Robot Collision Avoidance

Bounding volume based approaches in velocity obstacle (VO) provide a good solution for collision avoidance of mobile robots with uncertainty. However, the VO built with the bounding footprint always has over-constraining problems which may lead to conservative maneuvers of the mobile robots. Addressing this problem, a vertical ellipse based velocity obstacle (VEVO) collision avoidance method is proposed in this paper. The method mitigates the over-constraining situation by building the footprint probability ellipse whose minor axis is vertical to the direction of the obstacle to minimize the VO area. Based on VEVO, a DWA (Dynamic Window Approach) integrated method is proposed to provide a set of available velocities in speed selection. According to different collision avoidance objectives like collision safety, shortest time consumption and shortest trajectory length, a multi-objective velocity selecting strategy is proposed to provide optimal velocities for motion planning. Furthermore, a dynamic local path adjustment method is proposed to help robots react to the closest obstacle (dynamic or static) according to different collision safety requirements. We validate our methods in a simulated workspace with different numbers of robots going to their goal points. Experimental results show VEVO method could improve the collision avoidance performance in crowded multi-robot environment and robots could achieve their different objectives when suitable parameters are set in the velocity evaluation function. The proposed dynamic local path adjustment method only affects the trajectories in local areas and could ensure collision avoidance safety and performance at the same time.

An Urban Autodriving Algorithm Based on a Sensor-Weighted Integration Field with Deep Learning

This paper proposes two algorithms for adaptive driving in urban environments: the first uses vision deep learning, which is named the sparse spatial convolutional neural network (SSCNN); and the second uses a sensor integration algorithm, named the sensor-weighted integration field (SWIF). These algorithms utilize three kinds of sensors, namely vision, Light Detection and Range (LiDAR), and GPS sensors, and decide critical motions for autonomous vehicle, such as steering angles and vehicle speed. SSCNN, which is used for lane recognition, has 2.7 times faster processing speed than the existing spatial CNN method. Additionally, the dataset for SSCNN was constructed by considering both normal and abnormal driving in 7 classes. Thus, lanes can be recognized by extending lanes for special characteristics in urban settings, in which the lanes can be obscured or erased, or the vehicle can drive in any direction. SWIF generates a two-dimensional matrix, in which elements are weighted by integrating both the object data from LiDAR and waypoints from GPS based on detected lanes. These weights are the integers, indicating the degree of safety. Based on the field formed by SWIF, the safe trajectories for two vehicles’ motions, steering angles, and vehicle speed are generated by applying the cost field. Additionally, to flexibly follow the desired steering angle and vehicle speed, the Proportional-Integral-Differential (PID) control is moderated by an integral anti-windup scheme. Consequently, as the dataset considers characteristics of the urban environment, SSCNN is able to be adopted for lane recognition on urban roads. The SWIF algorithm is also useful for flexible driving owing to the high efficiency of its sensor integration, including having a resolution of 2 cm per pixel and speed of 24 fps. Thus, a vehicle can be successfully maneuvered with minimized steering angle change, without lane or route departure, and without obstacle collision in the presence of diverse disturbances in urban road conditions.

Multi UAV Cluster Control Method Based on Virtual Core in Improved Artificial Potential Field

In this paper, a method of multi UAV cluster control based on improved artificial potential field (APF) is proposed. The k-means method is used to integrate and optimize the attractive force between UAVs, and the concept of virtual core is introduced to realize the cluster control and adaptive formation flight of multiple UAVs. The attractive disturbance component of the target point is introduced and the backtracking-filling method is proposed to solve the local minimum problem in the APF. The repulsion force in the APF can realize obstacle avoidance and collision avoidance, and the virtual core can control the UAV cluster to fly to the target point under the attractive force of potential field, so as to realize the track planning and multi aircraft cooperative task. In the process of cluster flight when the UAV fails, merges or dispatches, the method can realize cluster reconfiguration and the cluster control effect and task execution success rate can be improved. The simulation experiments in virtual APF and urban environment APF show the effectiveness of this method.

Distributed Drone Traffic Coordination Using Triggered Communication

This paper proposes a low complexity distributed multi-agent coordination algorithm for agents to reach their target positions in dense traffic under limited communication. Each single-integrator agent is limited to communicating with only one other agent at a time in consideration of limited bandwidth. We adapt the Velocity Obstacle collision avoidance method from literature to the limited communication problem by incorporating Voronoi Cells and repulsion in our hybrid algorithm. We also introduce a priority system for distributed coordination to avoid deadlocks and livelocks by having agent pairs make mutual decisions based on each agent’s conditional priority. An event trigger-based communication protocol is designed to determine when and to whom to communicate. Our method’s effectiveness is demonstrated in simulations including 100 randomized scenarios of 50 agents. The simulations show that our proposed algorithm enables agents to reach their assigned target positions without deadlock and collision while requiring an average communication rate that is significantly lower than the control frequency.

Who’s the boss? Arbitrating control authority between a human driver and automation system

Progress toward the fully automated highway will first require that manual and automatic control be successfully combined. Determining a combination that preserves the best performance features of human and automatic control yet allows either driver to cover for the faults of the other is a challenging problem. In this study, we invited 11 participants to drive a simulated vehicle through a course with obstacles to investigate the ability of human-automation teams to cover for human and automation faults. We developed the automation system using model predictive control and implemented three schemes under which the human would share control with the automation. In Autopilot, the human driver initiated a takeover with a button press whereas in Active Safety the automation initiated a takeover when it anticipated an obstacle collision. In Haptic Shared Control the human was free to invoke a transition by activating or relaxing muscles. In addition, we included two baseline conditions in which control was given in whole to either the human or the automation. We compared performance in the five conditions by analyzing obstacle hits and metrics related to driving maneuvers around the obstacles that were avoided. Relative to individual human or automatic driver performance, we found that control sharing reduced obstacle hits under fault conditions but also occasionally resulted in obstacle hits under no-fault conditions. Our findings further indicated that team performance suffered most under Autopilot for automation faults and suffered most under Active Safety for human faults. Haptic Shared Control supported the best overall team performance.

Impact of Caffeine Ingestion on the Driving Performance of Anesthesiology Residents After 6 Consecutive Overnight Work Shifts.

Residency training in anesthesiology involves care of hospitalized patients and necessitates overnight work, resulting in altered sleep patterns and sleep deprivation. Caffeine consumption is commonly used to improve alertness when fatigued after overnight work, in preparation for the commute home. We studied the impact of drinking a caffeinated energy drink (160 mg of caffeine) on driving performance in a high-fidelity, virtual reality driving simulator (Virginia Driving Safety Laboratory using the Driver Guidance System) in anesthesiology resident physicians immediately after 6 consecutive night-float shifts. Twenty-six residents participated and were randomized to either consume a caffeinated or noncaffeinated energy drink 60 minutes before the driving simulation session. After a subsequent week of night-float work, residents performed the same driving session (in a crossover fashion) with the opposite intervention. Psychomotor vigilance task (PVT) testing was used to evaluate reaction time and lapses in attention. After 6 consecutive night-float shifts, anesthesiology residents who consumed a caffeinated energy drink had increased variability in driving for throttle, steering, and speed during the first 10 minutes of open-road driving but proceeded to demonstrate improved driving performance with fewer obstacle collisions (epoch 2: 0.65 vs 0.87; epoch 3: 0.47 vs 0.95; P = .03) in the final 30 minutes of driving as compared to driving performance after consumption of a noncaffeinated energy drink. Improved driving performance was most apparent during the last 30 minutes of the simulated drive in the caffeinated condition. Mean reaction time between the caffeine and noncaffeine states differed significantly (278.9 ± 29.1 vs 294.0 ± 36.3 milliseconds; P = .021), while the number of major lapses (0.09 ± 0.43 vs 0.27 ± 0.55; P = .257) and minor lapses (1.05 ± 1.39 vs 2.05 ± 3.06; P = .197) was not significantly different. After consuming a caffeinated energy drink on conclusion of 6 shifts of night-float work, anesthesiology residents had improved control of driving performance variables in a high-fidelity driving simulator, including a significant reduction in collisions as well as slightly faster reaction times.

Accurate Estimation of Ground Obstacle Collision Probability During ILS Approach

Accurate Estimation of Ground Obstacle Collision Probability During ILS Approach

Real-Time Feasible Footstep Planning for Bipedal Robots in Three-Dimensional Environments Using Particle Swarm Optimization

Footstep planning in various three-dimensional environments is formulated as an optimization problem, which is solved using a particle swarm optimi-zation. The objective function for optimization is designed to achieve real-time footstep planning considering arrival at the goal with effective not only foot placements but also walking periods, kinematic constraints: obstacle avoidance and hardware limitations, and dynamic constraints for the bipedal dynamics: stability while walking and feasibility of footsteps in a walking pattern generator. Specifically, optimization objectives are to minimize remaining distance to the goal, lateral movement, rotational movement, and walking period variation. Three penalties are also consid-ered depending on the situations. The first penalty is for obstacle collision avoidance along with prevention of unstable walking due to excessive footstep height variation. The second penalty is for walking satisfying stable zero-moment point condition along with the foot collision avoidance. The third penalty is for feasible footstep planning in a walking pattern generator. Any approxi-mation or precomputation is not required for the proposed footstep planning method. The validity of the proposed method is verified through experiments in various 3-D environments with static and dynamic obstacles.

Improvement of the anti­collision method "Velocity Obstacle" by taking into consideration the dynamics of an operating vessel

This paper has proposed an algorithm that accounts for the dynamics of an operating vessel within the Velocity Obstacle collision prevention method. This algorithm provides the basis for selecting joint maneuvers by course and speed with the assigned start for divergence with multiple targets by determining, applying a sorting method, a representative set of acceptable maneuver options. In order to employ the method of sorting, we have selected the ranges of change in the maneuver parameters (course and speed) and performed their sampling at a small enough step. One finds, for all pairs of discrete values of change in the course and speed, taking into consideration the dynamics of an operating vessel, a trajectory and duration of a maneuver, with determining, at the time it ends, the location of an operating vessel and targets, as well as it is to be established if it is accompanied by the intersection of target danger domains. If none of these domains is crossed, the maneuver option is considered acceptable. The totality of such joint changes in course and speed, derived from sorting, creates a set of permissible maneuver options. When finding this set, the dynamics of an operating vessel is taken into consideration simplistically. It is believed that the turns are performed at a constant angular speed, a change in the linear speed at braking can be represented by a power polynomial of second order, and changes in course and speed in the joint maneuver are independent. The targets involve circular danger domains whose center is shifted, from a target's center of mass to the stern, by 1/3 of the domain's radius. This radius has been amended to include the size of a target and an operating vessel.To test the resulting algorithm, the software has been developed in the Borland Delphi programming language. Employing it for calculations has confirmed the operability of the algorithm. That enables the derivation, in real time, of a set of velocity vectors for divergence taking into consideration the dynamics of own vessel, which makes it possible to improve the accuracy of forecasting and the safety of calculated maneuvers. The use of displaced circular danger domains for targets makes it possible to take into account the unequal degree of risk when crossing their course along the bow and stern

Mobility-Related Gaze Training in Individuals With Glaucoma: A Proof-of-Concept Study.

PurposeOlder adults with glaucoma show inappropriate gaze strategies during routine mobility tasks. Furthermore, glaucoma is a risk factor for falling and colliding with objects when walking. However, effective interventions to rectify these strategies and prevent these adverse events are scarce. We designed a gaze training program with the goal of providing proof-of-concept that we could modify mobility-related gaze behavior in this population.MethodsA total of 13 individuals with moderate glaucoma participated in this study. We taught participants general and task-specific gaze strategies over two 1-hour sessions. To determine the efficacy of this gaze training program, participants performed walking tasks that required accurate foot placement onto targets and circumventing obstacles before and after training. We used a mobile eye tracker to quantify gaze and a motion-capture system to quantify body movement.ResultsAfter training, we found changes in the timing between gaze shifts away from targets relative to stepping on them (P < 0.05). In the obstacle negotiation task, we found a greater range of gaze shifts early in walking trials and changes in the timing between gaze shifts away from obstacles after training (P < 0.05), each suggesting better route planning. A posttraining reduction in foot-placement error and obstacle collisions accompanied these changes (P < 0.05).ConclusionsOur results demonstrated that it is possible to modify mobility-related gaze behavior and mobility performance in older adults with glaucoma.Translational RelevanceThis study provides proof-of-concept for a gaze training program for glaucoma. A larger, randomized controlled trial is warranted.

Obstacles avoidance for autonomous marine vehicles based on the model predictive control

In this paper, a new obstacle collision avoidance algorithm based on the model predictive control is proposed. The algorithm has the following advantages. First, the vehicle nonlinear dynamics as well as restrictions on the phase variables and the controls defining forbidden areas for the vehicle are taken into account. Second, an adaptation of the algorithm to the time-varying number of obstacles according to information coming in real-time from sensors is implemented. Third, the return to the prescribed trajectory after obstacles avoidance is provided. The ship collision avoidance on the sea surface is considered as an example.

A Review on IoT Deep Learning UAV Systems for Autonomous Obstacle Detection and Collision Avoidance

Advances in Unmanned Aerial Vehicles (UAVs), also known as drones, offer unprecedented opportunities to boost a wide array of large-scale Internet of Things (IoT) applications. Nevertheless, UAV platforms still face important limitations mainly related to autonomy and weight that impact their remote sensing capabilities when capturing and processing the data required for developing autonomous and robust real-time obstacle detection and avoidance systems. In this regard, Deep Learning (DL) techniques have arisen as a promising alternative for improving real-time obstacle detection and collision avoidance for highly autonomous UAVs. This article reviews the most recent developments on DL Unmanned Aerial Systems (UASs) and provides a detailed explanation on the main DL techniques. Moreover, the latest DL-UAV communication architectures are studied and their most common hardware is analyzed. Furthermore, this article enumerates the most relevant open challenges for current DL-UAV solutions, thus allowing future researchers to define a roadmap for devising the new generation affordable autonomous DL-UAV IoT solutions.

FumeBot: A Deep Convolutional Neural Network Controlled Robot

This paper describes the development of a convolutional neural network for the control of a home monitoring robot (FumeBot). The robot is fitted with a Raspberry Pi for on board control and a Raspberry Pi camera is used as the data feed for the neural network. A wireless connection between the robot and a graphical user interface running on a laptop allows for the diagnostics and development of the neural network. The neural network, running on the laptop, was trained using a supervised training method. The robot was put through a series of obstacle courses to test its robustness, with the tests demonstrating that the controller has learned to navigate the obstacles to a reasonable level. The main problem identified in this work was that the neural controller did not have memory of past actions it took and a past state of the world resulting in obstacle collisions. Options to rectify this issue are suggested.

Cooperative Aerial Payload Transport Guided by an In Situ Human Supervisor

This paper defines and analyzes a cooperative aerial payload transport mission in which an onsite human supervisor provides haptic motion cues. While multiple small aerial robots (e.g., quadcopters) carry the weight of the payload, a human specifies traversal direction changes to the quadcopter team by exerting a pushing force on the payload. We show how the principle of linear momentum can be used for limited human intent recognition in this supervisory control application. This problem is formulated as real-time path planning of the quadcopter team using a continuum deformation cooperative control protocol that assures avoidance of interagent and obstacle collisions as well as safety of the human supervisor.

TOWARDS CONTINUOUS CONTROL FOR MOBILE ROBOT NAVIGATION: A REINFORCEMENT LEARNING AND SLAM BASED APPROACH

Abstract. We introduce a new autonomous path planning algorithm for mobile robots for reaching target locations in an unknown environment where the robot relies on its on-board sensors. In particular, we describe the design and evaluation of a deep reinforcement learning motion planner with continuous linear and angular velocities to navigate to a desired target location based on deep deterministic policy gradient (DDPG). Additionally, the algorithm is enhanced by making use of the available knowledge of the environment provided by a grid-based SLAM with Rao-Blackwellized particle filter algorithm in order to shape the reward function in an attempt to improve the convergence rate, escape local optima and reduce the number of collisions with the obstacles. A comparison is made between a reward function shaped based on the map provided by the SLAM algorithm and a reward function when no knowledge of the map is available. Results show that the required learning time has been decreased in terms of number of episodes required to converge, which is 560 episodes compared to 1450 episodes in the standard RL algorithm, after adopting the proposed approach and the number of obstacle collision is reduced as well with a success ratio of 83% compared to 56% in the standard RL algorithm. The results are validated in a simulated experiment on a skid-steering mobile robot.

Route selection and obstacle avoidance with a short-range haptic sensory substitution device✰

Abstract The design of Sensory Substitution Devices (SSDs) often relies on the belief that the information supplied by the devices should allow the construction of spatial mental representations on the basis of which routes are planned. This study, in contrast, illustrates that navigation using an SSD can be conceived as an on-line, dynamic process, without the need for establishing a predefined plan or model of the task prior to its execution. We analyzed route selection performed with a vibrotactile SSD that could detect environmental surfaces only within a short spatial range, limiting the availability of information about remote parts of the environment to be navigated. Sixty sighted participants performed a navigation task that involved the goal of reaching a target destination while avoiding five obstacles (placed in randomly predetermined configurations). Three groups of participants differed in the sensory modality used (restricted visual, acoustic + vibrotactile, and restricted visual + vibrotactile). While participants in the visual condition had fewer obstacle collisions and reached the target location sooner, the groups coincided to a large extent in terms of the routes that they followed. Furthermore, the routes selected by participants in all groups conformed well to routes predicted by a dynamic model of visually-guided locomotion (Fajen and Warren, 2003). These findings show that local, limited information about environmental layout can support route selection equivalent to that seen when information about the full environmental layout is available.

An Improved A-Star Algorithm Considering Water Current, Traffic Separation and Berthing for Vessel Path Planning

A traditional A-Star (A*) algorithm generates an optimal path by minimizing the path cost. For a vessel, factors of path length, obstacle collision risk, traffic separation rule and manoeuvrability restriction should be all taken into account for path planning. Meanwhile, the water current also plays an important role in voyaging and berthing for vessels. In consideration of these defects of the traditional A-Star algorithm when it is used for vessel path planning, an improved A-Star algorithm has been proposed. To be specific, the risk models of obstacles (bridge pier, moored or anchored ship, port, shore, etc.) considering currents, traffic separation, berthing, manoeuvrability restriction have been built firstly. Then, the normal path generation and the berthing path generation with the proposed improved A-Star algorithm have been represented, respectively. Moreover, the problem of combining the normal path and the berthing path has been also solved. To verify the effectiveness of the proposed A-Star path planning methods, four cases have been studied in simulation and real scenarios. The results of experiments show that the proposed A-Star path planning methods can deal with the problems denoted in this article well, and realize the trade-off between the path length and the navigation safety.

Velocity Obstacle Approaches for Multi-Agent Collision Avoidance

This paper presents a critical analysis of some of the most promising approaches to geometric collision avoidance in multi-agent systems, namely, the velocity obstacle (VO), reciprocal velocity obstacle (RVO), hybrid-reciprocal velocity obstacle (HRVO) and optimal reciprocal collision avoidance (ORCA) approaches. Each approach is evaluated with respect to increasing agent populations and variable sensing assumptions. In implementing the localized avoidance problem, the author notes a problem of symmetry not considered in the literature. An intensive 1000-cycle Monte Carlo analysis is used to assess the performance of the selected algorithms in the presented conditions. The ORCA method is shown to yield the most scalable computation times and collision likelihood in the presented cases. The HRVO method is shown to be superior than the other methods in dealing with obstacle trajectory uncertainty for the purposes of collision avoidance. The respective features and limitations of each algorithm are discussed and presented through examples.

Optimal Path-Planning of Nonholonomic Terrain Robots for Dynamic Obstacle Avoidance Using Single-Time Velocity Estimator and Reinforcement Learning Approach

A single-time velocity estimator-based reinforcement learning (RL) algorithm, integrated with a chaotic metaheuristic optimization technique is proposed in this article for the optimal path-planning of the nonholonomic robots considering a moving/stationary obstacle avoidance strategy. The additional contribution of the present study is by employing the Terramechanics principles to incorporate the effects of wheel sinkage into the deformable terrain on the dynamics of the robot aiming to find the optimal compensating force/torque magnitude to sustain a robust and smooth motion. The designed systematic control-oriented system incorporates a cost function of weighted components associated with the target-tracking and the obstacle avoidance. The designed velocity estimator contributes to the finite-state Markov decision process (MDP) in order to train the transition probabilities of the problem objectives. Based on the obtained results, the optimal solution for the Q-learning in terms of the adjusting factor for the minimized tracking error and obstacle collision risk propagation profiles is found at 0.22. The results further confirm the promising capacity of the proposed optimization-based RL algorithm for the collision avoidance control of the nonholonomic robots on deformable terrains.

ObstacleWatch

Walking while using a smartphone is becoming a major pedestrian safety concern as people may unknowingly bump into various obstacles that could lead to severe injuries. In this paper, we propose ObstacleWatch, an acoustic-based obstacle collision detection system to improve the safety of pedestrians who are engaged in smartphone usage while walking. ObstacleWatch leverages the advanced audio hardware of the smartphone to sense the surrounding obstacles and infers fine-grained information about the frontal obstacle for collision detection. In particular, our system emits well-designed inaudible beep signals from the smartphone built-in speaker and listens to the reflections with the stereo recording of the smartphone. By analyzing the reflected signals received at two microphones, ObstacleWatch is able to extract fine-grained information of the frontal obstacle including the distance, angle and size for detecting the possible collisions and to alert users. Our experimental evaluation under two real-world environments with different types of phones and obstacles shows that ObstacleWatch achieves over 92% accuracy in predicting obstacle collisions with distance estimation errors at about 2 cm. Results also show that ObstacleWatch is robust to different sizes of objects and is compatible to different phone models with low energy consumption.