China
Africa
Explore chapters and articles related to this topic
Fuzzy Behavior Organization and Fusion for Mobile Robot Reactive Navigation
Published in Takushi Tanaka, Setsuo Ohsuga, Moonis Ali, Industrial and Engineering Applications of Artificial Intelligence and Expert Systems, 2022
Jiancheng Qiu, Michael Walters
In behavior-related architectures, multiple behaviors usually need to be fused to create one set of output for actuators. Several schemes have been used for this, such as hierarchical switching[Brooks86], weighted averaging [Arkin90], context dependent blending[Saffiotti93] and fuzzy multiplexing[Goodridge94]. Context dependent blending developed by [Saffiotti93] has been effectively used to combine fuzzy reactive behaviors and goal-directed behaviors. In their application, a reactive behavior mainly refers to obstacle avoidance. In order to finish a task, a set of subtasks are exhaustibly planned by an off-line planner. A goal-directed behavior is activated for a subtask sequentially and combined with obstacle avoidance. The weight for goal-directed behavior is subdued with the obstacle avoidance weight and the outputs are defuzzified with the centroid calculation. Although the scheme works efficiently, only a few behaviors can be active at the same time. The success of this scheme depends on a detailed task planning. For example, a set of walls of a concave-shaped barrier must be provided as different subtasks for wall following behavior to execute in order to reach the other side of the barrier. Otherwise, the robot may be trapped in the concave area moving endlessly.
Autonomous Inspection for Industrial Assets
Published in Diego Galar, Uday Kumar, Dammika Seneviratne, Robots, Drones, UAVs and UGVs for Operation and Maintenance, 2020
Diego Galar, Uday Kumar, Dammika Seneviratne
In robotics, obstacle avoidance is the task of satisfying some control objective subject to non-intersection or non-collision position constraints. In unmanned air vehicles, it is a hot topic. What is critical about obstacle avoidance in this area is the growing need of the usage of UAVs in urban areas for especially military applications where it can be very useful in city wars. Normally, obstacle avoidance is considered to be distinct from path planning in that one is usually implemented as a reactive control law while the other involves the pre-computation of an obstacle-free path along which a controller will then guide a robot. With recent advances in the autonomous vehicle sector, a good and dependable obstacle avoidance feature of a driverless platform is also required to have a robust obstacle detection module (Wikipedia, 2019).
Safe and Effective Autonomous Decision Making in Mobile Robots
Published in Jitendra R. Raol, Ajith K. Gopal, Mobile Intelligent Autonomous Systems, 2016
It is important to note that the whole map is not updated in every cycle even if obstacles are detected. Only the circle with a radius of 15 m and robot at its centre is cleared. Next, the path from current robot position to the goal is recalculated. When the cell has big cost, the robot steers away from it to the adjacent cell with lowest cost. This facilitates obstacle avoidance while planning the path effectively. This procedure is repeated in every cycle. The flowchart shown in Figure 19.13 explains this in detail.
Navigation system with SLAM-based trajectory topological map and reinforcement learning-based local planner
Published in Advanced Robotics, 2021
Wuyang Xue, Peilin Liu, Ruihang Miao, Zheng Gong, Fei Wen, Rendong Ying
Our sub-target module is based on VFH. Compared with other traditional obstacle avoidance approaches, VFH is the most appropriate to our DRL-based local planner. Our local planner needs a target position as input. As mentioned in Section 1, there are several types of popular traditional obstacle avoidance approaches, such as APF, VFH, DWA. APF directly computes a command velocity instead of a target position, which is not appropriate to the input of our local planner. VFH divides space into several sectors and evaluate the sectors to choose the best direction. We can compute a sub-target with the best sector. DWA samples and evaluates different trajectories. The end of the best trajectory can be the sub-target. However, compared with VFH, the sampling step of DWA is computationally inefficient. Other optimization-based approaches are also inefficient to the task of generating a sub-target.
Path Planning for Multiple Targets Interception by the Swarm of UAVs based on Swarm Intelligence Algorithms: A Review
Published in IETE Technical Review, 2021
Abhishek Sharma, Shraga Shoval, Abhinav Sharma, Jitendra Kumar Pandey
Various approaches have been proposed for solving the issues related to target interception by a swarm of the mobile robot [51,52]. This section presents the different parameters and assumptions related to this problem as shown in Table 4. The parameters presented in the table are: Prior knowledge of the environment – the level of information available before and during the interception process. Information can be given a-priori or can be acquired during the process using real-time data flow.Type of communication – agents can share information (cooperative) or act independently without sharing information with other agents (non-cooperative).Obstacle avoidance – some methods focus on the actual interception process only, while others also consider collision avoidance with other agents and obstacles.Location of targets – like the prior knowledge of the environment, some methods require prior knowledge about the locations of the targets (as well as their trajectories) while others use real-time data acquisition tools for determining targets’ locations.Number of targets – the number of targets has a significant impact on the process complexity.
A new strategy for rear-end collision avoidance via autonomous steering and differential braking in highway driving
Published in Vehicle System Dynamics, 2020
Qingjia Cui, Rongjun Ding, Xiaojian Wu, Bing Zhou
To overcome the limitations of the AEB system, the other issue for collision avoidance focuses on emergency steering [10,11]. Compared with obstacle avoidance by braking, the steering manoeuvre is more complex, but it can effectively prevent the collision, even when the host vehicle is near the preceding vehicle. The general framework of obstacle avoidance consists of two aspects: generating a collision-free trajectory for automated driving and controlling the vehicle to follow this path. In [12], the author presented a 3-D virtual dangerous potential field to generate a collision-free trajectory and designed a path-tracking framework using a multiconstrained model predictive control (MPC) for automated steering. Nilsson et al. [13] focused on the obstacle avoidance path-planning problem with a framework of MPC. The researcher proposed the approach that was formulated as two MPC problems for longitudinal motion and lateral motion planning, whereas other papers considered only lateral motion planning. To decrease the computational cost, Hayashi et al. [14] and Brannstrom et al. [15] developed a method with collision-free trajectory as the identical acr and the required steering input, which was calculated according to the vehicle's characteristics based on a linear model in steady-state cornering. This autonomous steering system can avoid the collision well at lower speeds, but in highway driving, the vehicle dynamics exhibit nonlinear characteristics when the vehicle performs an emergency steering manoeuvre. As a result, the steering controller, regardless of the limits of tyre adhesion, cannot satisfy the accuracy requirement of path tracking at high speeds.