In this study, an autonomous robot navigation system is designed for live working on distribution line. The developed system features a real‐time detection and motion planning system, incorporating a manipulator capable of grasping power components. In order to accurately identify targets, the authors propose an object detection method based on the Larger Scale ‘You Only Look Once’ Version 4 (LS‐YOLOv4) algorithm for detecting the insulators and drop fuses. The LS‐YOLOv4 extracts features of power components by Convolutional Neural Network (CNN), and then performs feature fusion. Then the authors develop a motion planning method based on the Node Control Optimal Rapidly Exploring Random Trees (NC‐RRT*), which can drive the robot to realise the autonomous robot motion planning and obstacle avoidance. On the grasping function, the authors present a reliable Lightweight‐based Convolutional Neural Network (L‐CNN) grasping point detection method. Finally, the authors evaluate fully autonomous robotic system in both simulated and real‐world experiments. The experimental results demonstrate that the proposed system can effectively identify the target and complete the grasping task in an efficient way. Notably, the proposed motion planning method can take into account both planning efficiency and accuracy to manipulation tasks.

A systematic solution is developed to improve the autonomous capability of the quadruped robot with a manipulator, such as navigation, recognition and operation. The developed system adopts novel software, hardware system and system architecture, including a specially designed environment awareness system (EAS). Based on the camera and LiDAR on the EAS, the recognition of multiple common targets, such as the leader, door, window and bag, is achieved. In terms of navigation, a location method is built, that combines the laser odometer and global positioning system. A mapping and path planning module is designed by the Robot‐centric Elevation Mapping algorithm and the rapidly exploring rand tree algorithm. For operation, a real‐time target grasp detection system is proposed based on the You Only Look Once v5 algorithm to improve the success rate of tasks. The whole system is integrated based on the task relevance scheduling strategy to reduce the computational complexity. The tightly integrated system and the subsystems are evaluated by conducting simulations and physical experiments in robot recognition, navigation and operation. Extensive experiments show that the proposed framework can better achieve the autonomous navigation and operation of the quadruped robot with a manipulator. Notably, the proposed framework is still effective when facing dynamic objects. In addition, the system can be easily extended to other forms of mobile robot.

Quadruped robots equipped with manipulators need fast and precise grasping and detection algorithms for the transportation of disaster relief supplies. To address this, we developed a framework for these robots, comprising a Grasp Detection Controller (GDC), a Joint Trajectory Planner (JTP), a Leg Joint Controller (LJC), and a Manipulator Joint Controller (MJC). In the GDC, we proposed a lightweight grasp detection CNN based on DenseBlock called DES-LGCNN, which reduced algorithm complexity while maintaining accuracy by incorporating UP and DOWN modules with DenseBlock. For JTP, we optimized the model based on quadruped robot kinematics to enhance wrist camera visibility in dynamic environments. We integrated the network and model into our homemade robot control system and verified our framework through multiple experiments. First, we evaluated the accuracy of the grasp detection algorithm using the Cornell and Jacquard datasets. On the Jacquard dataset, we achieved a detection accuracy of 92.49% for grasp points within 6 ms. Second, we verified its visibility through simulation. Finally, we conducted dynamic scene experiments which consisted of a dynamic target scenario (DTS), a dynamic base scenario (DBS), and a dynamic target and base scenario (DTBS) using an SDU-150 physical robot. In all three scenarios, the object was successfully grasped. The results demonstrate the effectiveness of our framework in managing dynamic environments throughout task execution.

The quadruped robots have superior adaptability to complex terrains, compared with tracked and wheeled robots. Therefore, leader-following can help quadruped robots accomplish long-distance transportation tasks. However, long-term following has to face the change of day and night as well as the presence of interference. To solve this problem, we present a day/night leader-following method for quadruped robots toward robustness and fault-tolerant person following in complex environments. In this approach, we construct an Adaptive Federated Filter algorithm framework, which fuses the visual leader-following method and the LiDAR detection algorithm based on reflective intensity. Moreover, the framework uses the Kalman filter and adaptively adjusts the information sharing factor according to the light condition. In particular, the framework uses fault detection and multisensors information to stably achieve day/night leader-following. The approach is experimentally verified on the quadruped robot SDU-150 (Shandong University, Shandong, China). Extensive experiments reveal that robots can identify leaders stably and effectively indoors and outdoors with illumination variations and unknown interference day and night.

This study presents a path-planning approach toward efficient obstacle avoidance in dynamic environments. The developed approach features the awareness of the topological structure of the dynamic environment at a planning instant. It is achieved by employing a homology class path planner to generate a set of non-homotopy global paths. The global paths are cast into tree structures separately and optimized by the developed sampling-based path-planning methods. This mechanism can adaptively adjust the optimizing step size according to the change in the dynamic environment, and the sampling module uses the Gaussian Mixture Model (GMM) Optimizer to control the sampling space. The approach seeks the globally optimal path as it maintains and optimizes homology classes of admissible candidate paths of distinctive topologies in parallel. We conduct various experiments in dynamic environments to verify the developed method’s effectiveness and efficiency. It is demonstrated that the developed method can perform better than the state of the art.