2. Special Issue Summary

Construction and mining are the industrial sectors that have been using many advanced technologies in field robotics, particularly for earthmoving processes. Autonomous systems have been increasingly applied to such equipment as haul trucks, excavators, bulldozers, loaders, telehandlers, graders, and compactors that present robotic platforms operating in various modes of autonomy. The literature survey by Nguyen & Ha provides a comprehensive review of significant contributions and recent advances using robotics and automation technologies, with an emphasis on operations in volatile environments and on the collaboration of those platforms [Reference Nguyen and Ha1].

For onsite execution of robotic tasks, ground-based platforms are often required to navigate in offroad, unstructured environment subject to terrain constraints and dynamic obstacles, such as in emergency robotics, search and rescue, agriculture, forestry, or in military operations. For unmanned ground vehicles autonomous operating in these conditions, a 3D navigation system is proposed by Zhou, Yi et al., using an octree-based terrain constraint analysis module with an octomap library for path planning and trajectory generation to enable mapless navigation with negative obstacles [Reference Zhou, Yi, Zhang, Chen, Yang, Han and Zhang2].

For the sake of enhancing work capacity and task efficiency in industry, research attention has been paid to the interactions, either among machines or between machines and humans. Addressing this problem for robotic collaboration with manipulators, Kanik, Ayit et al. propose an admittance control scheme for human–robot interface via a shoulder haptic device [Reference Kanık, Ayit, Dede and Tatlicioglu3]. The dynamics ratio of the end-effector’s velocity and the interaction force change is controlled to follow a desired admittance function while avoiding the obstacles and driving the trajectory tracking error to zero with a PI controller. The results are promising in the context of achieving high performance and safe collaboration between a human operator and robots.

In dealing with interactions between the robot’s end-effector or platform’s tool and the environment, the fourth paper of this special issue by Muñoz, López et al. presents a new approach to geometrically constrained path planning for robotic grasping using fast marching square (FM $^2$ ) and differential evolution (DE) techniques [Reference Muñoz, López, Quevedo, Barber, Garrido and Moreno4]. For this, geometrical constraints imposed by the robot arm and hand are coded in some cost functions correspondingly for the optimization purpose. The FM²-based path planning DE-based evolutionary algorithms are then applied to obtain an optimal path for a certain grasping task.

Addressing also the robotic grasping problem, Lu, Cai et al. propose a target-oriented push-grasping system to sort and pick out impurities in a dense environment using the synergy between pushing and grasping strategies [Reference Lu, Cai, Lu, Cao, Guo and Wang5]. An attention module is first developed using a vision-based algorithm for target saliency detection and density-based occluded-region inference. The identified targets are then picked out by a push–grasp synergy framework incorporating a convolutional network to enable the robot’s ability to learn pushing actions and isolate the target for grasping.

To improve the grasping capacity of robots, a new self-adaptive underactuated robot hand with rigid-flexible coupling fingers is designed, as reported in Su, Wang et al. [Reference Su, Wang, Lu and Wang6]. The robotic hand, driven by four servomotors, consists of three fingers with seven degrees of freedom. As per the design, it can perform flexion and extension motion to present a four-bar linkage. A belt is fitted in the inner direction of the finger to make it flexible while maintaining a rigid structure in the outer direction. This rigid-flexible finger is aimed at grasping objects of different shapes, sizes, and hardness.

To manage such tasks as delivering components in an uneven environment, some service robots are required to possess the capability of walking as humans or animals besides scooting along on wheels. Motivated by advances in wall-climbing robots, Bisht, Pathak, and Panigrahi present the work on modelling, simulation, and experimental validation of wheel and arm locomotion for wall-to-wall transition, obstacle avoidance, and uneven surface navigation [Reference Bisht, Pathak and Panigrahi7]. The robot consists of two mobile modules connected with an inchworm-inspired robot arm mechanism. Particle swarm optimization and PID control algorithms are used to effectively control the process of switching between arm and wheel configurations to accommodate the wall-climbing navigation of the robot.

The last paper of the issue is by Le, Vu, and Vo, devoted to the development of an access control system using 3D face recognition. Unlike 2D face recognition, the 3D recognition algorithm can identify features with improved robustness against variations in pose and illumination conditions as well as facial expressions [Reference Le, Vu and Vo8]. Covering preprocessing, feature extraction, and classification stages, the proposed algorithm combines the principal component analysis and linear discriminant analysis for feature extraction and k-nearest neighbour for classification. The resulting access control system may find many applications in enhancing security in a variety of work environments.