Publications

Jumping-height-and-distance (JHD) active adjustment capability is important for jumping robots to overcome different sizes of obstacle. This paper proposes a new structural-parameter-based JHD active adjustment approach for our previous jumping robot. First, the JHD adjustments, modifying the lengths of different legs of the robot, are modelled and simulated. Then, three mechanisms for leg-length adjustment are proposed and compared, and the screw-and-nut mechanism is selected. And for adjusting of different structural-parameters using this mechanism, the one with the best JHD adjusting performance and the lowest mechanical complexity is adopted. Thirdly, an obstacle-distance-and-height (ODH) detection method using only one infrared sensor is designed. Finally, the performances of the proposed methods are tested. Experimental results show that the jumping-height-and-distance adjustable ranges are 0.11 m and 0.96 m, respectively, which validates the effectiveness of the proposed JHD adjustment method.

In this paper, the position tracking control problem of pneumatic servo systems is investigated. These systems usually have high nonlinearities and unmeasurable piston velocities. Firstly, by using adding a power integrator technique, a global finite-time state feedback controller is proposed. Secondly, based on homogeneous theory, a nonlinear observer is developed to estimate the piston velocity. Finally, the corresponding output feedback controller is derived, which local finite-time stabilizes the position tracking error system. Compared with the conventional backstepping output feedback control scheme, the developed nonsmooth output feedback control scheme offers a faster convergence rate and a better disturbance rejection property. Numerical simulations illustrate the effectiveness of the proposed control scheme.

Wireless sensor node deployment in unconstructed environments affects the network's connectivity and longevity. This paper presents the 3D self-deployment of jumping robot sensor nodes (JRSNs), which can jump over and onto obstacles to enhance network performances. The effects of JRSNs' 3D deployment on network connectivity and energy consumption are simulated first. Then, we propose a localization method by the fusion of ultra-wideband (UWB), inertial navigation system (INS), and jumping error model for JRSNs' precise locomotion during deployment. Moreover, we propose path planning and deployment algorithms considering the JRSNs' locomotion pattern and obstacles. Finally, we conducted experiments on the localization and deployment of JRSNs in an outdoor environment. The results verified the feasibility of the proposed method and algorithms. The JRSNs' deployment error was smaller than 25 cm. The network connectivity was improved significantly by 57.70% with the 3D deployment. The JRSNs can deploy autonomously for various applications in obstacle-rich environments.

The aerial posture adjustment of jumping robots are very important for them to land safely. The body of our previous jumping robot rotates in the air which may lead to damage of its fragile parts during landing. In this paper, the aerial posture adjustment of the robot is investigated by modeling and simulation. Firstly, inspired by aerial posture adjustment of animals and insects, the robot model with a pole leg and an additional weight (AW) is introduced. Then, the simulations of the model are conducted. Specifically, the jumping without and with active posture adjusting in the air are simulated. The effects of the length of the pole leg, the mass of the AW, and the leg's driving torque on aerial posture adjusting performances are studied. The simulation results show that the active adjusting with the pole leg system can change the aerial posture of the robot and make it land with a safe posture with proper adjusting the parameters. The results of this paper verify the feasibility of the proposed aerial posture adjustment method which will help us to develop more robust jumping robot for applications in unstructured environments.

Thanks to their powerful feature extraction capabilities, deep convolutional neural networks havebecome increasingly popular in sensor-based human activity recognition (HAR) in recent years.However, their high computational cost has hindered their deployment in real-world applications. Thispaper proposes a novel approach to channel pruning using the artifificial bee colony (ABC) algorithm.The ABC algorithm is a metaheuristic optimization algorithm inspired by the foraging behaviorof honeybees. Unlike traditional pruning methods that prioritize channel importance, our methodseeks to identify the optimal number of channels at each layer, i.e. the optimal pruning structure.To reduce human interference, we transform the search problem of the optimal pruning structureinto an optimization problem and solve it automatically through the ABC algorithm. Our method cansignifificantly decrease the model size and computational cost without compromising performance.We on four benchmark HAR datasets showcase the effffectiveness of ABCSearch. For example, on theWISDM dataset, we achieved an 89.42% reduction in computational cost, an 88.70% reduction inmemory consumption, and a 5× acceleration on embedded platforms while maintaining comparableaccuracy